← Live demos Free live demo
Codex builds an AI quote generator from one download Download the build pack, give it to Codex, and it builds a working quote generator in your Make account by itself. Your price list lives in Make, the AI picks the items and writes the scope, Make does the maths, and a small app on your computer shows a branded quote you save as a PDF. Then we copy it for a second business.
10 min Codex Make MCP Make AI Data stores Node.js
Your browser does not support video playback. Back 10 sCaptions on 0.75× 1× 1.25× 1.5×
In this demo 00:00 What we’re building, and what you need01:11 One voice prompt to Codex01:55 A blank Make scenario02:47 While Codex builds: the Fixit pack03:20 From a local test app to a product04:03 Use your real supplier prices05:06 What Codex built in Make05:41 The quote desk app07:05 Running a real quote08:02 Make as your back end, and selling it
Keyboard shortcuts When the video or its buttons have focus:
Space or K play or pauseJ back 10 secondsL forward 10 secondsC captions on or offTranscript 00:00 Let's build an automatic quote generator 00:02 for tradesmen or contractors, plumbers, 00:05 electricians, builders, et cetera, just 00:07 using ChatGPT Codex and Make Automations 00:10 platform. 00:11 You'll need a ChatGPT account, and you'll 00:13 need a Make account. 00:14 All the links are in my bio. 00:16 Just click on it, and you can use the 00:18 course pack, which is free, on our 00:20 build.cogniz.online AI course. 00:23 you can download it and immediately start. 00:25 So it'll be, on our course, on our course 00:28 website under live demos. 00:29 You can download this pack, extract it, 00:31 and then you are essentially ready to 00:33 begin. 00:33 So let's head over to ChatGPT to one-shot 00:36 this from, the context that we provide 00:39 with this pack. 00:40 So I'm gonna instruct ChatGPT with voice, 00:43 and then we're gonna ask it to build a 00:45 local version of this. 00:46 So it's not gonna deploy it automatically, 00:48 but so that you can see how an automation 00:50 works for an automatic quote generator 00:52 that uses your prices and your rates, 00:54 which you build into this from context to 00:56 quickly generate a quote for you. 00:58 Now, you can also build this and sell this 01:00 to contractors in your local, in your 01:03 local town or your local area, when you 01:04 deploy a solution like this. 01:06 This is essentially free advice on a 01:08 business that I'm giving away that you can 01:10 start tomorrow. 01:11 Okay, let's head in. 01:12 ChatGPT, so we've got a folder, in, a, a 01:15 pack in this, quote generator folder. 01:17 I want you to examine all of its contents 01:20 and then generate the local host, so that 01:22 we've got a modern interface where we've 01:25 got this automatic quote generator. 01:27 You can invent pricing, from now, but read 01:29 the materials, and then we're gonna deploy 01:31 a Make scenario. 01:33 I'll share the link shortly. 01:35 I want you to start on this blank canvas 01:38 and then build out the Make automation 01:40 from there. 01:41 read everything and get started. 01:43 If you have any clarification questions, 01:45 just ask me anytime. 01:46 Okay. 01:46 I'm gonna stop the voice recording. 01:48 You can automatically send it, but I just 01:50 want to quickly add, the blank Make 01:53 scenario where we're gonna start off. 01:54 we can briefly wait and see if we get any 01:56 questions. 01:57 I'll quickly shift to the Make scenario so 01:59 that you can see it is essentially a, 02:02 absolute blank scenario. 02:03 There you go. 02:04 there's nothing on here. 02:05 ChatGPT will use this now, to generate, 02:09 this, the all the, instructions that are 02:11 included in that pack that you can 02:13 download. 02:14 Again, on our build.cogniz.online, we're 02:17 extending the live demos. 02:18 all the available, material in the live 02:20 demos, which will be free, there's 02:22 obviously exclusive, exclusive content in 02:24 the member section as opposed to, the free 02:27 section, which goes in much more depth. 02:29 but let's see if ChatGPT, is, on its way. 02:32 Yes, it's looking at the inventory, 02:34 looking at the skills, examining our 02:36 skills. 02:36 There it's already finding, the Make 02:38 scenario. 02:38 I'm gonna give it a moment, then I'll 02:39 refresh the Make scenario, and from there 02:41 we can see what, ChatGPT builds directly 02:44 from that without me giving any further 02:46 instructions. 02:47 Okay, ChatGPT's building. 02:49 it identified the pack information. 02:51 Fixit Plumbing is our, is our, fictional 02:53 company that we use throughout the course 02:55 for training. 02:56 so you actually practice on a, on a 02:58 made-up sample company. 03:00 and in this pack that you can download, 03:01 there's ten sample rates, so it'll use 03:03 that in the platform. 03:04 Let's quickly have a look if there's 03:06 anything been... 03:07 that's been populated in the Make scenario 03:10 by, by ChatGPT. 03:11 It's using the MCP. 03:13 it's actually building the quote, desk 03:15 layout, so that'll probably be, yeah, the 03:17 HTML that we can run locally as the demo. 03:20 Now that, that HTML that's locally hosted 03:23 gives you an opportunity to test, the 03:24 interface that can be connected to the 03:26 Make scenario, and that will be the base 03:29 for deploying your solution to the web, if 03:31 you want to actually sell as a service or 03:33 use this, with your customers where they 03:35 can go and generate sample quotes. 03:37 Obviously, when you, when you do that, 03:39 any, any trader, contractor will know that 03:42 there's always installation variables that 03:44 might in- might influence the quote. 03:46 extra longer pipeline runs, different 03:48 cable requirements, components that need 03:50 to be replaced that might not be part of 03:52 the quote. 03:52 So obviously there's a notice that you'll 03:54 give with this, to say, "This is an 03:56 estimate. 03:56 final quote only on inspection after call 03:59 out." but it gives your customers an idea, 04:01 and it can be a tool that you use as well. 04:03 So in the previous video, we did the, 04:06 invoice extractor, invoice or receipt 04:08 extractor. 04:09 So you as a contractor can actually take 04:12 all your other invoices that you, that you 04:15 used when buying from Build It or your, 04:18 whatever your supplier is, whether it's 04:19 the plumbing, electrician, electrical 04:21 industry, building industry. 04:22 You can upload all of those, extract the 04:24 information, and then use that as a source 04:27 of context for your future pricing. 04:28 So then you can actually connect your 04:30 WhatsApp to this as well and build your 04:32 own quoting engine where from real prices 04:34 that you recently paid, you can build 04:37 estimates on the spot if you, if you want 04:39 to quickly do a quote without needing to 04:41 go and sit at the computer, run through 04:43 everything, and then put your margins on 04:45 there. 04:45 So you can expand this, and ChatGPT or 04:47 Claude Code, can do all of that for you. 04:50 This pack that you can download, is the 04:52 starting base for this. 04:53 You can see it says it's, it's actually 04:55 updating the scenario as we speak. 04:58 might have used a different scenario. 04:59 So we'll just wait and see if it refreshes 05:02 this one or if it gives us a link for the 05:05 separate scenario. 05:06 Okay, so ChatGPT actually, created a new 05:09 scenario for us. 05:10 I did not save the previous, blank 05:12 scenario, so it created the new one, and 05:13 I'm gonna quickly show it. 05:15 So this is quite a complicated one that he 05:16 built. 05:16 not that complicated, but, there we've got 05:18 an AI agent in the loop. 05:20 So it receives via the webhook from this 05:22 platform that, that Codex is building on 05:24 the localhost. 05:25 you can see it already tested it. 05:26 It used a hundred and fifty-three credits. 05:28 so it's already been, being tested by 05:31 ChatGPT itself. 05:32 So it's got data stores, and then it saves 05:34 information, analyzes information, looks 05:37 at the pricing tables, if there's errors, 05:39 it moves along. 05:40 let's quickly see what the progress is on 05:43 from ChatGPT's side. 05:45 and it's already started to build our 05:47 localhost. 05:48 so we can see here. 05:49 So it did a mobile version, looked at the 05:51 mobile response, responsiveness. 05:53 Simple. 05:53 Obviously, this is not a full platform. 05:55 It's, just a locally hosted interface so 05:58 that you can play around with and test, 05:59 but this can eventually evolve into, into 06:02 your platform or for your own use or for, 06:04 for reselling it. 06:05 So that would already then be connected. 06:07 We'll just ask ChatGPT to provide us with 06:10 the localhost port. 06:11 Okay, it's actually running on, four eight 06:13 seven zero. 06:14 here we go. 06:15 So that's the localhost running. 06:17 So here's our Fixit, Make automation, and 06:20 this is now actually connected to this 06:22 localhost quote demo. 06:24 All right? 06:24 so we're gonna Here you can enter. 06:26 If you build this out for yourself, you 06:27 can go and play around with this. 06:29 obviously in the pack you'll see the 06:31 format of how the pack is laid out in 06:32 terms of sample pricing. 06:33 you can go and modify those actually to 06:35 your own scenario. 06:36 So I'm gonna just load. 06:37 Got, just got a message from, Codex. 06:39 We'll check it now. 06:40 So you can load the example and then run 06:41 it. 06:42 but let's quickly just see what the 06:43 feedback is from, from ChatGPT. 06:45 It's finished. 06:46 so it built the localhost. 06:47 we can load the example, prepare the 06:49 quote, and save it as a PDF. 06:51 it's connected to the active Make scenario 06:53 which it built, and this was all Codex, 06:54 right? 06:55 So from that prompt packed, which 06:57 structured it. 06:58 Obviously, my environment is set up, 06:59 correctly, which is part of the courses 07:01 that we do in terms of getting everything 07:03 connected so that this can happen. 07:04 But let's quickly look at the, the 07:06 localhost. 07:07 Okay, so we've populated the, the sample 07:09 details here, arbitrary information, and 07:12 then we are going to ask it to prepare the 07:15 quote. 07:15 Now, this information is being sent to 07:17 Make. 07:17 Right. 07:18 So we can actually see, we'll be able to 07:20 see the scenario run. 07:21 There it's going. 07:22 All right. 07:22 So it's actively being run now. 07:24 So we can look at the history, and there 07:26 you can see, instant success. 07:28 Literally just ran. 07:29 And here is, here is the results. 07:32 So that's the quote. 07:33 Fixit Plumbing, fictional company, quote, 07:35 fictional numbers. 07:36 And here we've got the test client, all of 07:39 this test details that were in here. 07:40 scope of work based on what happened here. 07:43 And then if you've got your pricing, it'll 07:45 s- it'll give you the call-out fees. 07:46 It'll give you all the pricing that you 07:48 add, your components that you need to add 07:50 based on your pricing that you provided, 07:52 your service fees, et cetera, and provide 07:54 an estimated quote. 07:55 Now, this is obviously an arbitrary 07:56 number, VAT included, et cetera, and the 07:58 notes that you need. 07:59 So now you can go and play around with 08:01 this. 08:01 and that is essentially how you connect 08:04 your, your platform or a platform to a 08:06 Make scenario that in a s- in a sense can 08:09 run a, a semi back end for you. 08:11 A lot of, logic can be run there, 08:13 analysis, document extraction. 08:15 You can connect information, databases, 08:17 and then you can have an AI agent 08:19 orchestrate and, actually work for you. 08:22 So this is, this is an administrative 08:24 role. 08:24 We've got an admin, AI agent in here, 08:27 running on a prompt connected to data that 08:30 performs a real task. 08:32 Now, this run ins- runs instantly, so if 08:34 you give it the right information, your 08:36 pricing information, your service fees, 08:38 your call-out fees, distance travels, and 08:40 kilometers that you add additionally, 08:43 you've got a, a quick and easy way to 08:45 generate prompts on-site when you're at a 08:47 customer or when you are overburdened with 08:50 a lot of quote requests. 08:51 and then you can go further, and you can 08:52 expand on this. 08:53 You can then automatically, when that 08:55 quote is accepted, save that information. 08:57 you can then, work that into your 08:59 accounting, whatever you wanna do. 09:00 The, the, the limits of this platform is, 09:03 is almost limitless. 09:04 Your limitation is gonna be your 09:06 imagination in trying to solve the 09:08 problems that you face daily within your 09:11 business or what, the, the problems that 09:13 your customers face, which you can solve 09:14 using something like this. 09:16 so again, links are in the bio. 09:18 you can go and download this. 09:19 You can go and play around with this. 09:20 You can go and sell this if you deploy it. 09:23 So you can deploy to AWS or whatever 09:25 hosting platform you wanna use. 09:26 and you can actually approach local 09:27 businesses to start selling this. 09:29 I might do the same. Build pack · ZIP · 15 KB
Download the build pack The same pack used in this demo: instructions for your AI, the build specification, the app and the tests.
Unzip it and open START-HERE.md. Put your business details and prices in the business folder. Open Codex in the folder and paste the prompt from CODEX-PROMPT.txt. When Codex is done, run node app/server.mjs and open http://localhost:4870. Your AI never needs your passwords or keys: it uses the connections already in your Make account. Use made-up client details while you test.
Download the packWant to build this yourself? The Cogniz course teaches each piece step by step: connecting ChatGPT or Claude to Make, reading a scenario, and checking what the AI built before you rely on it.
Try a free step → See membership