WEBVTT 00:00.145 --> 00:04.053 At eleven forty at night, a restaurant sends a voice note: 00:04.053 --> 00:05.908 two cases of cooking oil, 00:05.908 --> 00:09.408 five kilos of chicken, and "the usual" sauces. 00:09.408 --> 00:10.487 Tomorrow morning, 00:10.487 --> 00:13.764 someone has to work out what "the usual" means. 00:13.764 --> 00:17.927 Here's how AI can turn a distributor around, step by step. 00:18.305 --> 00:20.621 Picture a food supplies distributor. 00:20.621 --> 00:24.450 Thirty staff, a few hundred restaurant and café customers. 00:24.450 --> 00:28.411 Orders arrive every way imaginable: chat messages, 00:28.411 --> 00:32.542 voice notes, photos of handwritten lists, phone calls. 00:32.542 --> 00:34.898 Every one gets retyped by hand. 00:34.898 --> 00:36.845 Late-night orders get missed. 00:36.845 --> 00:38.266 Wrong items go out. 00:38.266 --> 00:41.161 The phone rings all morning with "where's my delivery?". 00:41.161 --> 00:42.516 Payments come in late. 00:42.516 --> 00:45.796 And the order desk is under constant pressure. 00:46.325 --> 00:48.759 It starts by sitting with the order desk: 00:48.759 --> 00:50.641 counting how orders arrive, 00:50.641 --> 00:53.049 how long each one takes to key in, 00:53.049 --> 00:55.312 and how many need correcting later. 00:55.312 --> 00:57.970 Then the awkward cases: "the usual", 00:57.970 --> 00:59.628 an item that's out of stock, 00:59.628 --> 01:01.404 an urgent same-day delivery, 01:01.404 --> 01:03.536 a new customer with no account, 01:03.536 --> 01:05.694 an account over its credit limit. 01:05.694 --> 01:08.365 Each step gets sorted: automate it, 01:08.365 --> 01:12.158 let AI do it with a person checking, or keep it human. 01:12.705 --> 01:16.298 Then the AI learns what an experienced order clerk knows: 01:16.298 --> 01:18.403 every product and pack size, 01:18.403 --> 01:21.376 each customer's usual order and delivery days, 01:21.376 --> 01:24.100 which items can be swapped when stock runs out, 01:24.100 --> 01:26.771 the cut-off times, the credit terms. 01:26.771 --> 01:29.749 Now "the usual" actually means something. 01:30.319 --> 01:31.608 Here's the heart of it. 01:31.608 --> 01:33.648 Whether an order comes as a message, 01:33.648 --> 01:35.411 a voice note in Malay, 01:35.411 --> 01:39.174 English or Mandarin, or a photo of a handwritten list, 01:39.174 --> 01:43.042 the AI turns it into a draft order: products matched, 01:43.042 --> 01:46.661 quantities checked, "the usual" filled in from history. 01:46.661 --> 01:49.924 And here's the expert part: it shows how sure it is. 01:49.924 --> 01:53.240 Anything it isn't sure about goes to a person to check. 01:53.240 --> 01:54.858 Then the order goes straight into 01:54.858 --> 01:57.213 the distributor's existing system, 01:57.213 --> 01:59.717 and the customer gets a confirmation. 02:00.258 --> 02:02.982 Calls get answered too, at any hour. 02:02.982 --> 02:04.324 "Do you have stock?" 02:04.324 --> 02:06.192 "Can I add one more case?" 02:06.192 --> 02:07.706 "What time is my delivery?" 02:07.706 --> 02:11.416 The AI answers from live stock and delivery information. 02:11.416 --> 02:14.880 Price disputes and account questions go to a person. 02:15.425 --> 02:17.070 Most "where's my delivery?" 02:17.070 --> 02:19.952 calls disappear when customers simply know. 02:19.952 --> 02:21.675 So each customer gets a message 02:21.675 --> 02:23.833 when their delivery is on the way, 02:23.833 --> 02:25.359 with an arrival window. 02:25.359 --> 02:27.688 Drivers capture proof of delivery. 02:27.688 --> 02:30.794 If an item was short, it's flagged straight away, 02:30.794 --> 02:33.732 not discovered weeks later in a dispute. 02:34.258 --> 02:37.140 Invoices are built from what was actually delivered, 02:37.140 --> 02:38.811 so they're right the first time. 02:38.811 --> 02:42.114 Customers get a friendly reminder before a payment is due, 02:42.114 --> 02:44.087 by message and email. 02:44.087 --> 02:46.258 And if an account goes over its limit, 02:46.258 --> 02:48.995 a person reviews it before the next order. 02:48.995 --> 02:53.038 No harsh automation, just no more forgotten invoices. 02:53.592 --> 02:55.026 Then the relationship. 02:55.026 --> 02:56.948 The system notices patterns, 02:56.948 --> 02:59.053 so a customer who orders oil every 02:59.053 --> 03:02.198 Monday gets a timely nudge: "the usual?". 03:02.198 --> 03:04.040 Customers who've opted in hear about 03:04.040 --> 03:06.448 new products that suit their kitchen. 03:06.448 --> 03:09.079 And new restaurants find the distributor too, 03:09.079 --> 03:10.776 through an online catalogue, 03:10.776 --> 03:12.579 a proper business profile, 03:12.579 --> 03:14.224 and ads aimed at new restaurants 03:14.224 --> 03:16.925 opening in its delivery areas. 03:17.458 --> 03:18.419 And the team? 03:18.419 --> 03:20.827 Order clerks stop retyping and start 03:20.827 --> 03:24.024 handling the exceptions and the relationships. 03:24.024 --> 03:27.721 Sales reps visit customers instead of chasing orders. 03:27.721 --> 03:30.550 The warehouse gets clean pick lists earlier. 03:30.550 --> 03:33.659 When the pressure comes off, good people stay. 03:34.205 --> 03:36.429 One delivery route goes first. 03:36.429 --> 03:37.824 Everything is measured and 03:37.824 --> 03:40.314 adjusted before the rest follow. 03:40.838 --> 03:42.944 So here's the end of the turnaround. 03:42.944 --> 03:45.733 Orders in within minutes, in any form. 03:45.733 --> 03:48.470 Every order captured, even at midnight. 03:48.470 --> 03:49.720 Fewer mistakes. 03:49.720 --> 03:52.273 Customers who know where their delivery is. 03:52.273 --> 03:53.588 Faster payments. 03:53.588 --> 03:55.579 And a team that wants to stay. 03:56.138 --> 03:58.507 That's what AI can do for a distributor. 03:58.507 --> 04:01.441 It doesn't replace the people who know your customers. 04:01.441 --> 04:04.178 It takes away the retyping and the chasing, 04:04.178 --> 04:05.691 so they can look after them. 04:05.691 --> 04:06.691 If you're wondering what this 04:06.691 --> 04:08.349 could look like in your business, 04:08.349 --> 04:11.340 I'm always happy to talk it through on LinkedIn.