AI + MCP tools have made me feel like a different kind of PM.
I was ramping up on a new product and immediately hit a foundational question: what does an "active user" actually look like here?
Answering that meant wading through @Slack and @Notion , then turning it all into testable hypotheses. Fast.
With @Claude by Anthropic and @Venn AI , I got better research than I would have found on my own. It surfaced docs I hadn't seen yet, synthesized the most relevant data points, and drew on growth frameworks from Reforge and Elena Verna to help formulate the definition ideas.
The result: multiple candidate active user definitions, each with a validation approach. Something that would have taken 10+ hours took about 30 minutes.
I could have gotten there eventually. But 30 minutes instead of a day and a half changes what's possible in a fast-moving startup environment.
If you've explored MCPs, what workflows have you found most helpful?
Replies
This resonates so much. On the recruiting ops side, I've started using AI agents to pull context from multiple sources before I even touch a candidate pipeline. What used to be a full afternoon of stitching together notes, job specs, and market data now takes about 20 minutes. The biggest unlock for me has been letting the AI synthesize first and then layering my judgment on top, rather than doing all the grunt work myself.
Venn.ai
@ceciliatran exactly! With Venn and all its connectors, I can pull in all the relevant context for AI to synthesize, saving a ton of time. Then output the result layered with my judgement directly into Gmail drafts, slack messages, etc.
My favorite of all, I can do this in the phone while on the go.
Hello Aria
This resonates. MCP tools and AI agents have fundamentally changed the PM loop. The old rhythm: write spec → wait for eng → get build → review. The new rhythm: describe intent → see prototype in minutes → iterate in real-time.
The PM superpower now is not requirements writing. It is context-setting and taste. Knowing what to reject and why. The AI can build many things — only a human with deep user empathy can tell it what actually matters.
Building Hello Aria (AI productivity assistant for WhatsApp/iOS, ~3k users, launching on PH April 10th) has made us experience this shift first-hand. Our PM now spends time on user conversations and product taste decisions, not Jira tickets. The velocity increase is real.
The risk I watch for: over-building. When prototyping is fast and cheap, the temptation is to build too many things. You still need the discipline to ship one thing well.
Venn.ai
@sai_tharun_kakirala Totally! I use @Lovable on several side projects, and have been excited to see that they now support MCP and added @Venn.ai for my most recent hackathon to pull in the PRD from a Google Doc and the flow from @Figma . I plan to figure out how to connect Stitch with Google as well since I use that as a design too.