AWS re:Invent 2022 - Supply chain and logistics (INO105)

Aug 16, 2023

AWS re:Invent 2022 - Supply chain and logistics (INO105)

This session provides an overview of Amazon.com and AWS supply chain and logistics strategies, starting with an overview of digital supply chain strategy. For the Amazon supply chain, the session provides an overview of Amazon global operations, automation and scale of the supply chain, and the Amazon fulfillment center strategy including robotics. For the AWS supply chain, the session provides an overview of the complexity of producing and delivering equipment to global data centers. The session then builds on this information to provide guidance for how AWS services can be used to support the operation and forecasting of a digital supply chain.

Learn more about AWS re:Invent at https://go.aws/3ikK4dD.

Subscribe:
More AWS videos http://bit.ly/2O3zS75
More AWS events videos http://bit.ly/316g9t4

ABOUT AWS
Amazon Web Services (AWS) hosts events, both online and in-person, bringing the cloud computing community together to connect, collaborate, and learn from AWS experts.

AWS is the world’s most comprehensive and broadly adopted cloud platform, offering over 200 fully featured services from data centers globally. Millions of customers—including the fastest-growing startups, largest enterprises, and leading government agencies—are using AWS to lower costs, become more agile, and innovate faster.

#reInvent2022 #AWSreInvent2022 #AWSEvents

Content

0.72 -> - Hey thank you all very much for coming along

3 -> to this session on supply chain.

6.03 -> There's one problem with being a supply chain person.

8.16 -> It's that you're always looking out for those opportunities

10.62 -> and today on the way to this session,

12.3 -> I really wanted a cup of tea.

14.517 -> Now not American tea, not cold.

16.14 -> Warm, with milk.

18 -> And I stopped off at that very famous provider of coffee,

21.18 -> and they had no English tea.

23.76 -> It's okay, it's all right, I can live with that

25.65 -> 'cause there's always another one.

27.42 -> So I carried on my journey towards the conference room

29.85 -> and I finally get to the next store.

32.182 -> Queue up as you do.

33.84 -> Dutifully wait, I'm British, I love it.

36.09 -> Finally get to the front of the queue say,

37.957 -> "English breakfast tea."

39.09 -> Of which there are none.

40.8 -> However, cunningly I always carry two teabags with me.

45.12 -> For this very reason when I travel in America.

47.76 -> But fortunately I'm in Vegas where you pay for everything.

50.97 -> So I even had to pay $1.78 for a cup of hot water,

55.83 -> brilliant.

57.12 -> So that was a great start to the day

58.47 -> and I won't even talk about losing my bags on the way here.

61.23 -> So John McFall,

62.55 -> I look after AWS' business unit

65.37 -> that specializes in supply chain transportation

67.77 -> and logistics.

68.61 -> We really did try to keep the business unit focused.

71.85 -> So as you can imagine, it's a fairly broad series of topics.

75.48 -> So I'm not gonna be able to cover everything in great depth.

78.06 -> However, I can see from the audience

79.53 -> that it would be guys who are super interested

81.18 -> in different aspects.

82.44 -> There are no points

83.37 -> for getting to the end of the presentation.

85.38 -> So if you have got questions

86.367 -> and we want to turn it into a dynamic session,

88.86 -> then Nash has very kindly agreed to act

91.8 -> as a mobile microphone.

93.27 -> So go ahead and feel free to stop me, ask a question.

97.17 -> You can probably tell I have a bit of an accent,

99.12 -> so I'm from Belfast and that means breathing is optional

101.94 -> when talking.

103.02 -> So as a result I can really squeeze a lot of stuff in

106.02 -> to a short period of time.

107.67 -> So again, if you don't understand something

109.44 -> or you want me to repeat something, just shout.

111.48 -> I'm very accustomed to it.

114.22 -> All right, so what's the agenda of today?

116.73 -> So we're gonna talk a little bit about

118.2 -> what you already know about Amazon

119.88 -> and I'm gonna maybe ask you some questions

121.56 -> about what you think you know.

123.03 -> And then we'll go and have a little look

124.2 -> at our digital supply chain spectrum

126.48 -> before we dig into some of the other areas.

128.94 -> So let's go straight to it.

130.08 -> What do you already know about Amazon?

132.9 -> So maybe we get asked a question, what is Amazon?

138.15 -> What do you think?

138.983 -> What comes to your mind when someone says Amazon?

143.91 -> Any thoughts?

146.07 -> Delivering packages yep, screensaver for mom.

150.09 -> - [Audience Member] An addiction?

151.32 -> - Sorry? - An addiction.

152.94 -> - Yep.

155.885 -> (laughs) Addiction.

157.59 -> It's coming up to Christmas time.

159 -> Just to be clear,

159.833 -> for all the gentlemen in the room,

160.92 -> I'm sure you'll be ordering two days before Christmas.

163.5 -> So stand by, it's quite rushed.

165.78 -> Any other thoughts?

168.45 -> Film distribution, logistics company.

170.43 -> Freight forwarding, global, it's huge right?

173.13 -> But of course we're also a media producer,

175.26 -> making our own movies.

176.58 -> We make our own devices.

178.32 -> So there's all sorts of complication

179.91 -> within even just the Amazon supply chain

181.62 -> and it's everything from the distribution of film digitally

184.74 -> right through to maintaining our own distribution

188.88 -> of our physical products that we manufacture, et cetera.

191.25 -> So it's a fairly complex piece

192.87 -> but we're gonna focus mostly on the retail business

194.82 -> because it's that that we're most interested in

197.309 -> in terms of understanding how Amazon operates.

199.59 -> So more than you know, billions of packages, 185 countries.

202.8 -> In Q4 2019, in Europe alone, 19 billion packages.

207.81 -> So a huge number of packages being delivered to customers.

210.54 -> Hopefully mostly on time.

212.7 -> And hopefully mostly what you actually ordered

216.09 -> because we do strive for that customer centricity.

219.15 -> We don't always get it right but when we don't,

220.95 -> we hope that the connected customer experience helps us.

224.64 -> So in terms of the, ope I've gone the wrong way.

228.06 -> In terms of Amazon itself,

230.37 -> when we talk about this notion of unprecedented scale,

233.61 -> this is when you're driving by Texas

235.44 -> and you suddenly see the million square foot facility

237.69 -> and you're like,

238.924 -> "Wow Amazon, unprecedented scale."

241.08 -> But for me, that's just one way to measure it.

244.62 -> We have more than 250 million unique products.

248.43 -> Represented in 450 million at the SKU level,

251.01 -> so physically on the shelf.

252.72 -> So that's another way to measure complexity but you know,

255.93 -> it doesn't really scratch the surface.

257.91 -> For anyone here who's working in supply chain,

259.98 -> you know what I'm about to go into here.

262.23 -> We need to move the product from China.

264.21 -> We need to be moving it across

265.32 -> to the U.S. like everybody else.

266.82 -> We don't own our global mile.

268.62 -> So we are dependent upon Maersk,

270.671 -> Kuehne+Nagel, MSC and everybody else

272.13 -> to move that volume for us.

274.17 -> We don't own all aspects

275.34 -> of our freight forwarding capability

277.17 -> and nor do we own all aspects of our middle mile.

279.84 -> So yes we have Amazon Transportation Services

282.36 -> but we are dependent upon Eristobarts,

284.4 -> J. B. Hunt, everybody else to move the volume.

286.77 -> And of course there are not enough tractor units,

288.69 -> nor are there enough drivers.

290.19 -> So if you want to go and buy a tractor unit

292.2 -> and you want to go and sell your services to us,

293.94 -> we will happily buy it from you.

295.32 -> So therefore we are also using mom and pop shops

297.99 -> with one or two tractor units

299.64 -> and we're having to do that in every single country

301.59 -> where we operate.

303.3 -> And then it gets even more complex right?

304.95 -> Because that's one aspect.

306.912 -> But what about getting it to the end customer?

308.49 -> So when it comes to getting it to the end customer,

310.11 -> look at Europe.

311.444 -> We are using post couriers, Belgium Post, UPS,

314.4 -> we're using our own delivery service partners.

316.47 -> We're using flex drivers, Amazon Logistics,

318.75 -> Royal Mail, you name it, we're using them

320.76 -> to get it into the hand of the final customer

323.04 -> because there's insufficient capacity for us,

325.92 -> never mind everybody else

327.09 -> that needs to move volume around the network.

329.85 -> So when I talk about unprecedented scale,

331.65 -> for me it's more about that communication effect.

334.44 -> The data transfer effect.

336.15 -> The ability to coordinate all of that, seamlessly,

339.18 -> so that the customer is delighted

340.65 -> with a standard package in their hand, on time.

344.04 -> On promise, at the right quality.

347.55 -> The next aspect is hyperspeed and you know,

350.55 -> when I listen to all my solution architect types,

353.04 -> you know 29 million transactions per second

355.02 -> over the Amazon website they're all,

356.557 -> "That's fast John."

357.78 -> Now fortunately, as you can probably tell,

359.88 -> I'm not a solution architect.

362.43 -> So for me, that's super interesting

364.17 -> but I step back and I say to myself,

365.827 -> "Do you know what though?

367.297 -> "We're also underpinning NASA

369.337 -> "and we're also underpinning large banks."

371.82 -> Massive streaming services.

373.32 -> So I hear you about the transactions over Amazon

375.78 -> but in the context of AWS,

377.91 -> you know it's got to be proportionate.

380.34 -> So really the big thing for me and for many of you

382.95 -> will be the physical arrival of a product in your hand.

386.366 -> So let's get a question from the audience.

389.76 -> So the U.S. is our busiest market.

393.21 -> Q4, not surprisingly, the busiest quarter.

395.97 -> Last year we decided to put ourselves

397.5 -> under a bit of pressure,

398.333 -> so we put Prime Day in Q4 and then we thought,

401.347 -> "Do you know what?

402.18 -> "We also have this thing called the pandemic."

403.44 -> And I don't want to make light of it,

404.52 -> I know it's a very devastating piece

406.32 -> but we have the pandemic

407.52 -> and it's causing operational challenges right

409.41 -> that we all faced.

410.46 -> Separation of people,

411.57 -> limitation on capacity and everything else.

413.7 -> So in the context of the pandemic,

415.65 -> busiest day, busiest quarter, busiest market.

418.23 -> How quickly, fastest time, not average,

420.3 -> do you think Amazon was able to put a product

422.46 -> into the hand of a customer from point of order?

425.25 -> What do you reckon?

428.081 -> Thoughts?

431.22 -> Shout.

432.96 -> Sorry?

434.67 -> One hour.

436.26 -> Any takers?

438.18 -> Processing the order, picking the order,

440.34 -> packing the order,

441.27 -> physically taking it to the customer door.

444.72 -> Height of the pandemic.

448.17 -> Two hours?

450.18 -> 10 hours?

451.62 -> The answer is 29 minutes and 54 seconds.

455.64 -> Okay.

456.473 -> Now let's reflect on that for just a moment.

458.34 -> There are companies that can do it faster than that.

461.997 -> But the companies that are doing it faster,

463.56 -> are doing it normally with a discrete product line

466.14 -> for a dedicated purpose.

467.58 -> In other words, I've got a legal document.

469.02 -> I need you to take it from one office to the other,

470.79 -> because lawyers still like to create paperwork.

473.632 -> So there's a guy in Spandex ready to go,

476.64 -> you hand him the legal document, he jumps on his bike.

478.92 -> Cycles over, hands it over.

480.15 -> Job's a good.

481.23 -> But we're trying to do it with 250 million unique products,

484.17 -> 450 million at the SKU level

485.73 -> and we're trying to identify which of those products

487.68 -> can actually be delivered fast at scale.

489.93 -> We are not delivering a washing machine any time soon

492.87 -> in Manhattan in 29 minutes, okay?

495.78 -> So we need to identify exactly what can be delivered.

499.2 -> So when we talk about speed in that context,

501.27 -> it's not necessarily

502.32 -> the physically handing the product over to you that matters.

505.23 -> What matters is that we're able to use AI and ML at scale.

509.7 -> Because what we're able to say is,

511.267 -> "What product types could we deliver in sub-hour?"

515.67 -> What are people likely to require in Boston,

518.79 -> in the sub-hour requirement?

519.93 -> And by the way, Boston is not homogenous.

522.39 -> So the north, the south, the east, the west are different.

525.33 -> So what are people likely to require

526.89 -> in those parts of the city?

528.96 -> At the block level.

530.46 -> And then of course that's only part of the jigsaw right?

532.86 -> Because we also need to know what volume is present

535.53 -> in the fulfillment center.

536.363 -> Not more or less,

538.14 -> not get your notebook out and tell me.

540.84 -> I need to know.

542.79 -> I need to know the weather conditions,

544.8 -> I need to know the traffic conditions

546.48 -> and I need to be able to provide the customer with a promise

548.97 -> before they've even ordered the item.

552.72 -> AI, ML at scale.

554.01 -> So to me that is a better indication of speed

556.44 -> than virtually anything else.

558.51 -> Then the next aspect is this notion

560.34 -> of relentless innovation.

562.38 -> Now I am told that I should never mention the Fire Phone

564.36 -> even though it was obviously a big hit.

566.25 -> But we have innovations with things like AWS

569.7 -> which is a massive innovation that's a very obvious one.

572.13 -> But even with a flop like the Fire Phone,

575.07 -> we were able to draw out components like Alexa

577.83 -> which then subsequently go off to be very successful

581.49 -> but those are things that you could just, they're big.

584.46 -> But what about all the little innovations

585.84 -> that you just walk by?

588.39 -> If you log onto Amazon and you go to buy an inflatable kayak

593.64 -> the very first thing we're gonna offer you,

595.874 -> two paddles so you don't get stuck up the creek

598.17 -> without them.

599.76 -> The very next thing we're gonna offer you,

601.71 -> a blowup vest so you don't drown.

604.08 -> Then we're gonna go and get you

604.98 -> one of those little baggy things

606.03 -> that you can put your cellphone in

607.515 -> and we're gonna offer you all sorts of stuff.

608.94 -> Why, because you didn't want an inflatable kayak.

612.544 -> What you wanted was to go and spend time

615.24 -> on the water with your family.

617.13 -> And in order to do that,

618.33 -> there's lots of other things that you require.

620.7 -> That's a cool innovation.

622.68 -> I mean it really is.

623.513 -> It's one that you walk by it,

624.346 -> you wouldn't even think about it.

625.41 -> But it's a cool innovation and what we do

627.99 -> is we take that sort of innovational thinking

630.24 -> and we bring it out to our customers.

632.1 -> So I'm working with a remote operation up in Canada

635.406 -> and we say to the supply chain team,

637.927 -> "What's the challenge?"

638.76 -> You know, repairing our vehicles, whatever it is.

641.16 -> And the reality is,

641.993 -> is that the engineer in that remote location

644.82 -> doesn't want an oil filter.

647.31 -> They might be ordering an oil filter

648.93 -> but they don't want an oil filter.

650.73 -> What they want is a serviceable vehicle.

653.204 -> So if more often than not, when you buy an oil filter,

656.55 -> you also require a special oil filter tool and oil

660.27 -> and a sump tray and various other components,

663.33 -> offer that all up front to the engineer

665.76 -> so that they can make a selection

667.17 -> of what they actually require

668.76 -> so that when the parts arrive, the vehicle gets serviced

671.732 -> because that's what we actually want.

674.43 -> And taking that innovative thinking into your supply chain

677.01 -> helps change the way that you think about delivery.

680.19 -> It's not about putting the cellphone

681.51 -> in the hand of your customer.

683.1 -> It's about making the customer

684.51 -> able to take a picture of the lunch

686.19 -> so they share it with the world

687.36 -> 'cause we all need to see more pictures of people's food.

690.24 -> All right, so it's that ability to make sure

691.89 -> the phones are activated.

693.6 -> Then of course we like to think of ourselves

695.13 -> as being a customer-centric organization.

697.8 -> Customer-obsessed to use an even stronger word.

700.68 -> I'm gonna let you be the judge of that

702.42 -> because you're gonna hear me talk about customers as we go

704.88 -> but we'll see what you think.

707.632 -> So in terms of the digital supply chain spectrum,

711.48 -> this is a little bit like the political spectrum.

713.73 -> Now before you take a deep breath,

715.32 -> Northern Irish person talking about politics on stage,

717.75 -> in America, it's okay.

719.91 -> I'm not going to.

721.14 -> I'm just going to say this.

722.88 -> You're neither left wing or right wing on every issue.

725.64 -> And therefore depending upon the issue

726.99 -> will depend upon where you are in the spectrum

728.88 -> and it's the same as this.

729.87 -> You're neither nascent in everything.

731.91 -> You're likely to be quite advanced in some areas,

734.64 -> and quite nascent in others.

736.35 -> And when you look at this spectrum,

738 -> the sorts of symptoms that you hear

740.52 -> for someone in nascent is things like,

743.077 -> "I'd love to be able to find out what's on that truck

746.227 -> "but I would have to pull out one Excel spreadsheet

748.597 -> "that tells me all the trucks that I've got.

750.367 -> "I would have to pull out another Excel spreadsheet

752.077 -> "that tells me what all the purchase orders are.

753.637 -> "Another Excel spreadsheet that tells me

755.137 -> "where all the customers are.

756.367 -> "I've got a little pie chart to make by lunchtime

758.677 -> "and then I need to do all of these things

760.057 -> "just to find out what's on one truck."

762.09 -> And it's really painful.

763.29 -> So you can't really react to dynamic situations.

766.14 -> In other words, I phone you up and I say,

767.647 -> "Hey I'm sorry to tell you but the ship

769.747 -> "that's coming into Boston, two-week delay."

773.22 -> Probably not an awful lot

774.45 -> that most businesses in nascent can really do about that.

777.33 -> It's a very manual process to react.

780.6 -> But you see, there's a trick to get out of this isn't there?

783.39 -> And the trick to get out of this

784.912 -> is to go and speak to your CFO

786 -> who are always very friendly people and say,

787.897 -> "Give me $4 million

789.397 -> "because what I'm gonna go and do

790.507 -> "is I'm gonna buy a whole load of

791.737 -> "best of breed SaaS offerings."

793.29 -> Software as a service offerings.

794.797 -> "And I'm gonna go

796.179 -> "and buy myself out of that nascent requirement."

797.91 -> That's cool.

799.334 -> Problem then becomes,

801.15 -> I've got an advanced fuel management system.

804.03 -> I have an advanced fleet scheduler.

806.1 -> I have an advanced labor scheduler and so forth.

810 -> But when I want to actually find out what's going on,

812.55 -> I need 1500 passwords.

813.99 -> I need to download all the data into Excel,

816.03 -> so that I can then combine the data sources

817.71 -> to then be able to make a decision

819.45 -> and now everybody's frustrated.

821.25 -> Because we didn't get the chance to get this benefit

823.53 -> that we were talking about from the cloud.

825.51 -> We're getting best of breed but siloed.

829.358 -> The way that Amazon operates

831.12 -> is that Amazon is predominately autonomous, predominately.

836.46 -> If you go into one of our FCs,

838.05 -> we have magnets with people's names on them,

840.39 -> stuck on whiteboards managing scheduling in some areas.

843.96 -> We do even use Excel.

846.78 -> But more often than not,

848.16 -> for the bigger things, we have automation

850.62 -> and I'm gonna show you some of that as we go.

852.78 -> The final type is integration.

855.15 -> And in integration,

856.08 -> this is where we're trying to make it seamless.

858.63 -> So I work with a corrugate manufacturer, I consume boxes.

863.43 -> I provide the corrugate manufacturer

865.41 -> with my inventory holding dynamically

867.33 -> so they know exactly what I've got.

868.77 -> They know my rate of consumption.

870.39 -> And their sole job is to make sure

872.04 -> that I've got the right corrugate at the right time

873.81 -> to sustain my operation.

875.58 -> What that means is that I don't phone them up in November

877.71 -> with 200 fulfillment centers and say,

880.087 -> "We all need more boxes."

881.85 -> And we suddenly find that our supplier

884.55 -> is unable to meet the requirement.

886.53 -> Not because they don't want to, but they just can't.

889.29 -> So by actually saying,

890.587 -> "Just give me what I need."

891.75 -> We hope that we can actually manage the process

894.545 -> and that becomes integrated right?

895.378 -> So I'm just paying for it seamlessly.

896.211 -> I don't need to phone anybody up, it just happens

898.41 -> and that's a lovely place to be.

899.7 -> But for Amazon, that's something that we're pushing into

901.8 -> but we're not there in all regards.

905.527 -> Then how do we do this?

907.71 -> When we think about supply chain and certain pillars

910.02 -> and these won't surprise you, this is not rocket science.

912.66 -> In the first pillar we have planning

914.82 -> and when we talk about planning, it's everything right?

916.41 -> We're talking about labor planning,

917.7 -> we're talking about capacity-planning, truck planning.

920.07 -> We're talking about spot logistics, fuel purchase.

922.74 -> EV charging, all of it.

924.15 -> Plan, plan, plan

925.83 -> and all of that planning needs to be, in some way,

928.524 -> integrated right?

929.67 -> Because there's no point in planning EV charging

932.58 -> but not thinking about your right bind flows.

935.88 -> Because all the vehicles will be on the charge

937.35 -> at the wrong time.

938.52 -> And of course if you're in California,

939.87 -> where maybe you're only allowed to charge certain times

942.18 -> there might be also challenges

943.26 -> around how you're actually gonna sustain the management.

946.44 -> The next aspect is this notion of collaboration.

949.83 -> It doesn't matter how big you are

952.86 -> or how small you are in this space.

956.04 -> You have to work with everybody.

958.47 -> Now I outlined that notion of not owning all aspects

961.23 -> of your supply chain.

962.79 -> So that means you can't walk around with a big stick

964.95 -> demanding that people deliver stuff for you.

967.08 -> It ain't gonna work.

968.73 -> Transactional relationships aren't going to work.

971.34 -> You have to find a way

972.63 -> of being able to coordinate your activity with everyone.

975.941 -> Let me give you an example.

976.774 -> I was running a fulfillment center,

978.03 -> just outside of Peterborough in England and I,

981.36 -> I was on a Friday call at 1700.

983.49 -> I was that guy.

984.57 -> And all of my fulfillment centers begin with the letter E,

986.73 -> so I'm gonna be there for a long time.

987.99 -> So I've already told my wife I ain't coming home

990.09 -> and I get the call that says

992.22 -> my delivery estimated accuracy has fallen.

994.17 -> Not only fallen, you John,

996.628 -> are actually dragging down the whole European average.

1000.89 -> So to be super clear, you're fixing that problem

1003.92 -> and you're fixing it now okay?

1007.1 -> So when I did the work,

1008.96 -> I find that one of our service providers was the root cause.

1013.04 -> They were simply messing up

1015.05 -> and it would be very easy to phone them up

1016.37 -> and get all angry and everything else.

1018.62 -> And I probably was a bit grumpy but you know,

1020.42 -> it was late on a Friday.

1022.04 -> So what we did is we went out

1024.064 -> and actually visited their delivery station.

1025.4 -> What can we do to make this easier for you?

1027.564 -> And it turned out that they didn't want to let us down.

1030.8 -> They didn't want to let Amazon down.

1032.39 -> So they wanted to make sure

1033.44 -> that they were just taking the volume.

1034.52 -> They didn't want to complain but what we were doing

1036.74 -> is we were deliberately storing all of the packages

1038.78 -> until 1700, putting them on the truck

1040.85 -> and then sending the truck.

1042.35 -> Which meant that at 1730,

1044.57 -> they would open the door and it was like Christmas.

1046.1 -> They were like oh.

1047.42 -> They had all this day where they were doing nothing

1049.25 -> and suddenly they've been swamped with volume.

1052.1 -> So a very simple fix of course is free flow the volume.

1055.85 -> So they just started taking the volume off the dock

1058.07 -> and it resulted in my delivery estimated accuracy

1060.5 -> not resulting in Friday calls anymore

1062.24 -> and everybody was good.

1063.073 -> But it's just a simple example right?

1064.31 -> It's just a simple example

1065.48 -> of not blaming anyone else in your supply chain

1067.52 -> but working together in a collaborative way

1069.05 -> to solve a problem.

1070.04 -> And sometimes it doesn't need technology.

1073.37 -> The other part is to do with execution

1076.79 -> and the execution bit

1077.93 -> really then falls down to this notion of so what

1080.66 -> when the situation changes?

1082.52 -> And there are great bumper stickers in the U.S.,

1084.71 -> that sort of say it right?

1085.61 -> It happens.

1087.5 -> So when it does happen, what are you going to do?

1090.071 -> What are you going to do when the truck is late?

1092.36 -> What are you going to do when the road is blocked?

1095.93 -> The border is closed, the port is congested.

1100.31 -> And that means we have to be able to take into account

1102.89 -> all of our planning.

1104.51 -> We have to be able to take into account

1105.95 -> all of our collaboration and we have to make a change.

1109.49 -> So for instance, we might say,

1111.906 -> "Well let's just take the volume

1112.739 -> "and send it to a different fulfillment center."

1114.29 -> For argument's sake.

1115.43 -> But we need to take into account

1116.84 -> that the inventory is now in the wrong place.

1118.64 -> That's gonna change the customer order profiles,

1120.5 -> that's gonna change the cost,

1121.55 -> the variable cost of unit for product, et cetera

1123.59 -> and it also means that the next time the volume comes in,

1125.66 -> we don't need to send it to that fulfillment center anymore.

1127.37 -> We need to send it to a new one.

1128.93 -> So that whole notion of execution needs to be dynamic

1131.54 -> and it needs to understand the planning

1133.04 -> and it needs to understand the collaboration.

1135.2 -> In other words what you must be able to do

1137 -> is to draw insights out of your supply chain

1139.43 -> that enable you to be able to make accurate decisions.

1143.918 -> But how do we do it?

1147.32 -> Well we have these mechanisms that we use.

1150.151 -> What are mechanisms?

1151.054 -> Well if I tell someone to go and do a better job,

1153.681 -> they already think they're doing a good job.

1156.35 -> So in the end, they're not gonna try to improve

1159.8 -> whenever they're already doing a good job.

1161.75 -> So we need to have processes, tools

1163.31 -> and inspection methodology

1164.9 -> that enables us to hold ourselves to account

1166.91 -> so that we actually do the things that we do.

1169.1 -> But what I want to nail quickly is architecture.

1172.79 -> And the architecture is important

1174.53 -> because it underpins how I'm gonna set up

1176.81 -> the rest of the conversation.

1178.67 -> So when we talk about architecture,

1180.23 -> we use microservices.

1182.245 -> And the microservices, the best way to think about them

1184.88 -> is a small widget of technical capability

1187.7 -> that solves a specific problem.

1191.99 -> Here is one.

1193.43 -> So this lady has been sent

1195.805 -> to the right part of the fulfillment center

1197.03 -> to pick a product to put it into the tote.

1200.06 -> That is the pick microservice.

1203.24 -> It's all it does, it's all it will ever do.

1204.59 -> There's a whole team of people that look after pick,

1206.874 -> they innovate in pick

1207.707 -> and they do all sorts of cool things with pick.

1210.939 -> But to pick something, you need to know what it is,

1214.827 -> you need to know where it is, et cetera.

1215.87 -> In other words, you need other microservices

1218.06 -> that communicate effectively.

1220.04 -> You bring that together,

1222.262 -> now you're able to go pick a product.

1223.43 -> If you want to do labor planning,

1224.81 -> cross-border tax reconciliation

1226.79 -> and all of these other things, you end up with amazon.com.

1231.077 -> Right?

1232.91 -> Now let's reflect on this for one second

1234.62 -> because it's always good to do the opposite test.

1236.81 -> If we were now to go and close the doors

1238.31 -> and don't worry, the guys aren't gonna do this.

1239.51 -> If we go close the doors

1240.77 -> and we got lots of clever people here,

1242.24 -> and I said to you, "Let's go and draw end to end,

1244.977 -> "every single aspect of your business' supply chain."

1248.51 -> We'd fail right?

1250.52 -> We'd fail.

1252.52 -> But even if we were remotely successful,

1254.178 -> somebody like me would come along and completely break it

1256.338 -> because I would do something you hadn't anticipated.

1258.11 -> So this notion of the microservices

1259.97 -> enables you to be quite dynamic

1261.5 -> in how you go about solving complex problems

1264.32 -> and it also enables you to think differently

1265.94 -> about how you use it.

1267.84 -> So you might be using an access control card

1269.3 -> for simply gaining security to the building.

1271.97 -> But that access controlled card could then be used

1274.28 -> to help allocate someone to a work station

1276.8 -> or to have their coffee ready

1278.09 -> whenever they walk through the door or to ensure

1280.52 -> that there's actually English breakfast tea available.

1283.58 -> So in that notion of Amazon,

1286.37 -> you can see the microservices work.

1288.41 -> So what?

1290.336 -> There is an absolute explosion of data.

1292.64 -> There are lots of people that want to come

1294.47 -> and interact with your data

1296.18 -> and because there's an explosion of data

1297.95 -> and people interacting with your data,

1299.93 -> guess what they want to do?

1301.891 -> They want to make decisions.

1304.181 -> So that means that it is no longer good enough

1307.22 -> to have a 24-hour batch processing inventory decision.

1311.57 -> You need to know

1312.403 -> if you're gonna make a decision, you need to know.

1314.643 -> And I give you an example.

1316.04 -> I was working with one customer and at midnight

1318.95 -> they would virtually send the inventory

1322.97 -> to the store manager.

1324.74 -> And then at 10 o'clock the next day or so,

1326.66 -> the truck would arrive with the inventory.

1329.06 -> The truck may have the inventory

1330.47 -> that was virtually sent to you

1331.82 -> but there was no reconciliation

1333.2 -> between those two positions.

1335.06 -> So you have a very clever algorithm

1336.71 -> that is correctly identifying

1338.81 -> which place to send the inventory

1340.52 -> and then you have an execution

1341.69 -> that's not necessarily linked to the two.

1344.06 -> So you have a breakdown in this notion

1345.77 -> of decision-making versus the accuracy of your data.

1351.11 -> So in Amazon, we have more than 50 different systems

1354.02 -> that are doing nothing but scanning the horizon

1356.3 -> for inbound flow.

1359.45 -> Checking the capacity of our fulfillment centers,

1362.39 -> are we paying the right price for the products,

1364.517 -> for the logistics that we're using and so forth.

1367.52 -> Why?

1368.48 -> Well because we need to do inventory planning and control.

1372.62 -> And that means we need to be able to simulate the capacity.

1374.93 -> Capacity being space, physical space on the shelf.

1378.5 -> Capacity being process capacity.

1381.56 -> Capacity being specialist logistics.

1383.81 -> We carry HAZMAT so we got perfumes and so forth

1386.87 -> and it has to be sold under specific conditions

1388.94 -> and in certain jurisdictions,

1390.814 -> those restrictions are further convoluted.

1394.1 -> We have micro supply chains.

1395.78 -> Apparel is really challenging to move

1397.58 -> for anyone in that industry.

1398.9 -> You know, I feel your pain,

1400.4 -> especially when it comes to January.

1402.35 -> So we have all the apparel to deal with

1404.45 -> and it has to go through specialist supply chains.

1406.79 -> Beer, wine, spirits, et cetera.

1409.76 -> And this is what it looks like

1410.93 -> when you move away from Excel.

1413.06 -> This is a view that one of,

1414.44 -> an Amazon employee would have

1416.57 -> if they were looking at the inventory position.

1419.33 -> You can see the number of units,

1420.62 -> you can drill through the units.

1421.76 -> Hard lines, soft lines.

1423.05 -> You could go and see whether it's you know,

1424.76 -> consumer electronics, high value items.

1426.56 -> It doesn't matter.

1427.55 -> And therefore you're able to take a detailed view

1430.13 -> as to what's going on within the inventory

1433.13 -> and what's coming towards you.

1435.147 -> That's inbound operations only.

1437.121 -> Then the problem is though,

1440.45 -> is that you have some really big companies

1442.49 -> that we work with doing great things.

1444.59 -> Procter & Gambles and others

1446.27 -> who are moving volume into Amazon

1448.91 -> but we also have some small companies.

1451.25 -> John McFall and Co. who sell two mugs a year over Amazon.

1455.93 -> In all of those instances,

1457.07 -> we need to be able to take the volume

1458.81 -> and bring it in to our equation.

1461.45 -> So we need to be able to provide an API gateway

1463.85 -> for advanced customers

1465.23 -> that seek to just push the data to us electronically

1468.08 -> and we also need to be able to provide a gateway

1469.85 -> for people that need to log onto our portal and update it.

1472.517 -> And in that way,

1474.081 -> we're able to help small businesses

1474.914 -> to gain entry to a global market

1477.29 -> which is very exciting and you know how proud we are

1480.23 -> of our small vendors getting and becoming very successful.

1484.37 -> So those give you that ability to do it

1486.65 -> and then when you take a step away from it again

1488.93 -> and you move into the business area

1490.16 -> where many of you will operate.

1492.31 -> In this instance,

1493.143 -> we're able to then provide detailed summaries

1495.53 -> with confidence levels

1497.21 -> and for those of you that are working with data,

1498.92 -> you know how important it is.

1500.39 -> We need to be able to provide the confidence level

1502.28 -> so that the business user understands

1504.59 -> what model and simulation outcome is relevant to them.

1508.4 -> So in this instance,

1509.233 -> you can see that the confidence level is low

1511.04 -> for all of the data in that first slide.

1513.02 -> It's good to know it's low.

1514.01 -> So I don't go and hire 2000 people

1515.72 -> for the fulfillment center.

1517.01 -> Let's wait until we get that confidence level up

1518.81 -> so I know exactly what I need to hire.

1522.26 -> And we are able to then use that sort of information

1525.11 -> to be able to provide inventory, visualizations

1528.03 -> and recommendations.

1529.931 -> You may or may not have heard the announcement

1530.764 -> about AWS Supply Chain earlier on today

1533.898 -> and that's a screenshot of the sorts of tooling

1536 -> that is available.

1539.12 -> When you then think about why we're doing all of this,

1541.49 -> it's because we want to answer this one question.

1545.42 -> And the way that I reflect upon that, it's really,

1547.647 -> "What would you do if you knew for sure

1549.237 -> "something was about to happen?"

1553.7 -> But the real trick here

1555.23 -> is that you have to be able to reason on top of the data.

1559.266 -> It's not good enough to just be able to see it,

1560.099 -> you have to understand what it's telling you.

1562.91 -> Example, ports are somewhat congested at the moment

1567.2 -> and have been for some time.

1569.36 -> So Amazon identifies that along with everybody else

1571.58 -> and we recognize that yes,

1573.59 -> the big deep sea ports are.

1575.06 -> But the smaller ports are not.

1577.28 -> So we take smaller vessels out of charter

1580.31 -> and we reroll them to TUC-carrying vessels.

1583.55 -> Then we identify that there are insufficient

1585.2 -> containers in the world.

1586.61 -> So we have to manufacture our own.

1588.733 -> So we're now manufacturing our own containers,

1590.06 -> placing them on chartered vessels

1591.53 -> and sending those vessels into smaller ports.

1595.7 -> That's really expensive and we're a frugal company.

1600.17 -> So you can't do it for everything.

1603.054 -> So therefore you need to use data to reason

1604.91 -> as to which SKU does it make sense

1607.43 -> to ensure that you've got on your shelves

1609.41 -> at all times to delight your customers?

1611.99 -> And at what point can you step away from doing this

1614.48 -> to get back into normal operations

1616.22 -> where you're able to also ensure

1618.14 -> that you're passing on the right savings to the customers?

1623.06 -> So it's data driven everything.

1626.618 -> And this is my sandscript, so my apologies

1629.24 -> and if you look at what I'm interested in,

1630.8 -> I'm super interested in these operational efficiencies.

1633.83 -> I'm after them all day long.

1635.81 -> What would it mean for you

1637.01 -> if I was able to take one day of inventory holding

1639.89 -> out of your business?

1642.32 -> If I was able to improve your labor cost by 1%.

1647.6 -> If I was able to improve your truck fill rate by 1%,

1651.598 -> reducing your fuel bill, whatever?

1653.24 -> Point is, for most big businesses, that's big numbers

1657.14 -> and for some businesses,

1658.04 -> everybody in this room is retiring to the Bahamas

1659.99 -> if you can go and nail one day's worth of inventory.

1663.38 -> So it's worth getting at on a 1% use case

1666.826 -> but there's no point in spending $10 million

1669.38 -> because you think you're gonna get a $5 million savings

1671.51 -> per day of inventory.

1673.07 -> You need to spend a relatively small sum of money

1675.23 -> to go and experiment.

1676.94 -> To go and test the water and if it's successful,

1679.49 -> then you need to be able to rapidly scale it

1681.29 -> and that's the notion of experimentation

1682.85 -> within Amazon, give you an example.

1685.55 -> So Tide, we all love it.

1687.869 -> You walk into Target, other stores are available

1690.14 -> and it's got that really fantastic-shaped bottle.

1692.69 -> You can see, it stands out.

1694.04 -> It's kind of a bright color

1696.17 -> but as someone that works in ecommerce, it's a killer.

1700.192 -> Because now I've gotta get

1701.198 -> this really awkward-shaped bottle,

1702.32 -> I've gotta place it into another tub

1703.97 -> so that it doesn't spill

1705.35 -> and then I've gotta place it into a cube.

1708.093 -> Boy you're killing me.

1708.926 -> And I've gotta package it all up

1710.335 -> and I know it looks pretty but you're killing me right?

1712.28 -> So we work with Tide and you're saying,

1714.357 -> "Look guys, it doesn't matter.

1716.107 -> "It's coming from an ecommerce company,

1716.94 -> "it doesn't have to look pretty on the shelf.

1718.497 -> "It has to be functional when it arrives with the customer."

1721.61 -> Now this innovation is actually really cool.

1723.5 -> Look at the little legs at the bottom of the box

1725.3 -> that pop out.

1726.133 -> It mean you can put the cup under the box.

1729.38 -> Now I'm the guy that turns the thermostat down, way down.

1734.09 -> So when it comes to these sorts of things,

1735.65 -> I want every drop out

1737.66 -> and that's why I end up buying lots of clothes

1739.91 -> because I end with bleach and so forth all over me.

1743.33 -> So in this box, what you actually have

1745.22 -> is a little gravity ramp.

1746.54 -> So it actually automatically lets everything out.

1749.2 -> That's cool, I love it.

1750.813 -> But the reason why I'm just sharing with you

1751.67 -> is actually look at the numbers.

1753.74 -> Because for me, that's a cuboid all day long.

1756.2 -> Stackable, palatable, fantastic, I love it.

1759.74 -> We can get that out to customers super fast.

1761.99 -> Customers are gonna be happy with the innovation

1764 -> and we're also being environmentally sensible

1766.76 -> because we're saving the amount of water

1768.2 -> that we're shipping as well.

1770.197 -> Cool stuff.

1772.73 -> The other thing that we need to be able to do

1774.14 -> is we need to be able to use data

1775.67 -> to drive operational decision making.

1778.46 -> So this slide, the blue lines

1780.83 -> show you all of the various paths that our trucks.

1783.32 -> So we're not talking about all of the other trucks

1784.88 -> that we're using.

1786.278 -> Just Amazon Transportation Services.

1787.111 -> That our trucks could take,

1788.72 -> in order to move around Germany.

1793.2 -> The red is the arcs, the legs that we're going to use

1796.46 -> to be able to fulfill optimally

1798.47 -> the movement of the volume around the network.

1802.906 -> So how good are we?

1805.64 -> If you give us 72 hours notice, we're not so good.

1809.44 -> We get a 12% improvement.

1810.32 -> But for Amazon that's big money okay.

1812.93 -> These are significant improvement but all the same,

1817.176 -> it's a 12% improvement.

1818.18 -> But if you give us a little bit more notice,

1820.25 -> we're up there 29% improvement.

1822.62 -> And to be clear, those are improvements on

1824.6 -> an already advanced scheduling capability.

1826.82 -> So it's quite exciting.

1828.47 -> And it's that sort of pace that's driving it operationally.

1830.9 -> This is when your CFO's going,

1832.377 -> "Yeah thank you, this is genuinely impactful."

1835.31 -> This is when your transportation operational manager

1837.35 -> is actually benefiting from the technology

1839.27 -> that you're putting in front of them

1840.65 -> and gaining operational impact.

1842.63 -> It's when people are able to come and touch the cloud.

1846.2 -> Right?

1847.254 -> And that's what makes the difference for people's lives.

1849.53 -> The other piece that we can do

1850.58 -> is help with simple things like safety.

1853.67 -> You know, simple bit of AI.

1855.77 -> Guy's picking up his phone, that's a problem.

1858.53 -> The guy, what distance is he maintaining

1860.24 -> from the vehicles in front?

1862.34 -> You could use this for convoy marking.

1865.572 -> So if I'm gonna be I don't know,

1866.405 -> a remote builder and I'm operating in a rural area,

1869.57 -> I may choose to not send my trucks through a small village.

1872.9 -> So therefore I can actually assure that my trucks

1875.39 -> are not coming through the village

1876.74 -> by monitoring the path that they're taking and so forth.

1880.984 -> Simple, little camera, plugs on the windscreen.

1884.78 -> But you can also use computer vision

1886.4 -> for lots of other cool things in the supply chain

1888.92 -> and in this instance we're helping with PPE-wearing.

1892.34 -> How many times have you walked around a facility,

1894.86 -> you've got part industrial trucks moving,

1897.44 -> you've got areas where you've got high wrecking

1899.39 -> and whatever else you've got.

1900.5 -> Maybe you're in oil and gas

1901.46 -> where safety is so very important.

1903.5 -> In those instances,

1904.82 -> we can use a little just standard CCTV camera.

1907.1 -> It's not expensive anymore.

1908.81 -> And we can assure that your team are wearing

1911 -> the correct kit.

1913.36 -> You can even help highlight to managers

1914.21 -> hotspots of where people don't wear the right kit

1916.917 -> so you can go in signage

1917.913 -> or you can go and put someone down to help fix the problems.

1920.03 -> But the key thing here,

1921.43 -> at all times with computer vision

1922.97 -> is not just the ability of the computer

1924.74 -> to be able to sense what's going on.

1927.23 -> It's really very firmly about so what?

1930.881 -> You can read the registration plate of a truck, so what?

1933.65 -> You can identify who the truck driver,

1935.63 -> or who the truck operator is

1936.92 -> by reading the side of the vehicle.

1938.81 -> Mom and pop number one, cool.

1941.51 -> But we can also take data off the chassis.

1944.54 -> We can also take data off the sea container.

1947.99 -> Standard cameras, nothing fancy.

1950.69 -> We can use that in your gatehouse.

1952.97 -> So now the person doesn't have to come out and go,

1955.317 -> "Hey Bob, what's on your truck today?"

1958.31 -> I could just simply have a sign that says lane one

1961.443 -> 'cause I know who you are and I know what you've got.

1962.991 -> I can even link it to PO reconciliation.

1969.128 -> Pay the company now 'cause the truck has been received.

1971.33 -> Cool, right?

1972.47 -> Getting into the integration world.

1974.57 -> I can also use the same technology

1976.58 -> to go and scan my yard so I know where the volume is

1979.04 -> because we all may have lost a truck

1980.69 -> at one point or another

1982.43 -> when operating in the supply chain space

1984.74 -> and I have been that guy again that gets the phone call

1987.14 -> that says, "John there's a 40-foot truck

1990.328 -> "full of PlayStations, can you go and find it please?"

1992.09 -> You know, I am full of lots of mistakes.

1994.79 -> So that ability to go and help,

1996.14 -> actually identify where stuff is, is important.

1999.74 -> The other piece that I wanted to share

2000.97 -> is the notion of where things are coming from,

2003.637 -> dynamic planning.

2005.248 -> This is the operational laydown in Europe.

2007.864 -> You'll see lots of gaps.

2008.697 -> The gaps is to do with Post Corrias,

2010.51 -> French Postal Service, Belgium Post,

2012.25 -> Royal Mail, everybody else that we're using.

2014.632 -> So the gaps are filled, it's just not Amazon facilities.

2018.16 -> And what we need to be able to do

2019.75 -> is dynamically answer the question,

2022.515 -> "If a customer orders something, where's it coming from?"

2026.47 -> Now before we've even placed the order,

2028.18 -> what have we told you?

2029.62 -> When you're gonna get it.

2031.21 -> Right, we made you the promise.

2033.272 -> Order it before 11 and you will get it by tomorrow.

2036.896 -> So we've made you a promise.

2037.849 -> So we've already simulated all of this.

2040.436 -> So the green ones, we can actually get the order to you

2042.58 -> on time.

2044.354 -> The blue ones have the volume

2045.187 -> but we can't get it to you on time.

2047.89 -> Then we have a whole bunch of criteria along the side.

2050.993 -> Price, CO2 emissions.

2054.43 -> Load balancing.

2055.81 -> You know, we want to favor French Postal Service today

2058.78 -> because we've actually negotiated a good rate

2061.09 -> and therefore we want to make sure

2062.2 -> that we actually fill up to our rate.

2064.15 -> Once we've filled up to our rate we don't want a penalty,

2067.147 -> so we're gonna suddenly make it more expensive

2068.41 -> and we're gonna drive artificially volume

2070.03 -> to another well-negotiated rate and et cetera.

2074.104 -> So that's all going on simultaneously in the system.

2078.552 -> But what is also cool

2079.51 -> is that when someone comes along as a contractor

2082.3 -> and they do this all the time

2083.89 -> and they drill a hole through your power cable

2086.2 -> and your facility stops working, it's okay.

2089.44 -> You can look after your people.

2091.18 -> People first because it will autonomously switch to LIL1.

2096.468 -> And that's cool.

2097.301 -> And I've had it happen to me twice with amazon.com.

2102.31 -> So how do we get there?

2105.375 -> 'Cause there's lots of people will say to me,

2106.39 -> say, "John well that's really good and it's very interesting

2108.137 -> "but how do we get to a service

2111.047 -> "that is going to give us the visibility

2113.267 -> "and the transparency that we need

2114.947 -> "for our business to be successful?"

2117.52 -> Well every good data company is gonna come

2119.26 -> and they're gonna say to you what?

2121.085 -> Build a data lake.

2123.58 -> But to be super clear, first of all,

2125.38 -> you can't swim in a data lake.

2127.12 -> And secondly, it doesn't actually solve anything on its own.

2131.02 -> So you can get all your technical team around

2132.67 -> and they'll tell you how beautiful it is and it is,

2134.38 -> a beautiful thing, fantastic.

2136.525 -> What does it do?

2138.063 -> Brings data together.

2139.332 -> But what we actually need

2141.94 -> is we need good business questions.

2144.721 -> If you've got a good question

2147.19 -> and you're bringing the right data out

2148.6 -> to answer that question,

2150.859 -> now it's worthwhile, now I am in.

2153.22 -> Because now I am into a data-driven decision making process.

2156.58 -> I have to answer the question based on the data.

2159.7 -> If I do that, then I can start using AIML

2163.549 -> and if I can start using AI and ML,

2166.81 -> I can get decision support but more importantly,

2169.66 -> I can get automation.

2172.63 -> So you have to build the business through that process.

2176.4 -> Painful as it is,

2177.233 -> you have to build the business through that process.

2178.81 -> There are no shortcuts, there is no easy button.

2182.296 -> Here's an example.

2185.601 -> This is the Amazon supply chain.

2186.678 -> Again, stuff on the left is our equivalent of an ERP.

2189.533 -> Some cool AWS services, inevitable data leak,

2191.23 -> some persona running the supply chain and a problem.

2195.19 -> This is the supply route running from Rome, Milan

2197.41 -> to Vienna.

2199.228 -> Thickness of the line is an indication

2200.061 -> of the volume being moved,

2201.38 -> the arrow the direction of travel.

2202.69 -> The Italian government, during the pandemic,

2204.91 -> for a whole host of national reasons closed the border.

2209.29 -> Our supply chain self-healed.

2213.661 -> Now you were all,

2214.624 -> you would've been really impressed by that

2215.77 -> but you're not anymore because you're just like,

2217.007 -> "Well John all you did just TLD one item,

2218.597 -> "we just went to RE4."

2220.63 -> Correct, correct.

2223.69 -> It's at scale making all of those small decisions

2227.2 -> so that it is autonomously selecting the next best option.

2230.38 -> Is this optimal?

2233.796 -> No chance.

2235.337 -> No chance.

2236.796 -> We have not negotiated the best price for the logistics.

2239.304 -> We have not negotiated the right position of the trucks

2240.46 -> and so forth.

2241.293 -> So now if you're wondering what those persona are doing,

2243.43 -> they're not worrying about satisfying the customer,

2245.23 -> notifying the customer, that's all done.

2247.588 -> Customer's happy, that's done.

2248.62 -> They're focused on getting the business

2250 -> into a good position.

2251.881 -> Negotiating with carriers, contacting them,

2253.612 -> making sure that we've got the harder problems.

2254.98 -> What you really need strategic thinkers to solve for.

2257.65 -> Is this about to happen in another market?

2260.23 -> Are we about to see this happen in the U.S.?

2262.692 -> You remember we had

2263.539 -> that whole sort of pandemic domino thing going on?

2265.772 -> When is this going to happen in North America?

2268.06 -> Could we see states closing borders, et cetera.

2272.509 -> Small steps.

2275.76 -> Jeff did not wake up one day

2276.7 -> with a fully autonomous supply chain in his mom's garage.

2280.57 -> Let's dispel that myth, didn't happen.

2283.529 -> Nor did we actually get it right.

2287.549 -> We built a monolithic application and it failed.

2290.44 -> So then what we did is

2292.281 -> we built three monolithic applications and they failed.

2297.04 -> And that's whenever some very clever chap,

2298.63 -> somebody called Andy came along and said,

2300.437 -> "I know how to solve for this."

2302.44 -> And he created AWS.

2303.94 -> Eventually, because he said,

2305.865 -> "Well actually, we can also offer this to market."

2308.9 -> And that's where we end up where we are today.

2310.81 -> So how do you go about this?

2312.16 -> Well the first thing that you need are use cases.

2314.606 -> What are those problems,

2316.3 -> what are those things that you want to solve for?

2319.42 -> Every one of those, multimillion dollar savings

2321.82 -> for most big businesses.

2324.85 -> Then what you need to do

2326.684 -> is build the technology component necessary

2328 -> to solve for that use case.

2330.408 -> That use case.

2332.38 -> Then you need to ingest the data necessary

2334.99 -> to solve for that use case.

2338.56 -> What you don't do is ingest all the data,

2342.22 -> build a big solution

2345.547 -> and then walk around the business saying,

2346.38 -> "Bring out your dead."

2347.44 -> Because you're gonna and fix all these problems right?

2350.02 -> That is super expensive, super time-consuming

2353.17 -> and I can guarantee you,

2354.67 -> everything's gonna look like a nail

2356.14 -> if you've only got a hammer.

2358.66 -> So in that instance, what we drive for here

2360.88 -> is use case identification and I can also guarantee you,

2364.09 -> with 45 years of experience,

2366.61 -> it is easier to get fatter than it is thinner.

2369.58 -> So you're going to add more and more capacity over time

2373.39 -> and you're gonna build more and more capability over time

2376.72 -> and your platform

2377.62 -> is gonna be more and more resilient over time

2379.9 -> and most importantly, because you will be wrong,

2383.2 -> you can quickly course correct as things change.

2388.09 -> And getting into that mindset helps you become dynamic.

2394.45 -> But we can also make use of sophisticated technology

2397.39 -> and we can also leverage things

2399.07 -> like Amazon's physical supply chain.

2402.73 -> So we might have an option that says,

2404.417 -> "Well I could go and spend $250 million

2406.667 -> "and build a facility in New York."

2409.09 -> Or I could just put my volume into Amazon

2411.91 -> and have it distributed by Amazon on my behalf.

2416.639 -> And take that 250 million and build a facility in Brazil

2419.56 -> where we don't have the distribution capability

2422.125 -> that you need.

2423.73 -> So you can make big capital decisions

2426.01 -> by thinking dynamically about how you want to execute

2429.61 -> and that's fun right?

2431.194 -> That is fun.

2432.167 -> That gives you ways of doing things differently.

2435.37 -> The other thing you can do

2436.45 -> is you can start taking a very different view of life

2438.76 -> when it comes to things like inventory management.

2441.993 -> Does this look familiar to most people

2442.826 -> for inventory management?

2445.402 -> Yeah, yeah.

2446.235 -> And what's the safety buffer?

2449.191 -> How do we gauge safety buffer?

2452.54 -> I mean that's how nervous

2453.523 -> the chief supply chain officer is right?

2456.367 -> That's the, I hear the CFO,

2457.78 -> he's telling me to keep the inventory down, I hear him

2459.97 -> but at the same time no one's gonna fight me

2461.95 -> if I stock up.

2463.24 -> So the safety buffer is,

2465.4 -> how close can the chief supply chain officer

2467.516 -> keep themselves together when things are going wrong?

2470.86 -> And they do go wrong.

2472.33 -> So really what you end up with

2473.49 -> is a safety stock that's based on your planned demand.

2476.897 -> And the problem then, and we've seen it writ large

2479.44 -> in the pandemic is that when actual demand

2482.47 -> outstrips planned demand.

2485.553 -> So what can you do when actual demand

2486.61 -> is outstripping planned demand, what can you do?

2490.87 -> Plot the crash?

2493.729 -> Absolutely, it's coming.

2496.631 -> You can rob Peter to pay Paul.

2498.56 -> So you can start prioritizing one customer over another,

2500.942 -> making somebody unhappy but making sure

2501.775 -> you're prioritizing the happy customer.

2503.71 -> You may have seen that play out in semiconductors.

2507.4 -> You know, but really,

2508.57 -> the trick here is that you're going to stock out.

2512.53 -> And the only way to get around this

2513.76 -> to buy excessive amounts of stock

2516.667 -> which is not good for your PML.

2521.633 -> So that's not how Amazon does it at all.

2524.638 -> We are continuously running models

2526.06 -> that are changing the replenishment policy for every SKU.

2531.19 -> So that should mean, should,

2534.128 -> because we do stock out.

2535.57 -> It should mean that when you get to the actual demand

2539.016 -> exceeding planned demand by a little bit,

2541.424 -> you can dynamically change the replenishment policy

2542.74 -> for that SKU.

2545.475 -> Which should give you a first mover advantage

2549.16 -> to secure the right levels of inventory

2551.8 -> to protect the customer experience, should.

2555.053 -> We do stock out, we're not immune from it.

2559.75 -> Let's look at another thing.

2562.505 -> This is a great story.

2563.711 -> We have an apparel manufacturer that suddenly realized

2565.69 -> that they're actually also an agricultural business.

2568.66 -> So we want to make cocktail dresses, that's pretty cool.

2572.95 -> But we also want to be able to track cotton seed

2575.71 -> to the cocktail dress.

2578.38 -> Right?

2579.644 -> That's awesome.

2580.54 -> So we could go and find some really expensive technology,

2582.82 -> you know call out blockchain.

2584.56 -> We could go and find some really cool tags

2587.41 -> that we could put on every single grain of seed

2589.93 -> that's coming into the piece.

2591.34 -> We're not doing any of that, that's too expensive, no way.

2594.01 -> So you have to find very inexpensive mechanisms

2596.65 -> that enable you to be able to track this across the piece.

2600.22 -> So you end up in a situation

2601.75 -> where you're looking to get

2604.373 -> what is effectively end tier integration.

2606.748 -> That great thing that we all know and love.

2608.11 -> We're having to identify where the cotton's coming from.

2611.411 -> We're having to identify where it's been aggregated.

2612.4 -> We're having to identify

2613.969 -> where it's then going to the factory,

2615.311 -> we then have to recognize that actually,

2616.15 -> a cocktail dress is not made of just cotton.

2619.12 -> There will be other components, plastic buttons

2621.46 -> and various other things that will be added to it.

2624.088 -> So therefore the garment will actually be coming

2625.56 -> from lots of different sources

2626.393 -> and then we have to be able to build a picture

2628 -> that the consumer and others will value and that's cool.

2634.96 -> But it's also useful for a business perspective right?

2638.23 -> So yes, it's great for our customers

2639.61 -> to be able to see that we're working in an ESG way,

2642.13 -> and working towards sustainability

2644.17 -> but it's also quite useful,

2645.7 -> whenever you're talking about risk-adjusted pricing.

2648.216 -> And what do I mean by that?

2650.56 -> So I might be buying all of my cotton from,

2653.92 -> for argument's sake, Bangladesh.

2657.34 -> Good price, great relationship, fantastic product.

2662.29 -> But if it's subject to adverse weather, every year,

2666.782 -> and I end up with a supply risk as a result,

2669.496 -> I may need to diversify a little bit.

2670.92 -> So I'm gonna go and buy some volume

2672.387 -> from the next available cotton manufacturer.

2677.41 -> And in the end, I may actually find

2679.755 -> that I've got global mile risks.

2680.62 -> So I may need to go and find somewhere

2682.36 -> with really cheap labor like California.

2686.363 -> And have a proportion of cotton coming from there

2690.204 -> or whatever product that I'm manufacturing.

2691.78 -> I can then adjust the price according to risk.

2693.79 -> So I may artificially make my U.S. source cheaper

2697.6 -> and I may artificially make my foreign source more expensive

2701.53 -> so that my system adjusts the price

2703.36 -> so that I'm actually getting a reasonable amount of volume

2705.37 -> coming from all the different sources

2707.59 -> so that I'm able to protect myself if the situation changes.

2712.281 -> And there's lots of little things that can be done there

2713.2 -> to help businesses.

2715.476 -> Especially when it comes

2716.327 -> to protecting your strategic supplies.

2719.68 -> Then you need to be able to track it all.

2721.96 -> So this is a cool thing, this is Perfect Mile.

2724.39 -> Delivery Estimated Accuracy, DEA.

2726.73 -> And there is a test at the end of this presentation,

2728.92 -> you'll be pleased to know.

2730.24 -> So DEA.

2731.65 -> And what we want to do is we want to make sure

2733.48 -> that we're sustaining performance.

2735.19 -> Remember what I said at the start?

2737.512 -> Amazon does not own all aspects of its delivery.

2741.334 -> So this is actually monitoring the performance

2742.54 -> of third party logistics providers,

2744.52 -> fulfillment centers, everyone that touches a parcel.

2748.15 -> This is the type of technology

2749.86 -> that underpins what I showed you previously

2751.87 -> with the notion of where is the volume coming from,

2754.614 -> how is it moving, what's the performance of the network?

2756.07 -> And it's the type of technology

2757.27 -> that enables you to get visibility over your operators

2760.27 -> so that you're able to get and delight your customers

2763.469 -> with the movement of your volume.

2764.91 -> Whether you're outsourcing it or running it yourself.

2769.12 -> So with AWS, what we seek to do is to build this notion.

2773.53 -> I mean I call it, I mean I'm sure

2775.54 -> there's lots of technical people

2777.234 -> that will roll their eyes.

2778.067 -> I call it AWS Supply Chain OS.

2780.999 -> This notion of an operating system.

2783.276 -> This idea that you can take all sorts

2784.109 -> of different applications

2785.53 -> and they can all work seamlessly together.

2788.11 -> To be super clear, the business user

2790.03 -> kind of doesn't care what's under the hood.

2793.245 -> Do I have a steering wheel?

2794.488 -> Do I have an accelerator, brake,

2795.321 -> can I drive the vehicle?

2797.535 -> That's what I want and I want user interface to be agile.

2800.884 -> I want to be able to see what I'm doing

2802.091 -> and I want to be able to function.

2803.14 -> But under the hood,

2804.22 -> we might have bespoke built solution one

2807.52 -> for S&OP planning.

2809.65 -> We might have bought Pega Fuels for fuel optimization.

2813.94 -> We may be using SAP for our ERP system.

2818.47 -> We may be using N4 for our, it doesn't matter.

2822.807 -> We're using all of these different capabilities

2824.17 -> but the desire is to have them integrate,

2826.24 -> to communicate together

2827.32 -> so that we're able to provide the holistic view.

2833.2 -> And when we look at that,

2834.46 -> what I hope is coming across loud and clear here

2837.34 -> is this notion that we're bringing data to action

2840.37 -> to support the business.

2843.782 -> I hope what's come out loud and clear,

2846.76 -> there is no point in surfacing data at all

2851.11 -> if you're gonna do nothing with it.

2854.384 -> Right, pointless.

2855.217 -> If you're gonna let them crash, let them crash.

2857.494 -> If you're gonna go fix it, go fix it.

2860.35 -> And this is always goes back to the analogy

2862.63 -> of air traffic control.

2864.905 -> You've heard people talk about this notion

2865.738 -> of logistics control,

2867.001 -> tower supply chain controls and so forth.

2868.865 -> This is where we are with surfacing data

2870.52 -> and I never understand why it is

2872.5 -> when people use that analogy

2874.387 -> because when you think about it,

2876.82 -> what ends up happening

2877.78 -> is that you have dashboards everywhere

2880.03 -> of crashed airplanes in your business.

2883.18 -> Right?

2885.981 -> And I've seen them right.

2887.005 -> I've spent all my life

2887.838 -> walking into dashboards with crashed airplanes.

2889.33 -> John, you were wrong yesterday because.

2891.562 -> Your performance is down yesterday because.

2895.57 -> But yet the compute part is not helping me

2897.79 -> by looking ahead and giving me an indication

2900.07 -> of what I could do to make a difference.

2902.842 -> So for example,

2903.864 -> "John your delivery accuracy was poor yesterday."

2906.82 -> Thanks.

2908.71 -> Right?

2910.061 -> You know, seriously, thanks.

2911.05 -> Where, "John, pager's going.

2913.877 -> "If you don't put that box on that truck

2916.277 -> "in the next 30 minutes, you're gonna have a DEA hit."

2921.341 -> Thanks, right?

2922.174 -> I can go and help my outbound team.

2923.89 -> Or there's a 95% chance that that parcel

2927.19 -> is not going to be delivered on wave one.

2931.24 -> That's useful.

2932.95 -> Right, that's useful because I can phone ahead.

2935.447 -> "Hey Mrs Customer, are you gonna be in

2937.607 -> "'cause I've got a parcel for you."

2940.701 -> Right, so you can check these things out.

2942.701 -> I was working with one customer,

2943.534 -> their parcels happened to be MRI scanners

2946.03 -> and believe it or not they would arrive at hospitals,

2948.28 -> they'd be like, "Ding dong,

2950.267 -> "I've got an MRI scanner for you."

2952.33 -> Not here gov.

2954.37 -> The whole idea that you would have

2955.78 -> this multimillion dollar piece of equipment arriving

2958.87 -> and they're not ready to receive.

2962.35 -> It's that sort of level and we've all been there right?

2965.202 -> We've all been there with things like wind farms

2966.04 -> and everything else with huge items arriving

2968.32 -> that are not ready to receive.

2971.042 -> So that's about surfacing the data.

2972.01 -> The next thing is the notion of predictive analytics.

2975.13 -> So we surface the data that tells us

2977.979 -> where all our trucks are, that's cool.

2980.378 -> But then we get the prediction that says,

2981.863 -> "The truck's going to arrive at 12 o'clock."

2984.62 -> That's helpful, at least I've got a prediction.

2986.86 -> But it's not really helping us a lot

2989.83 -> because we're not getting any guidance from it.

2993.61 -> You see?

2994.99 -> So here's your trucks,

2998.524 -> the truck is not going to make it

2999.98 -> to the fulfillment center on time.

3001.361 -> There's your ETA.

3004.103 -> Guidance, because the truck is not going to make it

3006.843 -> to the fulfillment center on time,

3008.4 -> you're gonna move it to a secure parking area

3010.98 -> that is 60% the way there

3012.81 -> which is going to enable the driver to rest

3015.26 -> so that it's ready for delivery the next day.

3018.103 -> That's cool, that's actually helpful

3020.999 -> and then when you look at

3022.02 -> the next stage of automation it says,

3023.377 -> "Truck driver one, drive to secure parking area

3026.866 -> "that has been secured for you.

3027.699 -> "You're in parking bay 45 and your new destination will be

3031.057 -> "FC whatever at 12 o'clock tomorrow morning."

3036.02 -> Don't need to talk about it,

3037.143 -> don't need to do anything about it.

3038.242 -> I know what I'm gonna do, it's what I always do.

3039.529 -> That's when it gets helpful.

3043.042 -> And you know, for me,

3043.875 -> when I think about these things, it's really,

3046.17 -> it's simple logic about things like even weather.

3050.4 -> I know you're sad at a British guy

3051.426 -> not talking about the weather yet, it's okay.

3052.259 -> I will.

3053.25 -> If it's gonna be raining tomorrow

3054.81 -> and I've gotta walk to the train station

3056.28 -> and you know I'm going,

3057.39 -> it's okay to tell me it's raining tomorrow.

3060.185 -> It'd be better to say,

3061.202 -> "John bring your umbrella."

3062.685 -> But it would be even better

3063.687 -> if there was an Uber sitting at my door

3065.021 -> when I walked downstairs.

3066.15 -> So it's that ability to be able to step the logic

3069.543 -> and to be able to create the systems that step the logic.

3071.52 -> So if you're sitting there as a developer

3073.476 -> and if you're sitting there as a business user,

3075.72 -> it's okay to try to push the boundary.

3078.48 -> You may not get there yet, it's okay.

3080.781 -> Push the boundary and ask for guidance.

3083.5 -> Don't give me the ETA, tell me the so what.

3085.817 -> Yeah?

3088.64 -> Tell me the so what.

3089.805 -> And if the so what is a logical business piece,

3091.62 -> then let's go build for that.

3093.72 -> So that's whenever we try to move away from insights

3096.36 -> and we try to move much more towards automation.

3099.81 -> But the single most important thing

3101.61 -> underpinning all of this is this.

3106.221 -> You are not, you are not going to be able to start tomorrow

3110.73 -> with a fully automated something in your business.

3115.061 -> All right, so just get that out of your head.

3116.64 -> And nor am I gonna be able to walk into your business

3119.01 -> and say,

3120.359 -> "Oh I've solved every problem in your business.

3122.839 -> "I'll come and fix it for you right now."

3124.759 -> All right, first of all I'm not that arrogant.

3126.403 -> And secondly, I'm gonna fail

3127.41 -> and you're gonna enjoy watching me fail

3129.581 -> if I came in with that attitude anyway.

3130.74 -> So you're gonna fail.

3132.57 -> So what we need to be able to do

3134.04 -> is just take step by step by step

3137.519 -> and progressively get better insights,

3139.443 -> get better predictions, get better outcomes

3141.543 -> and continuously drive step by step by step and keep it at.

3145.11 -> 1%, 1%, 1%.

3147.51 -> That means not having waterfall bills.

3151.05 -> That means helping your IT team recognize

3153.06 -> that we're not having waterfall bills

3154.29 -> no matter which way they want to dress it up.

3156.69 -> Right?

3158.04 -> And it also means when you do bring in SI partners,

3161.239 -> GSI partners, it also means working with them to say,

3163.477 -> "I'm comfortable with you

3165.282 -> "not delivering all of this in one go.

3167.42 -> "Let's get into action.

3169.164 -> "Let's see the benefits, let's see the improvements.

3170.707 -> "Let's accept some risk together.

3173.719 -> "Let's accept some failure together

3174.997 -> "and let's start driving together."

3178.32 -> So if I had any final thoughts in this space

3181.3 -> it would be simply this.

3182.91 -> When we go and start doing all of this,

3186.285 -> there's three things that are gonna happen.

3188.439 -> The first thing that's going to happen

3189.272 -> is that you're going to fail.

3192.823 -> Get comfortable with it, you're gonna fail.

3195.42 -> The very next thing that's gonna happen

3197.28 -> is that someone's gonna be misunderstood.

3200.49 -> Now if I'm misunderstood it's brilliant.

3203.64 -> John's right, the world's wrong and I've got the solution

3207.485 -> and I'm gonna go fix the world.

3208.802 -> That's cool, it's really empowering for me.

3210.18 -> Other people, not so much but I like that notion.

3214.663 -> But as a leader,

3216.27 -> it's actually the other way around.

3219.092 -> What happens when my team,

3220.97 -> I don't understand what they're trying to achieve.

3223.44 -> And what is worse, they keep failing.

3226.682 -> Trying to achieve this thing that I don't understand.

3228.45 -> So how can I as a leader encourage people

3231 -> to work in an environment

3232.17 -> where I don't quite understand

3233.34 -> what they're trying to achieve to help dust them off,

3236.01 -> and say keep going and give them the confidence to try

3238.2 -> and keep experimenting.

3240.188 -> That's the hard bit about being misunderstood yeah?

3243.12 -> And in the final aspect of all of this

3245.28 -> is that innovation isn't something that you can go and buy.

3249.359 -> I tried it, I bought a campervan

3250.2 -> and I thought I would be able to surf immediately, I can't.

3253.41 -> Right?

3254.71 -> I bought a Mac thinking I would suddenly be

3255.543 -> a fantastic engineer and fantastic designer, I'm not.

3259.35 -> So you can't just go plug it in.

3261.6 -> But I can guarantee you now,

3263.34 -> that there is somebody that's just gone

3264.72 -> for a cigarette break somewhere in one of your offices

3267.18 -> that knows how to improve inventory holding by one day.

3271.05 -> And there's somebody that's currently sitting in a truck

3273.24 -> driving along the road

3274.23 -> that knows how to improve fuel optimization

3276.3 -> or price for the fuel if only we did X.

3279.72 -> So how do you get out of their way to help them innovate?

3284.43 -> Okay, anyway, that's me.

3286.11 -> Any questions that people may have,

3288.15 -> I'm happy to take.

Source: https://www.youtube.com/watch?v=cVgcG2c3r4E