AI for New Business Strategy: The Case for Structure Over Freeform Chat

Open a blank chat window, describe your new business idea, and ask whether it's any good. In seconds you get back something fluent, organized, and reasonable-sounding: a list of risks, a read on the market, a few next steps. It feels like strategy help. A lot of the time, it genuinely is.

The point here is not that AI is overrated. Freeform AI is very good at a real set of jobs: generating options you hadn't considered, summarizing a pile of research into something readable, drafting a first version of almost anything, and pressure-testing your thinking with a fast counter-argument. Used that way, it's one of the most useful tools to reach a strategy team in a long time.

The trouble starts when the chat window becomes the place where the strategy actually lives. Because the same freeform quality that makes AI so good at generating and summarizing is exactly what makes its raw output a poor container for strategy work. The answer is convincing, but it isn't tethered to anything you can build on.

Three things a chat answer struggles to do for your strategy

Strategy is not a single good answer. It's a set of decisions you have to compare, test, and revisit over time. Freeform output struggles with all three.

It isn't easily comparable at a portfolio level. Every chat is a fresh island. Ask about one idea on Monday and another on Thursday and you'll get two thoughtful write-ups that were never evaluated against the same criteria, in the same order, at the same depth. So when you try to decide which idea deserves the next dollar, you're comparing two essays rather than two options measured on a common scale, and across a whole slate of ideas that makes honest prioritization a real challenge.

It isn't testable. A fluent paragraph reads like a conclusion, but strategy needs the layer underneath it: the specific assumptions the conclusion rests on. "This looks promising" is an assertion. "This is promising if a specific customer has this specific problem, if we can deliver at this cost, and if the economics clear this bar" is a set of claims you can go check. Freeform text tends to hand you the confidence without isolating the assumptions, which is the one thing you most need in order to know what to validate next.

It doesn't accumulate reliably. While modern agentic tools genuinely can hold onto documents and artifacts, carry context forward, and build up a kind of memory across a project, the problem is that the memory is self-generating and non-deterministic. To keep from repeating itself, the agent has to keep capturing its own history and the rationale for every change, and the structure it improvises tends to drift, hard to keep consistent from one project to the next without a lot of scaffolding you build and maintain yourself. It's almost infinitely flexible, which is exactly what makes it so hard to keep uniform and durable. So a record can exist, but unless you impose the structure yourself, it isn't the dependable, comparable account of what's been validated that governance actually needs.

None of this is a knock on the technology. It's a mismatch between an open-ended medium and a task that needs continuity, and a discipline that benefits from a structure that can make the abstract more concrete.

What a structure adds

The fix is to give the work a structure the AI operates within, rather than letting an open chat be the structure. In our practice that structure is a strategy hypothesis model: you state the idea as an explicit set of choices and assumptions rather than a narrative, and you evaluate it across the dimensions that actually determine whether it will work, the desirability, feasibility, and viability of the idea, at a depth appropriate to how early you are.

That single shift restores everything freeform output was missing. Because every idea, however diverse in scope or focus, is framed consistently, the outputs are comparable and you can rank them honestly. Because each one is written as assumptions rather than a verdict, it's testable, and you know exactly which risky belief to go challenge first. And because the model persists, it accumulates: as evidence comes in, the record grows into a living account of what's been validated, what's been retired, and what still has to be proven. The structure is what turns a good answer into a decision you can defend and return to.

For the same reason, a structure helps not only people but generative AI. A structural framework is a map: it guides open-ended exploration and questions toward a consistent end you can evaluate, test, and validate, and an AI benefits from that guidance exactly as a person does. The explicit assumptions, the common dimensions, and the growing record that make a strategy legible to a human are the same things that make it legible to the AI tools working alongside them. What's good for human thinking and analysis turns out to be good for AI, which is why the most useful setup isn't AI or a structured framework, but AI working inside one.

The right pairing is AI inside the structure

In the noise of the "AI in innovation" hype, that distinction is easy to lose, so it's worth being concrete about what working inside the structure actually looks like.

AI tools can be genuinely helpful in new product or business strategy development, and Growth Forge® Software has AI built in too. What a standalone chatbot app lacks is the sharable, consistent framework and the persistent model that make its output comparable and cumulative. The practical question is how to get the speed of AI without giving up the structure strategy requires, and you get it by pointing the AI at the model rather than at a blank prompt.

That's the design principle behind the AI in Growth Forge Software: technology should serve the innovation team, not lead it. The AI Discovery Assistants help you gather and scan for relevant information, the AI Summary Assistants compress research into something usable, and the AI Industry Analyst Agent brings in external market and competitive analysis, and each one feeds the structured hypothesis model rather than producing a loose answer in a side conversation. Every one of those features is user-initiated, transparent, and gated on your explicit approval before anything it generates becomes part of the project. The AI accelerates the work; the model keeps the work coherent.

You could, in principle, assemble that structure yourself on a general agentic AI platform: build up a library of reusable prompts, skills, and document-based frameworks that force some consistency onto the AI. It's possible, and for a while it can even work. But doing it well is a serious undertaking. It takes real expertise, and the result has to be flexible enough for the messy diversity of actual business strategy while staying maintainable, sharable, private, and secure. That's a lot to ask of a team whose job is to evaluate ideas, not to build and maintain strategy infrastructure. We've already done that work in Growth Forge, grounded in decades of practitioner experience in new business strategy and innovation portfolio management.

That combination is also how we think a method should show up in strategy and innovation software: not as a chatbot bolted onto a blank page, but as an interactive, guided environment where the structure lives, the guidance is built in, and the AI works within it. If you want to see the specific tools, they're at Growth Forge; if you want the broader picture of where AI does and doesn't fit in the innovation process, that's the subject of our position on AI in innovation.

The short version for anyone weighing AI for strategy work: let AI do what it's genuinely great at, and give it a structure to do it inside of. Fluency without a framework produces answers you can't dependably compare, test, or build on. Fluency inside a persistent hypothesis model produces a strategy you can actually manage. Making that structure repeatable, with AI built in rather than bolted on, is what our consulting work and Growth Forge Software are built to support.

BRI Associates helps companies grow by drawing on decades of practitioner experience in corporate innovation and new business development — practitioners, not pundits or academics — through direct consulting, training workshops, and Growth Forge® Software, built for the unique requirements of corporate innovation and growth organizations.