Monday morning, and the browser has eleven tabs of AI tools in it. One writes PRDs, one summarises calls, one builds prototypes, one promises to tell you what to build next. Each has a free trial, and each trial wants your customer data. By Friday you have drafted more than ever, and the roadmap review still turns into a negotiation.
Most lists of AI tools for product managers stop at the tools. This one sorts them by the job they take off you, says what each costs, and spends some time on the one job that should stay yours.
Why this guide reads the way it does
Cyril Le Roux teaches Product Focus's flagship product management course and co-authored its AI course, and has been a senior product leader at BT, Photobox, Zoopla and TransferGo. The question he asks of any task, and the thread through this guide: is this a job for me, or something an AI could take care of?
His answer for most product managers: far more of the job than they hand over today, and never the choice of direction.
Three questions to ask of any AI tool
- Does it take a whole job off you? Sales enablement material, customer service training, first drafts of specs and release notes: none of these deserves much of a product manager's week any more. A tool that saves ten minutes inside a job you still do by hand is a nice-to-have.
- Can you check what it tells you? Never trust an AI completely. Good domain knowledge lets you spot what does not add up, and a tool that shows its sources lets you check the rest. Data analysis tends to be reliable; claims about markets and competitors need the most checking.
- Could you defend the result? You are the one who explains the roadmap to leadership. If you cannot say why an AI-drafted answer is right, you have not saved time; you have borrowed it.
Hand AI the work that feels like progress. Take back the time for the work that is.
Tactical work drowns strategy because it is easy and it feels like progress. AI is the first chance in years to claim that time back. The point of the tools below is what you do with the hours: understanding your market, your customers and your competitors well enough to choose a direction. Use AI there too, to move faster.
Job 1: a general assistant for drafting, research and thinking out loud
ChatGPT (OpenAI)
Free; Go $8 and Plus $20 a month; Pro from $100 a month
Best for: Drafting, summarising and analysing data, with the widest choice of plans and integrations
Watch for: Starts every conversation from what you paste in, unless you set up projects and memory with care
Claude (Anthropic)
Free; Pro $17 a month billed yearly or $20 monthly; Team from $20 per seat a month billed yearly
Best for: Long documents, careful writing and reasoning through a messy brief
Watch for: Same limit as any chat: it knows your product only as well as the context you give it
Gemini (Google)
Free; Google AI Pro £18.99 a month (the page showed UK prices only); for business, Gemini with Google Workspace from $14 per seat a month
Best for: Teams that already work in Google Docs, Sheets and Gmail
Watch for: Strongest inside Google Workspace; less of an advantage if your documents live elsewhere
Perplexity
Free; Pro $17 a month billed yearly
Best for: Fast market and competitor research with the sources listed beside each answer
Watch for: Cited sources make checking easier, not optional: open the ones your argument rests on
Job 2: customer research and feedback
Dovetail
No free plan listed; Bronze $4,800 a year plus credits, marked coming soon; Silver and Gold through sales
Best for: Turning interview recordings and transcripts into tagged, searchable evidence
Watch for: The synthesis is only as good as the interviews: it cannot fix leading questions
Productboard Spark
On request; Business includes 500 Spark credits per maker a month, Enterprise more
Best for: Teams already running feedback and roadmaps in Productboard who want an agent working over that data
Watch for: Its value depends on Productboard being where your feedback already lives
Enterpret
No published price; demo on request
Best for: Larger teams pulling support, sales and survey feedback into one place and ranking it
Watch for: Built for volume; a team with a few dozen requests a month may not need it
Job 3: meetings and notes
Granola
Basic free; Business $14 and Enterprise $35 per user a month
Best for: Notes from back-to-back calls without a bot joining the meeting
Watch for: Ask the other side before you record, and check the summary against what was agreed
Job 4: product analytics
Amplitude
Free up to 2 million events a month; Growth and Enterprise on request
Best for: Asking questions of usage data in plain language, with AI agents on every plan
Watch for: The answers are about what users did, not why; pair them with research
Job 5: mock-ups and prototypes
Figma Make
Free Starter plan with daily AI credits; Professional full seat from $16 a month, with 3,000 AI credits
Best for: Prototypes that stay next to your designers and their design system
Watch for: Agree with your designers where a PM prototype stops and design work starts
Lovable
Free plan with daily build credits; paid plans priced by credits, not per person (prices could not be read on the date of this guide)
Best for: A working web prototype from a conversation, to put in front of customers
Watch for: A clickable prototype can make an untested idea look finished
Job 6: discovery, documents and the roadmap
Jira Product Discovery
Free for up to 3 creators; $10.50 or $26.25 per creator a month
Best for: Ideas, insights and roadmaps next to the delivery work in Jira, with AI to organise and connect feedback
Watch for: Its home is the backlog and the roadmap, not the choice of direction
Notion AI
Included on Business, £16.50 per member a month (the page showed UK prices only); a trial on Free and Plus
Best for: Searching and drafting across the documents your team already keeps in Notion
Watch for: It answers from what is in the workspace, so out-of-date pages give out-of-date answers
Two techniques make these tools far more useful, and neither needs software. Before recording a customer interview, learn the approach in Rob Fitzpatrick's book The Mom Test: ask about what people did in the past, not what they would like in future, and no AI summary will be better than the questions behind it. And for competitors, have an assistant draft a profile of each rival that includes the strategy it appears to be following and the play you could make against it. A feature comparison dates within months. A rival's strategy tells you where its features are going.
Job 7: the strategy the other tools serve
Every tool above makes you faster at something. None of them decides where the product is going, and without that, faster means more features pulling in more directions: death by a thousand cuts, delivered sooner. There are three ways to do this job.
A general assistant and a disciplined process. Cheap, fast and enough for an experienced product manager who knows which techniques to use and in what order. The risk is a plausible strategy built in one conversation, with nothing to show what it rests on six weeks later.
A roadmap or discovery tool with AI. Jira Product Discovery, Productboard and Aha! all give a strategy somewhere to live next to the roadmap. They record the answers well. Arriving at them is still your work.
Scaffold. Scaffold is an AI product strategy tool for product managers and product engineers. Paste a link and drop in your documents: a free research run drafts your starting point, Scaffold suggests the approaches that suit your product's stage, and Strategic Options compares every direction you kept with changing nothing. You choose and validate. Here is what sits on every product's map:
- 01Context: where you areCompany Foundations, Persona, PEST Analysis, Competitor Profiles, Porter's Five Forces, Value Proposition Canvas
- 02Route: how you get thereStrategic Options, Strategy Narrative ("The Trip"), Now, Next, Later
- 03Destination: where you're goingProduct Vision, 3 Horizons
Scaffold owns the judgement, never the backlog. It does not track delivery: no sprints, no statuses, no code generation. That stays in Linear or Jira, and the research, meeting and prototype tools above stay useful alongside it.
For the AI tools that help with the strategy work itself, compared job by job, see the best AI tools to help a product manager build a product strategy.
Where the others beat Scaffold
- For everyday drafting, summarising and quick questions, a general assistant is faster and cheaper. Scaffold is not a chat window.
- It does not record or transcribe meetings, analyse usage data or build prototypes. Granola, Amplitude, Figma Make and Lovable do.
- It has no feedback inbox or roadmap for stakeholders to browse. Dovetail, Productboard and Jira Product Discovery serve you better there.
And it is the wrong tool for one kind of product manager: the one in a very early startup whose CEO runs the strategy hands-on. There, the job is to execute that strategy faithfully, not to write a separate one. Scaffold could still be useful to that CEO.
A note on lists like this one
Including ours. Feature lists date within months, and every vendor here ships AI features constantly. A tool's features today tell you less than where its maker is heading. Read this guide as a snapshot from 19 November 2026, and check each vendor's own page before you buy.
Choosing, in five lines
- Start with one general assistant, and use it for a week on real work before adding anything.
- Add a specialist tool for your biggest recurring time sink: interviews, meetings or analytics.
- Prototype only what you will put in front of customers.
- Check the sources behind any claim your argument rests on.
- Spend the hours you win on the strategy, and keep the choice of direction yours.
This week, list the five tasks that took most of your time and ask of each: a job for me, or for an AI? Then decide what you will do with the hours.
Related guides: how to build a product strategy, and the alternatives to Productboard and Aha! Roadmaps.