ai for strategy

The AI tools page covers daily workflows — writing specs, summarizing research, running SQL. This page is different. This is about using AI as a strategic thinking partner. The kind of work that used to take a strategy consulting team two weeks and a hundred slides.

I am not talking about asking ChatGPT “what should my product strategy be” and pasting whatever it says into your quarterly plan. That is not strategy. That is outsourcing your judgment to a language model that knows nothing about your market, your customers, or your competitive position.

What I am talking about is using AI to compress the research, challenge your assumptions, and stress-test your thinking — so that the strategy you develop is better informed and harder to break.

The strategic PM’s problem

Here is the reality for most PMs in India. You are expected to think strategically — about competitive positioning, market entry, pricing, expansion — but you do not have the support system for it. There is no McKinsey engagement. There is no strategy team. There is you, Google, and whatever your sales team tells you about competitors.

So you do competitive analysis once a quarter. You read whatever TechCrunch and YourStory publish. You hear secondhand from sales calls. And you build strategy on a foundation of stale information and partial data.

AI changes this equation. Not by giving you answers — it cannot — but by giving you the research speed to ask better questions.

Competitive analysis at speed

Traditional competitive analysis is a spreadsheet updated quarterly. By the time you finish it, a third of the data is wrong. AI lets you run a continuous competitive intelligence operation as a single PM.

The workflow that works:

Step 1: Build the competitor dossier. For each competitor, paste their homepage, pricing page, product changelog, recent blog posts, and any press coverage into an LLM. Ask it to extract: core value proposition, target customer segment, pricing model, recent feature launches, and stated strategic direction. This takes thirty minutes for five competitors. Manually, it takes two days.

Step 2: Generate the comparison matrix. Ask the LLM to build a feature-by-feature comparison. Then add the column it cannot fill: “implications for us.” The AI can tell you that Competitor X launched an API. It cannot tell you that Competitor X launched an API because their enterprise customers were threatening to churn — that context comes from your sales team, your network, and your market intuition.

Step 3: Run the “so what” analysis. This is where most PMs stop. They have a pretty spreadsheet and no strategic insight. Ask the LLM: “Based on this competitive landscape, what are three moves our competitors could make in the next six months that would hurt us most?” Then ask: “What would we do in response to each?” You are using the AI as a sparring partner, not an oracle.

Step 4: Repeat monthly. Update the dossier. Track what changed. The delta is more valuable than the snapshot. If a competitor suddenly starts hiring ML engineers, that tells you more about their roadmap than their blog does.

Scenario planning with AI

Scenario planning is one of the most powerful strategic tools a PM has — and one of the least used. The reason is simple: it is exhausting. Building three or four plausible futures, mapping the implications of each, identifying decision points — this takes a room full of smart people and an entire offsite.

AI compresses this. Not perfectly, but enough to make it practical for a PM who does not have the luxury of a two-day offsite.

How to run a one-hour scenario planning session with AI:

1. Define the uncertainties. Pick two uncertainties that matter most for your product. For an Indian B2B SaaS company, this might be: (a) will enterprise AI adoption in India accelerate or plateau, and (b) will our category consolidate around 2-3 players or remain fragmented.

2. Build the 2x2. Four scenarios from two uncertainties. Ask the LLM to describe each scenario in two paragraphs: what the world looks like, who wins, who loses.

3. Stress-test your roadmap. For each scenario, ask: “Does our current roadmap still make sense in this world? What would we need to change?” The AI will generate plausible implications. Your job is to separate the plausible from the probable.

4. Identify no-regret moves. These are investments that make sense in three or four of the four scenarios. They are your strategic foundation. Everything else is a bet.

The AI is not predicting the future. Neither are you. The point of scenario planning is to make your strategy resilient across multiple futures, not to guess which future is coming.

Market research synthesis

Here is where AI delivers the most strategic value with the least effort. You have access to more market data than any PM in history — industry reports, earnings calls, regulatory filings, news coverage, customer reviews, community discussions. The bottleneck is not access. It is synthesis.

A specific example from the Indian context. Say you are a PM at a healthtech startup and you need to understand the competitive landscape for remote patient monitoring. Without AI, you spend a week reading NITI Aayog reports, WHO documents, competitor case studies, and VC funding announcements. With AI, you feed all of that into a long-context model and ask specific questions:

• “What are the top three regulatory barriers for remote patient monitoring in India?”
• “Which competitors have raised funding in the last 12 months, and what segments are they targeting?”
• “What is the gap between what ABDM (Ayushman Bharat Digital Mission) promises and what is actually implemented?”

You get first-draft answers in minutes. Then you verify the critical claims and add the context the model does not have — what your customers are actually experiencing, what your sales team hears in the field, what the regulatory environment feels like on the ground versus what the policy documents say.

The verification step is non-negotiable. AI will confidently cite statistics that are outdated, misattributed, or fabricated. For market research that informs strategic decisions, every number must be traced to a source you can verify. Use the AI for structure and pattern recognition. Use your own judgment for the numbers that matter.

What AI cannot do in strategy

I want to be direct about the limits because the hype around “AI for strategy” is thick.

AI cannot tell you what to prioritize. It does not know that your CEO has a relationship with a key enterprise customer who needs Feature X by March. It does not know that your best engineer is about to quit if she has to build another dashboard. It does not know that the Indian regulatory environment is about to shift because a new RBI circular is being drafted. Prioritization requires context that no model has.

AI cannot replace customer conversations. You can ask an LLM to simulate a customer interview. The output will be plausible and useless. Real customers surprise you. They use your product in ways you never imagined. They care about things you considered trivial. No model can replicate the moment when a small business owner in Indore tells you she prints your reports because she does not trust digital records. That insight reshapes your product. An AI simulation never would have generated it.

AI cannot evaluate strategic fit. “Should we enter the SMB market?” is a question that requires understanding your team’s DNA, your unit economics, your sales motion, your support capacity, and a dozen other things that are internal and undocumented. The AI will give you a balanced pros-and-cons list. A balanced list is not a strategy. A strategy is a bet — and bets require conviction that comes from deep understanding, not from pattern matching.

The rule: Use AI to expand the inputs to your strategic thinking. Never use it to replace the thinking itself.

The prompt patterns that work for strategy

After running hundreds of strategy sessions using AI — both for my own products and in training thousands of PMs — these are the prompt patterns that consistently produce useful output.

The “steel man” prompt: “I believe [your strategic thesis]. Argue against this position as strongly as you can. What evidence would disprove it? What am I not seeing?”

This is the single most valuable strategic use of AI. It forces you to confront the weakness in your own thinking. Most PMs build strategy in an echo chamber — their team agrees, their manager agrees, nobody wants to be the person who says “this might not work.” The AI has no political incentive to agree with you.

The “second-order effects” prompt: “If we [strategic move], what are the second and third-order consequences? Think beyond the obvious first-order effects.”

Strategy fails most often at the second-order level. You launch a freemium tier and think “more users.” The second-order effect is your support team is overwhelmed by free users, your paid conversion rate drops, and your best customers feel neglected. AI is good at tracing these chains — not perfectly, but well enough to surface risks you had not considered.

The “historical analogy” prompt: “What historical parallels exist for [your situation]? What happened in those cases? What can we learn from the ones that failed?”

This works because the model has been trained on massive amounts of business history. Ask it about “Indian B2B SaaS companies that tried to move upmarket from SMB to enterprise” and you get specific examples, specific failure patterns, and specific lessons. Verify the examples — the model sometimes invents companies — but the patterns are usually directionally correct.

Test yourself

Where to go next

• Build the AI product strategy that follows from strategic insight: AI Product Strategy
• Ground strategy in real user evidence: User Research Methods
• Translate strategic insight into a product vision: Product Vision and Strategy
• Learn the daily AI workflow that feeds strategic work: AI Tools for PM Workflows