AI is everywhere. Whether that's in your LinkedIn feed, your news, your software, probably even your fridge. I noticed that the case for my washing machine.

However, not all AIs are built the same and give you the same real benefit. A lot of AI and AI for PMs, especially nowadays, is just smoke and mirrors, something put together more for the sake of marketing than actual functionality and solving real product management problems.

Thus, I went on a mission to identify the tools that will actually help product managers and improve the work output by checking if they improved my work output over the last few weeks. To get started, let me tell you about the methodology I use by telling you about the Logitech test:

Okay, so what is this logic test that I clearly made up and used to judge AI tools for proactive games? Well, many, many years ago when I was still a rookie new hire in my first startup, I had a mouse break down. I went to the shop to replace it so I could continue working. Yes, I was in my 20s. That's crazy, isn't it?

Anyway, the thing is that I've chosen something that stuck with me for years old, a mouse that had dedicated macro keys which I could assign any function to. I decided to go with copy and paste. Doesn't sound over the person, doesn't it? However, it's been with me ever since.

The gains I got from this simple change managed to improve my productivity at work so much that the company could fire two more people and keep me part-time and still have the same output that they had before I bought the mouse. And I'm not even joking. This is a real story.

So when I think of whether something is useful, it will stand the test of time. This mouse is my benchmark, and it tells me whether any tool I will use will actually help me and do that for years to come. When I chose tools for this article, I went with the ones that I believe will eventually pass my mouse test.

But what happens when you start using a tool that doesn't pass this test?

The dangerous tools are not the obviously bad ones. They are the ones that look helpful while costing you in ways you do not notice until later. Five traps to watch for:

When any of these creep in, you get the tail wagging the dog. You start shaping your work around the tool instead of the tool serving your work.

So, following the Logitech test, before a tool earns a spot, it has to clear all five of these. Not three. Not four. All five.

For me, integration is actually the most important when it comes to looking at those tools from the product management perspective. Truth be told, many great AI tools will not survive the test of time and usefulness if you have to jump through the hoops to use them, and they don't fit the tech stack and process that you already have at your company. Thus, it's critical for the tool to fit your existing process and tech stack in order to be actually useful long-term.

Here is the stack that passed the Logitech test. I have ordered it the way you move through the work itself, starting with the tools that make everything else faster, then walking the product process from discovery to delivery to alignment.

Oh, and quick mention: Some of the tools have sponsored my posts in the past, though none of them asked to be on this list, and every place is generally earned.

These are not tied to one stage. They speed up every other tool on this list, which is exactly why they earn their place first.

Wispr Flow (dictation).

System-wide voice-to-text that works in any app you already live in: Jira, Notion, Slack, Gmail, your AI chat window, anywhere your cursor lands. It removes filler words, fixes punctuation, formats as it goes, and learns your vocabulary so it stops mangling your product names and acronyms. It runs on Mac, Windows, and iPhone, with latency low enough that the text just appears as you talk.

Moreover, it allowed me to speed up my text input by a factor of two. I'm twice as fast.

Check it out using my link: https://ref.wisprflow.ai/dr-bart-pm

My verdict: this is the rarest kind of tool, because it makes every other tool on this list faster. If you think faster than you type, and almost all of us do, dictation is not a nice-to-have. Try it for one week and you will not go back to the keyboard. (I switched my own setup to Wispr Flow and the cross-app consistency is what sold me.)

ChatGPT (precise image creation).

Not designed for it, but in practice it understands a prompt better than the dedicated image tools. Midjourney makes prettier pictures. PMs do not need pretty, we need the exact image we pictured.

My verdict: after staring at my beautiful but useless Midjourney renders, ChatGPT gave me what I actually asked for.

Wan (short video and GIFs). It does not make headlines the way Sora did, but for small, moderated animations it is what works best for me.

My verdict: you would be surprised how much better a post performs with a GIF of subtle, non-obvious movement and none of the usual AI slop tells.

Gamma (fast presentations). Builds a good-looking deck from scratch or from the bullet points you already have. No, it did not make the Infoshare deck. I built that by hand.

My verdict: AI is not replacing my PowerPoint skills any time soon, but when I do not have time to hand-craft something, Gamma covers me.

This is where you figure out what is actually worth building. Three tools, in the order the work tends to flow.

Perplexity (market and competitor research). AI search that returns source-backed answers instead of a wall of blue links. Ask a complex question, get a structured summary with citations. Research mode goes deeper for competitive and market landscape work.

My verdict: great for getting oriented fast. Always verify the citations. It is more accurate than ChatGPT for this and it still hallucinates on niche topics.

Canny (feedback triage). Connects to Intercom, Zendesk, Gong, and Slack, then scans conversations for feature requests, deduplicates them, groups them by theme, and surfaces what is actually being asked for most.

My verdict: it is genuinely something to watch hundreds of thousands of scattered feedback fragments consolidate into a handful of meaningful insights. If you spend hours a week categorising tickets, this is your tool.

Dovetail (research synthesis). Upload interview recordings, transcripts, or notes. It auto-transcribes, tags themes, and lets you query across all of it: "what were the top pain points last quarter?"

My verdict: I hate watching good research go to waste because it is scattered everywhere and effectively lost. Dovetail puts an end to that. Best for organisations with more research data than they can manage by hand.

Productboard (roadmap drafting and feature scoring). Pulls customer feedback from multiple sources and links it to roadmap items, scores features by customer impact, suggests prioritization against your strategic goals, and generates roadmap summaries for stakeholders.

My verdict: great for a good-looking, evaluated roadmap draft. It takes the edge off the blank page so you can focus on the fine detail. Best for PMs who want a professional roadmap without attaching their heart and soul to a tool.

Bolt.new (rapid prototyping). Turns a plain-language description into a working full-stack web app in minutes, no code required. Use it to pressure-test an idea before engineering is ever involved. You can put a clickable prototype in front of users the same day you have the concept.

My verdict: it feels like working with a slightly slow development intern, except this one is entirely yours. Best for validating desirability before you write a single requirement, especially on zero-to-one features.

Thumba (tickets and user stories). Generates complete user stories, acceptance criteria, and test scenarios from plain English or a wireframe image. Integrates straight into Jira and Azure DevOps, with a zero data storage policy that keeps it safe for enterprise and regulated industries.

My verdict: an awesome time saver for when you know exactly what you want to build but do not want to type all of it out by hand. Best for teams shipping a high volume of tickets who want consistent, well-structured stories without the manual overhead.

This stage never really ends, which is why these two run in the background of everything else.

Granola (meeting notes and action items). Granola, using your actual PC audio transcribes in real time, then produces structured summaries with action items, decisions, and key moments. Searchable across every past meeting.

My verdict: how did I ever sit in a meeting without an AI notepad? If I had to keep only one AI tool, it would be a genuinely hard call between this and Wispr Flow.

Notion AI (updates and documentation). AI embedded inside Notion that drafts PRDs, weekly updates, and retros from rough notes, summarises long documents, adjusts tone for different audiences, and fills gaps using context already sitting in your workspace.

My verdict: I never personally saw the point of Notion in my day-to-day, and I still have to admire how useful and well-integrated its AI is. Best for any PM already living inside Notion.

Just to let you know, the thing, tools like Claude didn't make it to this list and probably many others, because all of AI tools are simply not built for product managers, but for developers. That was my focus here was to present the options you have to actually improve your work, and I can vouch for my selection.

However, perhaps some tools I have not tested yet, and I should. Which one should I look at? Let me know in the comments or email me.