WEBVTT - The ROI Rules of AI: Procuring Success (Sponsored Content)

0:00:00.480 --> 0:00:03.960
<v Speaker 1>Since you're a subscriber to this Bloomberg podcast, we thought

0:00:04.000 --> 0:00:07.480
<v Speaker 1>you'd be interested in a new four episode sponsored podcast

0:00:07.600 --> 0:00:12.640
<v Speaker 1>called The ROI Rules of AI, produced by IBM and

0:00:12.640 --> 0:00:16.759
<v Speaker 1>Bloomberg Media Studios. It explores how business leaders are thinking

0:00:16.800 --> 0:00:22.000
<v Speaker 1>about the return on investment of artificial intelligence projects. You

0:00:22.000 --> 0:00:26.439
<v Speaker 1>can subscribe wherever you listen to your favorite podcasts. Here's

0:00:26.560 --> 0:00:31.720
<v Speaker 1>a recent episode. Imagine you work in the procurement office

0:00:31.760 --> 0:00:34.839
<v Speaker 1>of a major company. You've been assigned to find a

0:00:34.920 --> 0:00:39.600
<v Speaker 1>supplier for a key component of your flagship product. You

0:00:39.680 --> 0:00:43.239
<v Speaker 1>need to limit your company's risk, so you begin by

0:00:43.320 --> 0:00:50.519
<v Speaker 1>asking is a potential supplier financially healthy? Are they being sued?

0:00:52.240 --> 0:00:56.360
<v Speaker 1>How do they score on environmental, social and governance metrics?

0:00:57.160 --> 0:00:59.840
<v Speaker 1>What are the odds that supplier could be temporarily shut

0:00:59.920 --> 0:01:06.600
<v Speaker 1>down by a war or a hurricane? And those are

0:01:06.640 --> 0:01:09.640
<v Speaker 1>just some of the questions you'd have to answer. It

0:01:09.680 --> 0:01:14.399
<v Speaker 1>could take you days to thoroughly investigate just one potential supplier.

0:01:15.160 --> 0:01:19.880
<v Speaker 2>The problem was efficiency. What we found is that every

0:01:19.959 --> 0:01:22.320
<v Speaker 2>one of these tasks are pretty time consuming.

0:01:23.920 --> 0:01:28.319
<v Speaker 1>That's Gary Kotovitz, chief data and analytics officer at Dunham Bradstreet.

0:01:29.000 --> 0:01:31.800
<v Speaker 1>His company is just out with a new product powered

0:01:31.840 --> 0:01:37.080
<v Speaker 1>by artificial intelligence that enables procurement professionals to research suppliers quickly.

0:01:38.520 --> 0:01:40.840
<v Speaker 1>This is the story of how they built it and

0:01:40.959 --> 0:01:48.040
<v Speaker 1>what they and their clients learned along the way. From

0:01:48.080 --> 0:01:51.960
<v Speaker 1>IBM and Bloomberg Media Studios. This is the ROI Rules

0:01:52.080 --> 0:01:57.160
<v Speaker 1>of AI and I'm your host, Edward Adams. On this podcast,

0:01:57.320 --> 0:02:01.120
<v Speaker 1>we're exploring how organizations of all sizes are using AI

0:02:01.320 --> 0:02:05.160
<v Speaker 1>to transform their operations, aiming to increase their return on

0:02:05.280 --> 0:02:15.800
<v Speaker 1>investment and that of their customers. There's no more storied

0:02:15.880 --> 0:02:18.480
<v Speaker 1>company in financial data than Done In Bradstreet.

0:02:19.000 --> 0:02:21.840
<v Speaker 2>Dun and Bradstreet is a data and analytics company that's

0:02:21.919 --> 0:02:25.040
<v Speaker 2>been around for almost two hundred years. We collect data

0:02:25.200 --> 0:02:28.760
<v Speaker 2>on over five hundred and ninety million private companies and

0:02:28.880 --> 0:02:34.640
<v Speaker 2>we provide our customers insights into supply chain management, credit decisioning,

0:02:34.880 --> 0:02:37.400
<v Speaker 2>lending decisioning, and sales and marketing.

0:02:38.520 --> 0:02:41.560
<v Speaker 1>Whether you're buying or selling, you need the kind of

0:02:41.639 --> 0:02:45.799
<v Speaker 1>information Done In Bradstreet collects. Sales staff use it to

0:02:45.919 --> 0:02:50.160
<v Speaker 1>prospect for potential customers, Banks use it to assess the

0:02:50.280 --> 0:02:53.560
<v Speaker 1>credit worthiness of a company applying for a loan, and

0:02:53.720 --> 0:02:57.960
<v Speaker 1>procurement professional to use it to de risk their supply chains,

0:02:58.560 --> 0:03:01.440
<v Speaker 1>and if the pandemic taught company is anything, it's that

0:03:01.560 --> 0:03:06.200
<v Speaker 1>supply chains have a host of risks, both foreseen and unforeseen.

0:03:07.080 --> 0:03:10.480
<v Speaker 1>It's the job of the procurement staff to anticipate what

0:03:10.720 --> 0:03:15.040
<v Speaker 1>could go wrong and mitigate those risks. Dunham brad Street

0:03:15.080 --> 0:03:18.440
<v Speaker 1>has long provided access to its data cloud through its

0:03:18.480 --> 0:03:23.400
<v Speaker 1>own digital interface and through third party procurement applications. A

0:03:23.480 --> 0:03:27.640
<v Speaker 1>procurement staffer researching a potential supplier, might I want to look.

0:03:27.560 --> 0:03:29.840
<v Speaker 2>At their EHG score, I want to look at their

0:03:30.280 --> 0:03:34.239
<v Speaker 2>credit score, I want to look at their supply chain profile,

0:03:34.480 --> 0:03:36.600
<v Speaker 2>or I want to look at where they're physically located.

0:03:36.720 --> 0:03:41.600
<v Speaker 2>So all those lookups that you would typically do take time.

0:03:48.360 --> 0:03:51.880
<v Speaker 1>To save procurement staff time. Dun and brad Street worked

0:03:51.920 --> 0:03:55.720
<v Speaker 1>with IBM and it's Watson x AI and data platform

0:03:56.080 --> 0:04:00.840
<v Speaker 1>to create a new natural language interface called Ask Procurement,

0:04:01.240 --> 0:04:04.800
<v Speaker 1>where procurement officers can ask questions as simple as.

0:04:05.080 --> 0:04:08.760
<v Speaker 2>Give me everything I need to know about company ABC.

0:04:09.560 --> 0:04:13.120
<v Speaker 1>Or staff can search for all their specific procurement criteria

0:04:13.280 --> 0:04:17.440
<v Speaker 1>at once, such as asking for widget manufacturers which have

0:04:17.839 --> 0:04:22.160
<v Speaker 1>strong credit, low debt to equity ratios, and are minority

0:04:22.279 --> 0:04:26.240
<v Speaker 1>owned from an initial list of suppliers generated by ask,

0:04:26.360 --> 0:04:31.200
<v Speaker 1>procurement staff can further narrow the prospects by asking additional questions.

0:04:32.240 --> 0:04:35.320
<v Speaker 1>The product took about six months to build and began

0:04:35.480 --> 0:04:38.960
<v Speaker 1>being offered to customers in early November. It's already paid

0:04:39.000 --> 0:04:41.880
<v Speaker 1>dividends for dun and Bradstreet. According to Code Ofmits.

0:04:42.600 --> 0:04:46.440
<v Speaker 2>Their return investment is two things. Accuracy as it relates

0:04:46.480 --> 0:04:49.320
<v Speaker 2>to their decision making. Do I have all the information

0:04:49.880 --> 0:04:52.560
<v Speaker 2>readily available to me in order to make the right decision?

0:04:52.640 --> 0:04:55.839
<v Speaker 2>The second is efficiency and productivity.

0:04:56.800 --> 0:04:59.640
<v Speaker 1>In the process of working with customers to build the product,

0:05:00.040 --> 0:05:03.479
<v Speaker 1>Done and Bradstreet learned a lot about customer workflows.

0:05:04.000 --> 0:05:06.640
<v Speaker 2>You start to understand do you typically look for an

0:05:06.839 --> 0:05:09.240
<v Speaker 2>HG score and a credit profile or do you typically

0:05:09.320 --> 0:05:12.960
<v Speaker 2>look for an EHG score and let's say corporate ownership

0:05:13.320 --> 0:05:17.000
<v Speaker 2>and those two questions the most important to majority of

0:05:17.080 --> 0:05:20.680
<v Speaker 2>our customers or is it something else so that starts

0:05:20.720 --> 0:05:23.800
<v Speaker 2>to overtime give you a lot of sort of intelligence

0:05:23.839 --> 0:05:27.720
<v Speaker 2>around how your customers interact with your data and the

0:05:27.839 --> 0:05:29.880
<v Speaker 2>kind of workflows you need to design.

0:05:30.880 --> 0:05:34.560
<v Speaker 1>And the customers also got an education about what generative

0:05:34.600 --> 0:05:36.480
<v Speaker 1>AI can and can't do.

0:05:37.160 --> 0:05:40.600
<v Speaker 2>Jennai itself, as we know, is a brand new concept

0:05:40.960 --> 0:05:44.880
<v Speaker 2>for many customers, and I think one of our biggest

0:05:44.960 --> 0:05:49.040
<v Speaker 2>challenges as we were building it is getting people to

0:05:49.160 --> 0:05:51.480
<v Speaker 2>understand the kind of value it can provide them. Now

0:05:51.520 --> 0:05:54.520
<v Speaker 2>that you know what it can do, customers have this

0:05:54.839 --> 0:05:57.480
<v Speaker 2>sort of aha moment and then from there they start

0:05:57.520 --> 0:06:00.440
<v Speaker 2>to kind of say, Okay, well I understand it, so

0:06:00.720 --> 0:06:02.240
<v Speaker 2>this is everything I want out of it.

0:06:03.200 --> 0:06:05.680
<v Speaker 1>Early users of the product have found that they are

0:06:05.760 --> 0:06:09.000
<v Speaker 1>reducing the time it took them to vet potential vendors

0:06:09.400 --> 0:06:12.120
<v Speaker 1>by an average of ten to twenty percent. Code of

0:06:12.160 --> 0:06:16.640
<v Speaker 1>It says in sizable companies where the procurement team can

0:06:16.720 --> 0:06:20.800
<v Speaker 1>number in thousands, that's the significant savings which can be

0:06:20.960 --> 0:06:31.919
<v Speaker 1>used to address more strategic procurement issues. Dun Bradstreet chose

0:06:32.000 --> 0:06:35.800
<v Speaker 1>IBM because it could play multiple roles in the process

0:06:35.880 --> 0:06:37.080
<v Speaker 1>of creating the product.

0:06:38.080 --> 0:06:44.800
<v Speaker 2>So IBM an amazing partner, and they partner with their

0:06:44.880 --> 0:06:50.120
<v Speaker 2>customers I think from multiple different dimensions. One is they

0:06:50.160 --> 0:06:55.560
<v Speaker 2>are a obviously technology provider. IBM is also a customer.

0:06:56.160 --> 0:06:59.960
<v Speaker 2>They are a consumer of this procurement product. There's certain

0:07:00.120 --> 0:07:05.440
<v Speaker 2>expertise that they brought. So as we started to use

0:07:05.800 --> 0:07:09.280
<v Speaker 2>Watson X platform and the tech that related to it.

0:07:09.960 --> 0:07:13.640
<v Speaker 2>They have a build team that helped us gather the

0:07:13.720 --> 0:07:16.720
<v Speaker 2>requirements as well as actually develop.

0:07:17.760 --> 0:07:21.920
<v Speaker 1>Dunn and Bradstreet's experience building ass procurement holds lessons for

0:07:22.040 --> 0:07:26.480
<v Speaker 1>other companies starting their AI journeys. According to Dave McDonald,

0:07:26.960 --> 0:07:30.680
<v Speaker 1>general manager of the US Industry Market for IBM.

0:07:31.120 --> 0:07:36.160
<v Speaker 3>First, I would say most transformational projects, like starting with

0:07:36.240 --> 0:07:39.960
<v Speaker 3>generative AI, are all about people, process and technology.

0:07:40.360 --> 0:07:42.000
<v Speaker 2>So let's start with people in process.

0:07:43.400 --> 0:07:48.720
<v Speaker 3>AI shouldn't be an it only led initiative because it

0:07:48.800 --> 0:07:51.600
<v Speaker 3>kind of becomes a science project and rarely gets to

0:07:51.720 --> 0:07:54.360
<v Speaker 3>that business benefit and the return on investment that people

0:07:54.400 --> 0:07:55.560
<v Speaker 3>are looking.

0:07:55.360 --> 0:07:56.520
<v Speaker 2>For that drive the value.

0:07:57.080 --> 0:08:00.040
<v Speaker 3>So our suggestion is you always need to have a

0:08:00.200 --> 0:08:03.400
<v Speaker 3>line of business sponsor who is going to directly benefit

0:08:03.480 --> 0:08:07.040
<v Speaker 3>from the outcome of the AI project. And it can't

0:08:07.160 --> 0:08:10.960
<v Speaker 3>just be kind of a simplistic ask a question to

0:08:10.960 --> 0:08:13.000
<v Speaker 3>get an answer. It's got to impact and change a

0:08:13.080 --> 0:08:16.400
<v Speaker 3>business process. So people in process are number one. Number

0:08:16.480 --> 0:08:21.160
<v Speaker 3>two is a large language model. If you're using one

0:08:21.200 --> 0:08:24.600
<v Speaker 3>that everybody has access to, is giving everybody the same answers.

0:08:24.960 --> 0:08:28.360
<v Speaker 3>It doesn't really give you competitive advantage. So being able

0:08:28.440 --> 0:08:33.640
<v Speaker 3>to combine private data that others don't have access to

0:08:33.960 --> 0:08:37.880
<v Speaker 3>with the traditional large language model capabilities of natural language

0:08:37.960 --> 0:08:42.080
<v Speaker 3>processing and speech that is what's going to drive it done.

0:08:42.160 --> 0:08:45.360
<v Speaker 1>Bradstreet is now turning its attention to creating Phase two

0:08:45.800 --> 0:08:49.720
<v Speaker 1>of ass Procurement, which will enable customers to integrate their

0:08:49.800 --> 0:08:53.120
<v Speaker 1>own data about suppliers into the data that done in

0:08:53.200 --> 0:09:00.319
<v Speaker 1>Bradstreet provides. Code. Itz believes that increasingly procurement department will

0:09:00.360 --> 0:09:04.120
<v Speaker 1>allow staff from other departments to interact with the product,

0:09:04.679 --> 0:09:06.559
<v Speaker 1>saving them yet more time.

0:09:07.640 --> 0:09:10.960
<v Speaker 2>The stakeholders are able to ask and get the questions

0:09:11.480 --> 0:09:16.440
<v Speaker 2>answered themselves. That alleviates a lot of the unnecessary tasks

0:09:16.480 --> 0:09:19.120
<v Speaker 2>that a procumber professional is engaged with today, which is

0:09:19.200 --> 0:09:21.560
<v Speaker 2>answering questions about where's my order.

0:09:27.520 --> 0:09:31.120
<v Speaker 1>This has been the ROI Rules of AI, a podcast

0:09:31.200 --> 0:09:34.880
<v Speaker 1>from IBM and Bloomberg Media Studios. If you like what

0:09:35.000 --> 0:09:39.160
<v Speaker 1>you hear, subscribe and leave us a review. I'm Edward Adams.

0:09:39.679 --> 0:09:40.840
<v Speaker 1>Thanks for listening.