Can ChatGPT Do Growth Strategy?

By David Isaac, Founder, AIPath. Last updated 20 July 2026.

TL;DR. Partly, and the parts it misses are worth doing, also.

General models in a chatbot interface are excellent at framing options, summarising research and stress-testing a narrative on demand.

For ongoing, mission-critical growth strategy you need a visualization of how the departments in your company interact as decisions are made. Understanding the impact on each team, and the new priorities that follow, is critical, and it is something chatbots were never designed to do.

Committing resources for the next quarter or the next six months is a mission-critical decision, and the context of what happened inside your company, and in your market, has to be rock solid before you can define the best next options for growth. There are always too many good ideas. Add the opportunity cost of wasted engineering to the lost revenue from the wrong strategy choice, and enterprise-grade pre-testing of strategy stops being a luxury and becomes a competitive advantage worth millions. AIPath puts that pre-testing in reach of growth-stage companies and their CEOs and C-suites, the USD 5 Mn to 50 Mn ARR bracket, not only the companies that can afford digital twin simulations from Palantir.

A wrong strategy does not announce itself for two quarters. By then the engineering is spent and the window has moved. AIPath is built to find that out in days.

You don't need more ideas; you need clarity on which ideas lead to growth. See the difference on your own company with a free demo session.

What does ChatGPT do well for growth strategy?

It does more than most strategy professionals concede.

A frontier model is a strong first-pass strategist: it frames a market entry question in seconds, surfaces considerations a team would take a week to list, summarises research, drafts positioning alternatives, and can examine the pros and cons of a particular option. For a leader who thinks by writing, it is the best sparring partner ever shipped, available at 11pm, which is when growth decisions actually get wrestled with after the daily operating work is finally paused. Used this way, a general model raises the floor of strategic thinking, and leaders should keep using it this way.

The trouble begins when a confident, fluent answer gets treated as a decision. It has not been compared statistically against the other options, and it has not been modelled against the trade-offs between the product, marketing, sales and engineering roadmaps. Even the sequence of initiatives on a roadmap is a set of variations that has to be computed. Cause-and-effect data from your market never reaches a chat model, so nothing it learns this quarter sharpens what it tells you next quarter.

Closed-loop compounding intelligence, built from real-world market tests, is the AI equivalent of gaining leadership experience. AIPath is built to compound exactly that way.

Can a chatbot give you tested options on roadmap trade-offs?

When you ask a chatbot for an answer on a mission-critical budget decision, what comes back is a plausible argument. A tested hypothesis, proven in your own market against a control, is a different object entirely.

The gap between those two objects is statistical validation, and that gap is where runway goes to die.

Strip it back and there are three levels of decision confidence, and AIPath is built to carry you to the third:

1. Valid options

2. Simulated winners

3. Statistically validated in-market winners
   

Ask your chatbot for the statistical power and the confidence interval between your leading growth strategy options, then judge what comes back. AIPath was built because that answer never arrives.

Where does a general model stop?

The model is the engine, not the car. A frontier model reasons like a brilliant generalist with no stake in the outcome: it produces plausible strategy the way it produces plausible prose, fluent and untested. Five gaps separate that from a decision a board can fund. No objective function: a raw model optimises for sounding right, not for whether this company should chase acquisition, retention or expansion this quarter, and the right answer changes with your stage. No ground truth: it holds no live model of your company, so each session starts from whatever you paste in, and its picture blurs rather than updates. No testing: it cannot run anything in your market, so there is no control arm and no verdict, only argument. No compounding memory: it resets, so nothing it learns about your company sharpens the next decision. And its confidence is opaque: a fluent answer with hidden reasoning is a slot machine with good grammar, which is a hard thing to defend to a board.

What is the difference between an answer and a decision?

An answer is text. A decision is a commitment a leadership team can fund and defend, and it has a bar to clear: evidence generated about the exact decision being made, at the resolution of the actual decision, tested in the market against a baseline before the money moved, and framed so a board can audit it. Strategy advice from a general model clears none of those bars, however sharp the prose. That is not a flaw in the model. It is a category difference between generating language and generating tested evidence.

How do ChatGPT and AIPath compare?

1. Objective. ChatGPT optimises for a plausible, well-formed answer. AIPath holds an explicit growth objective that reweights with your stage: acquisition, retention, expansion, profitability.
   

2. Company state. ChatGPT knows what you paste into the session. AIPath holds a living model of the company, updated as evidence arrives.
   

3. Option exploration. ChatGPT generates options on request, bounded by the prompt. AIPath systematically maps hundreds of thousands of configurations and ranks them by probability of paying off.
   

4. Testing. ChatGPT cannot act in the market; the answer is the end point. AIPath runs the strongest candidates as live in-market tests against a control before budget commits.
   

5. Memory. ChatGPT resets between sessions. AIPath compounds per company: every test result sharpens the next decision.
   

6. Execution link. ChatGPT advice ends at the chat window. AIPath hands a sequenced roadmap to product, engineering, sales and marketing, with rationale attached.

Does a better model close the gap?

No, and this is the part that surprises people. Every gap above is structural, not a capability the next model release adds. An objective function, a live company model, market-facing test infrastructure and per-company compounding memory are platform properties, not model properties. AIPath uses frontier models inside the system, so a better engine makes AIPath better. It does not make the car. The gap between a general assistant and a decision platform widens with model quality rather than closing, because better prose makes untested answers more persuasive, not more correct.

ChatGPT and AIPath: which one, when, and when you need both

Use both, for different jobs. Choose ChatGPT if the job is thinking: framing a question, drafting a narrative, summarising research, stress-testing your own reasoning, cheap breadth on demand. If the question commits no budget this quarter, the chatbot is the right tool, and the faster one. The moment a question commits real resources (a segment entry, a product investment, a repositioning, next quarter's budget), it has crossed into decision territory, and decision territory is what AIPath was built for. AIPath autonomously generates, ranks and pre-tests the options; the CEO decides. AIPath then sequences the winning path against the growth lever your business needs now and hands every team the instruction. The chatbot sharpens your thinking. AIPath pre-tests it before it costs you runway.

Questions leaders ask about ChatGPT and strategy

1. Is AIPath built on top of ChatGPT? AIPath uses frontier models as reasoning engines inside a larger system, the way a car uses an engine. The platform around them is AIPath's own: the growth methodology (Integrated Growth Execution), the living company model, the objective function, the in-market testing infrastructure, and the per-company memory.

   
   
   
   

2. We already pay for ChatGPT Enterprise. Why add AIPath? Keep it. It makes your team faster at drafting and analysis. It still does not know which growth lever your business needs this quarter, cannot pre-test a strategic initiative in your market, and cannot hand four teams a sequenced roadmap with evidence attached. AIPath does those specific jobs, and they are the jobs that set the budget for everything else.

3. Will better models make AIPath unnecessary? The opposite. Better engines make AIPath's exploration and reasoning stronger, while the platform properties (objective, model of the company, testing, compounding) remain the differentiator. Fluency was never the bottleneck in strategy; tested evidence was.

4. Can ChatGPT run in-market tests? No. A chat model produces text; it does not deploy live experiments, hold a control arm, or measure real acquisition cost. AIPath's tests return market evidence, such as an insurance deployment where recommendations cut acquisition cost from USD 240 to USD 43 in one quarter.

5. What should we keep using ChatGPT for? Framing questions, first drafts, research summaries, red-teaming your own narrative, and preparing sharper inputs for real decisions. It is the best thinking accelerant available. Deciding is a different job. AIPath exists for that one.

See what a tested growth decision looks like at library.aipath.one/home, or book a free 45-minute working session: AIPath maps your growth opportunities live, on your data, and you keep the analysis either way.

About the author. David Isaac is the Founder of AIPath, the Growth Decision Intelligence platform. He was formerly ASEAN Innovation Co-Lead at EY-Parthenon and Chief Growth Officer at GrowthOps (ASX:TGO).