AI module makers are turning a screenshot, a prompt, or a Figma file into a production ready, fully configurable HubSpot module in minutes instead of days, and that is changing what agencies charge for, what marketers wait on, and what developers actually spend their time doing.
Building a custom HubSpot module has always meant a fairly linear, manual process: a designer hands off a comp, a developer translates it into HTML, CSS, and JavaScript, then builds out the module fields, tests it across breakpoints, and pushes it to the Design Manager. Even experienced developers spend hours, sometimes days, rebuilding the same patterns (accordions, filtered grids, multi image sections) because each one still has to be coded and field mapped by hand.
That labor shows up directly in pricing. HubSpot developer rates typically run $75 to $250 an hour depending on experience and location, with fixed price custom builds ranging from a few thousand dollars into the tens of thousands for complex, integration heavy work.1 Individual custom modules commonly land between $500 and $2,500 depending on complexity, and even a landing page build can run $1,500 to $15,000 once design and development are combined.2
Joe Jerome, CRO of Computan, put a number on it during a recent episode of the company's Unwatchable podcast: "I know a complex mega nav would be around two thousand dollars. I know modules are gonna be, you know, anywhere from five hundred to twenty five hundred dollars depending on complexity, depending on testing and whatnot."
That cost and turnaround time is also why marketers so often end up waiting on developers for changes that feel simple on their end, like adjusting a layout option or adding a new content block, but require a code change on the module's back end.
An AI module maker is a tool that takes a plain language prompt, a screenshot, or a design file and generates a working HubSpot module from it, including the front end code and the module's configuration fields, rather than just a static mockup. Instead of typing "build me a pricing table," a user can hand the tool an image of an existing page section and ask it to recreate that layout, complete with editable content and layout options.
This is a meaningfully different category from generic AI code generation. A chatbot can write HTML and CSS for a section, but that output usually has to be manually wired into HubSpot's module and theme structure before a marketer can safely edit it. An AI module maker is built to skip that gap: it outputs a module that is already structured the way HubSpot expects, ready to drop into a page.
The shift lines up with what is already happening across software development more broadly. Roughly 84 percent of developers now use or plan to use AI coding tools, up from 76 percent in 2024, and a large share of shipped code in 2026 has some AI involvement.4 HubSpot specific tooling is simply applying that same shift to a narrower, more structured problem: building modules that fit a specific CMS.
Computan's internal AI module maker, built during an internal hackathon led by the company's CMS team, offers a chat style interface: describe the module or section you want, and it generates the code and pushes it straight to HubSpot's Design Manager.
Jerome demonstrated this live on the podcast using a screenshot of a skills marketplace page from Databox. The prompt asked for an identical layout with infinite scroll, header content, filters on each skill, and a hover animation for any item marked as featured, populated with 18 sample skills.
"I can make this thing in three minutes with two dollars worth of credits and one paragraph." - Joe Jerome, CRO of Computan
The result reproduced the original page's filtering, progressive search, card layout, and infinite scroll, including responsive behavior on mobile, without a developer touching a code editor. Sajeel Qureshi, CEO of Computan, reacted to the demo directly:
"What it did with one prompt is incredible. I mean there's just nothing that I've ever seen like that before. I've seen a lot. I haven't seen everything, but I've seen a lot. And I haven't seen anything like that." - Sajeel Qureshi, CEO of Computan
A second example from the same episode pushed the tool further. Given only a screenshot of a client's site colors and a short paragraph asking for "something that's gonna drive leads in a relevant way before he meets with people," the tool built a full interactive assessment module, complete with a scoring system and its own set of qualifying questions the client had never specified.
What separates this from a raw AI generated snippet is what happens after the code is written. The module maker in the demo automatically generated configurable module fields, including nested objects and repeaters for the marketplace's individual skill cards, so a marketer could add or edit skills without touching code. It also generated responsive layouts and multiple content arrangement options (photo top or bottom, 50/50 split, quote block on or off, accent color) inside a single flexible module, all natively compatible with HubSpot's Design Manager.
That native compatibility is the real unlock. Once the module exists, it behaves like any other custom built module: a marketing team can log in and update it directly, with no further AI or developer involvement needed to keep it running.
| Aspect | Traditional module development | AI module maker |
|---|---|---|
| Starting point | Design comp or written spec | Screenshot, prompt, or Figma file |
| Typical turnaround | Hours to days | Minutes |
| Typical cost per module | $500 to $2,5002 | A few dollars in usage credits |
| Fields and configuration | Manually built by a developer | Generated automatically, including nested objects |
| Editability after handoff | Native to HubSpot once built | Native to HubSpot once generated |
| Best suited for | Deep integrations, complex logic, edge cases | Standard sections, layouts, and marketplace style modules |
The module maker itself was built by Computan's own developers, not generated end to end by AI. Jerome was direct about that on the podcast: "This wasn't all just generated with AI, right? This probably had some AI help. Our developers made it." He was equally direct that the output of the tool still needs a human in the loop before it ships to a client: "These apps are still gonna be guided strategically and ultimately with a code review by real humans."
Qureshi framed the longer term shift in terms of what developers spend their time on, not whether they are needed at all: "People using the tools that are that they're building, helping them use them to do things better. That's what it's gonna come down to."
In practice, that means less time spent hand coding repetitive sections and more time on the harder eleven and twelve out of ten problems, as Jerome put it: complex integrations, custom logic, and building the AI powered tools themselves, rather than repeatedly building the same components.
Computan's team is already planning the next iteration: a tool that inspects a site's existing template code and theme files directly, without needing a screenshot at all, and generates full pages from a URL or a Figma link. A migration focused version of the same workflow is also in progress, aimed at moving existing websites onto HubSpot's CMS with less manual rebuilding.
Jerome acknowledged how far reaching that could be, including for HubSpot's own template and theme marketplace: "I really do. It's a very scary thing to contemplate that we could make something that would take out the legendary HubSpot marketplace." Qureshi agreed the theme and template side of the marketplace is the part most exposed, while the app marketplace stays intact: "The app marketplace would still be there, but the theme marketplace, yeah, you can just add prompt everything."
The direction of travel here is not unique to HubSpot. The broader low code and no code market is valued at roughly $65 billion in 2026 and growing at a compound annual rate above 24 percent, and Gartner has projected that 70 percent of new enterprise applications will use low code or no code approaches.56 Developers using AI coding tools report task level speed gains in the 25 to 55 percent range for scoped work like this.3
For agencies, that translates into a few concrete opportunities: faster project delivery on the module and section work that used to eat the most hours, lower production costs that can be passed on or kept as margin, and the ability to offer more genuinely custom work instead of pushing clients toward rigid template packs. It also opens the door to new AI powered services agencies can package and sell on their own, the way Computan is positioning its own module maker for early access and beta testing among its community.
None of this removes the need for review. Industry data on AI generated code more broadly is a useful caution here: in controlled testing across more than 100 models and 80-plus coding tasks, 45 percent of AI generated code introduced a known security vulnerability, and code acceptance rates for AI suggestions still sit well under half in professional settings.4 That is precisely why Jerome was careful to frame the module maker's output as something that still goes through code review, not something that ships untouched.
The open questions for AI module makers specifically are about governance and depth: how well generated code holds up under security and accessibility review, how it handles complex integrations that go beyond a single page section, and how consistently it maintains a brand's specific design system across dozens of generated modules rather than one. Jerome's own read on this was candid: there is still work at "the eleven and twelves on this ten point scale," meaning the deep integration and strategy work agencies are built to handle.
The likely trajectory looks less like developers being displaced and more like developers becoming the people who architect, review, and extend what AI tools generate, while marketers gain more direct creative freedom over layout and content without waiting on a build queue. Qureshi summed up the mood around the demo simply, "Very exciting stuff, sir. Very exciting. It is exciting." Jerome's response captured the flip side just as honestly: "Exciting and scary. It's all those things."
If AI module makers keep improving at the pace shown in these early demos, tools like this may end up as common in HubSpot development workflows as page builders and drag and drop editors are today.
AI module makers are not primarily a story about developers typing less code. They are reducing the repetitive build work that has always eaten the most hours in custom HubSpot development, while giving both developers and marketers more room to build genuinely sophisticated, on brand experiences. The biggest opportunity in this shift is not replacing developers. It is giving them more time to solve the integration, architecture, and strategy problems that actually require human expertise.
What is an AI module maker for HubSpot?
An AI module maker is a tool that turns a prompt, screenshot, or design file into a working HubSpot module, generating the HTML, CSS, JavaScript, and module fields automatically and pushing the result directly to HubSpot's Design Manager.
How is this different from just asking an AI chatbot to write code?
A general AI chatbot can generate front end code, but it usually still needs to be manually rebuilt into HubSpot's module and field structure. An AI module maker outputs a module that is already structured for HubSpot, including configurable fields and nested objects, so it is editable natively without further developer work.
Will AI module makers replace HubSpot developers?
Not based on how these tools are currently being used. The module maker discussed here was itself built by developers, and its output still goes through human code review before it ships. The role is shifting toward architecture, integrations, and review rather than repetitive module coding.
How fast can an AI module maker actually build a module?
In a live demo, a complex, multi component marketplace style module with filtering, infinite scroll, and responsive layouts was generated from a single screenshot and one paragraph of instructions in about three minutes.
Can marketers edit modules built this way without a developer?
Yes. Because the generated modules include native HubSpot fields, layout options, and content areas, marketing teams can update and maintain them directly inside HubSpot without going back to AI or a developer for routine changes.
Can AI eventually build entire HubSpot pages, not just modules?
That is the direction this technology is heading. Planned next steps include generating full pages from a URL, template, or Figma link, and using similar AI assisted workflows for website migrations onto HubSpot's CMS.
What should agencies watch out for with AI generated modules?
Code quality and security review, accessibility, complex third party integrations, and maintaining consistent brand design across many generated modules are all areas where experienced developer oversight still matters.
Sources: