My AI & PMM
Workflow Setup

The actual tools I use for product marketing, competitive intelligence, positioning research, and writing. Updated periodically. No affiliate links — just what I actually use.

AI I Use for PMM Work

Claude (Anthropic)
Primary AI assistant

My main tool for positioning work, messaging iteration, and long-form writing assistance. I use it for drafting positioning documents, stress-testing value propositions, and restructuring arguments. Better than GPT-4 for nuanced B2B writing tasks — it understands the difference between what's technically accurate and what's actually persuasive.

anthropic.com
Perplexity
Research & competitive intelligence

I use Perplexity for real-time competitive research — tracking what competitors are saying, finding recent analyst commentary, and synthesising news fast. Better than Google for research tasks that need synthesis rather than a list of links. I run competitive landscape updates through Perplexity before every major positioning review.

perplexity.ai
ChatGPT (Deep Research)
Market sizing & analyst report synthesis

ChatGPT's Deep Research mode is genuinely useful for analyst report synthesis — pulling together Gartner, Forrester, IDC coverage on a topic into a structured briefing. I use it to prep for analyst interactions and to understand the state of a market before building positioning. Takes 5 minutes to do something that used to take half a day.

openai.com
Notebook LM (Google)
Document analysis & synthesis

I upload customer call transcripts, analyst reports, and competitive intelligence documents and use NotebookLM to find patterns and answer specific questions across all of them. Particularly useful for win/loss analysis — synthesising themes from 30+ call transcripts in minutes instead of days.

notebooklm.google.com

How I Use AI in Specific Tasks

Competitive Intelligence

Perplexity for real-time tracking. Claude for synthesising competitive signals into positioning implications. ChatGPT Deep Research for quarterly landscape reviews.

Positioning & Messaging

Claude for first drafts and iteration. I write the core argument, Claude pressure-tests it and rewrites alternatives. I decide. Never Claude decides.

Customer Research Synthesis

NotebookLM for synthesising call transcripts and survey data. I upload the sources, ask specific questions, and use the answers to identify patterns I'd have missed reading linearly.

Writing & Essays

I write the first draft. Claude edits for structure and clarity, not for voice. If I can't tell the difference between my draft and the AI edit, the AI edit is wrong.

Analyst Prep

ChatGPT Deep Research to build the analyst's current view before a briefing. Perplexity to find their most recent published positions. Claude to structure my counter-narrative.

Where AI Doesn't Help

Strategic judgment. Understanding what a customer actually fears versus what they say they fear. The conversation that changes a CIO's mind. AI makes me faster at the craft. It can't replace the insight.

How I Write Essays

I write everything in plain text first. No formatting, no structure — just the argument in the order I'm thinking about it. Then I read it, find the real opening (almost always the second paragraph), and restructure from there.

The essays on this site are genuinely mine — the ideas, the specific stories, the arguments. I use Claude to stress-test structure and catch where the logic has gaps. The voice is mine because I've found that AI-edited writing without strong authorial control reads like AI-edited writing, which is not a compliment.

The test I use: if I read a sentence and I'm not sure whether I wrote it or the AI wrote it, I rewrite it until I'm sure it's mine.