Today’s newsletter is a collaboration post from Brian Balfour, who is the founder and CEO of Reforge, which has recently launched several AI-native SaaS tools for product teams. One of those is Reforge Build, an AI prototyping tool built for PMs. It’s designed for prototyping based on your real product. Capture a page with the browser extension, then start iterating on a feature that looks and feels exactly like your own product.
🎁You can try it free here and get a month of premium for free with the code PRODUCTIFY
Everyone talking about AI prototyping focuses on speed. Build prototypes in hours instead of weeks, anyone can do it now, costs dropped to nearly zero, etc.
This is all true, but it misses the point.
Speed by itself doesn’t make your product better. Shipping the wrong solution faster doesn’t help your customers. The real value of AI prototyping isn’t that you move faster. It’s that speed gives you capacity to explore more ideas before you commit.
For years, product teams operated under a brutal constraint. You could only afford to prototype one or two solution directions because design resources were limited and building even rough prototypes took weeks. So teams did their best to pick the most promising direction, commit resources and hope it worked.
AI prototyping removes that constraint entirely. You can now generate five different approaches to the same problem in an afternoon. This changes everything about how teams find the best solution to a known problem. You’re no longer choosing the first feasible approach. You’re choosing the best approach after deliberately comparing alternatives.
That’s the second-order effect of speed. You can put more ideas on the table, which allows for a thorough exploration of which is actually the best one for your customers. This article walks through how to capture this benefit. Not by moving faster, but by using speed to explore more before you build.
Introducing Reforge Build
This has been top of mind for us here at Reforge lately.
Early in 2025, we (like everyone else) were fascinated by the possibilities but frustrated that the market of vibe coding tools weren’t purpose-built for PMs working with existing products, stakeholders and organizational constraints. So we created Reforge Build, the AI prototyping tool for product builders.
You can learn more and start building here. Use the code PRODUCTIFY to get a month of premium for free.
The first-order effect: obvious speed
The immediate benefits of AI prototyping are clear. Traditional prototyping used to take weeks. A designer would create mockups in Figma, which required both tool knowledge and design skills. A developer would wire up the interactions. Back and forth, iteration after iteration. It took time and that time was expensive.
AI tools collapse this timeline from weeks to hours. You can go from a screenshot or rough description to a functional prototype in a single session. This removes the skill barriers entirely. PMs can prototype their own ideas. Founders can test concepts before hiring a team. Customer success reps can mock up features their clients are requesting.
The economics shifted dramatically too. Prototyping costs have fallen to near zero. What used to require expensive design and development resources now costs the price of an AI subscription. This means ideas that would never justify the resource investment can now get tested and these benefits are real and valuable. Teams that adopt AI prototyping ship faster because they reduce dependencies on constrained design resources. They can validate concepts far earlier in the development cycle.
But this is table stakes. The first wave of AI has been all about value capture (e.g. making existing processes 10% or 15% more efficient) and the second wave is all about value creation. Every team using AI prototyping can work faster because the speed advantage is so obvious. What separates teams that get exceptional value from AI prototyping is what they do with that speed.
The second-order effect: more ideas, better decisions
Speed creates capacity to explore divergent solutions and this is where the real value lives.
The old pattern was straightforward but limiting. You validated a problem through customer interviews and support tickets, then brainstormed some ideas about how to solve it. You picked the direction that seemed most promising and committed resources to build it. Maybe it worked well. Maybe it worked okay. Maybe it missed the mark entirely. Frustratingly, you wouldn’t know until after you’d invested weeks of design and engineering time.
AI prototyping opens the door to a completely different process. You can explain your feature idea or problem to AI, ask it for five different ways to address, then prototype each of those five solutions in a day. You can even ask for mutually exclusive solutions, then cherry pick the best parts of each.
Most importantly, you can actually compare these approaches to find the one that delivers the best experience and the most value to your customers. Once you get going, it’s easy to see the tradeoffs and benefits of each one clearly. And you can get them in front of stakeholders quickly to collect well-rounded feedback. When the time comes, users can test functional software rather than providing hypothetical feedback on a mockup.
This is fundamentally different from picking the first feasible solution. You’re choosing deliberately after exploration instead of committing early and hoping you guessed right.
The trap of moving too fast
There is one risk here. AI can take you to solution almost too quickly. You’ll generate something in 10 minutes that looks polished and feels complete. Your brain wants to say “done, let’s build it.” That instinct burns the entire advantage.
The teams getting exceptional value from AI prototyping have built the discipline to resist that urge. When they generate their first promising prototype, they force themselves to ask “okay, that’s one option, what are four more?” They use the speed to explore more, not to stop exploring faster.
This discipline is harder than it sounds because the prototypes look so good. You have to actively create a divergence step before you allow yourself to converge on a solution. You have to make exploring alternatives a requirement, not an option.
The payoff is worth it. You end up building the right solution instead of just a feasible solution. Your users get something that actually fits their workflow instead of something that kind of works. Your team doesn’t waste weeks building in the wrong direction only to discover the issues after launch.
How to force divergence
Most teams say they want to explore multiple directions but end up generating variations of the same idea. The prototype looks different but the underlying approach is identical. This happens because prompting AI without structure leads to convergence.
You need a deliberate process to force divergence. Start by using AI to brainstorm before you build anything. Ask it to suggest three to five fundamentally different approaches to your problem and be specific about what “different” means.
Here’s a template that works:
*Here’s an example of this prompt filled in with real product information. You can try it out free in Reforge Build here. Customize it or just run it as is to see how it works.*
That last part is critical. Don’t just ask for “different options.” Tell AI the axis of variation you want. For example, if you’re building a feature to help project managers track blockers, you might ask for approaches that differ in how proactive, visible and automated they are. Then specify what you mean by those terms. A passive approach surfaces information in existing tools. An active approach lets PMs query for blockers. A predictive approach anticipates blockers based on activity patterns. A collaborative approach has teams surface blockers together. And if you’re struggling to come up with different approaches, just ask AI
Once you have your directions mapped out, build lightweight versions of each. You’re not trying to create pixel-perfect designs here, just working prototypes that demonstrate each approach clearly enough that people can react to it. The goal is making each solution direction tangible.
You’ll know you’re doing this right when stakeholders can immediately articulate the tradeoffs between approaches. If they’re debating button placement instead of discussing which mental model fits user workflows better, you generated variations instead of divergent solutions.
The discipline here is resisting the urge to pick a winner too early. Keep all the options in play until you’ve shown them to stakeholders and tested them with users. The best solution often isn’t obvious until you see how different audiences react to each approach.
Where AI prototyping thrives
AI prototyping delivers the most value in three specific situations:
Stakeholder alignment: Show two or three directions to leadership before heavy engineering investment. Everyone sees the same thing instead of interpreting written requirements differently. This surfaces concerns and constraints early when they’re cheap to address. You get alignment on the concept before designers spend weeks in Figma.
Solution discovery: Use this when the problem is clear but the path forward isn’t. You’ve validated that users need better pipeline visibility but you’re not sure if they want a dashboard, notifications or a query interface. Prototype each approach and pressure test them to understand the tradeoffs. The best solution becomes obvious once you can compare real options.
Specification for design and engineering: The prototype becomes your reference implementation. Engineers look at working software instead of interpreting text documents. This removes the “is this what you meant” back and forth that wastes time. Your PRD still documents the business context and success metrics but the prototype shows exactly what users will see and do.
And here’s what not to do
A few common mistakes will kill the value here.
Don’t prototype before validating the problem exists. Multiple solutions to the wrong problem just means you explored the wrong space efficiently. Do your discovery work first.
Don’t confuse polish for proof. The prototypes look production-ready but they’re exploration tools. Fighting with AI to get pixel-perfect layouts wastes time and misses the point.
Don’t skip the divergence step because the first prototype looks good. This is the most common mistake and it burns the entire advantage. Force yourself to explore alternatives even when that first option feels complete.
Speed → exploration → happier customers
Speed is valuable but it’s just the enabler. The real value comes from using that speed to explore more ideas before you commit resources. Better decisions come from comparing approaches, not from finding feasible solutions faster.
Start small with your next feature. Force yourself to explore three genuinely different approaches before you pick one. Use the template above to structure your thinking. Build lightweight prototypes of each direction. Show them to stakeholders and users. See what you learn when you actually compare options instead of committing early.
That’s how you turn AI prototyping speed into better products for your customers.