Here's what a strategy consulting engagement actually consists of: someone interviews your leadership team. Someone runs competitive analysis using publicly available data. Someone builds a financial model in Excel using your numbers. Someone synthesizes findings into a slide deck using a template. Someone presents it back as strategic recommendations.
That's the $500,000 engagement. Every one of those tasks is now AI-executable. The only irreplaceable contribution: knowing which questions to ask. That's the Domain Translator role. The consulting industry sold you back your own judgment, dressed up in frameworks and PowerPoint. Now that the execution is free, your judgment is worth more than ever.
What Does Strategy Actually Consist Of?
Market Research — Industry sizing, regulatory landscape mapping, customer segment profiling, trend monitoring. This is information retrieval, synthesis, and pattern identification. AI at this task is not "good enough." It's categorically better than junior consultants — it processes more data, faster, with no billing incentive to extend the timeline.
Competitive Intelligence — Product/pricing analysis, positioning analysis, talent and hiring pattern monitoring, win/loss analysis. This is structured analysis of publicly available data. Perplexity plus Claude performs this function in hours, not weeks.
Scenario Modeling — Financial scenario construction, risk identification and quantification, strategic option mapping. Claude and GPT-4o with code execution produce scenario models faster and with fewer errors than an analyst team — and never charge for revisions.
Decision Documentation — Strategic recommendation synthesis, assumption documentation, implementation roadmap construction, board decks. This is structured writing. It is the thing consultants charge most for and that AI does most competently.
The Fog Machine has been hiding a simple fact: the execution of strategy work is not where the value lives. The value lives in knowing which questions to ask. You've always had that value. The consulting industry sold you back your own judgment, dressed up in frameworks and PowerPoint. Now that execution is free, your judgment is worth more than it has ever been.
What Is the AI Strategy Stack in 2026?
Research Layer: Perplexity Pro ($20/mo) for web-wide research with source citations — competitive intelligence, market sizing, regulatory analysis. Configure recurring research agents for weekly monitoring. Claude Opus for deep synthesis of uploaded documents — feed it your existing data, industry reports, competitor filings.
Analysis Layer: ChatGPT-4o with Code Interpreter for financial modeling and data visualization — you provide inputs and the analytical frame, it builds the model. NotebookLM (Google) for deep analysis of uploaded source documents — identifies contradictions and generates analytical frameworks.
Documentation Layer: Claude for strategic writing — brief it with findings and audience, it produces first drafts of board materials and executive summaries in 15 minutes. Gamma or Beautiful.ai for AI-driven presentation production.
Monitoring Layer: Perplexity agents running recurring research tasks automatically, delivering summaries weekly. Zapier connecting monitoring outputs to your reporting structure.
Total cost: $180–$220/month.
Decompose strategy into its components — market research, competitive intelligence, scenario modeling, decision documentation — and assign each to a specific AI. Perplexity for research, Claude for synthesis and documentation, GPT-4o for financial modeling. Total cost: $180–$220/month. The Domain Translator role — knowing which questions to ask — is the only human contribution required.
The expert who articulates the actual strategic question, provides context and constraints, directs AI tools to produce relevant analysis, evaluates AI outputs against domain knowledge, and makes judgment calls that translate analysis into action. This is what senior strategists have always done. What's new is that the junior analyst execution layer beneath them is gone.
No. AI cannot know which questions matter in your context, apply judgment from years of industry experience, detect anomalous data patterns, read a room, or integrate qualitative knowledge never written down. The execution layer is replaceable. Your domain judgment — the Domain Translator function — is not.
5 steps: Define specific strategic questions (not "what should our strategy be" — specific, answerable questions). Build research briefs. Run AI research passes. Build scenario models with your assumptions. Draft, review, decide. AI produces the documentation; you add qualitative judgment and make actual decisions.
