Paweł Huryn

Thanks for finding me here! 😅

Okay, let me say this:
product Bluesky has failed.

I can believe there is still some engineering and design left.

Maybe AI.

Temporarily.

Meanwhile, X is booming 🤷‍♂️

-

I'll check again in a few months.

I expect the platform to fade away, as it fails to retain the users

MCP for PMs: How To Automate Figma → Jira (Epics, Stories) in 10 Minutes Case Study: Using an LLM Agent & MCP (Model Context Protocol) to transform Figma into 6 epics and 20 user stories in 10 minutes. Without touching the keyboard.

Want to learn more?

You can explore MCP yourself. It's not hard.

Or get ready-to-use configuration to copy from my new post: www.productcompass.pm/p/mcp-case-s...

Plus, my support on Slack anytime.

(7/7)

Its popularity has exploded (Google, MCP AI):

(6/7)

Before I share more, why is MCP so powerful?

It's banal to use, with 300+ supported systems:

(5/7)

What this means:

1. Spending too much time working with the backlog?
- You might easily save 10+ hours/week.

2. Is your job about creating User Stories, rather than discovery?
- Bad news. AI can do that for you. Learn fast.

(4/7)

High-level steps to do the same:

1. Download Claude Desktop
2. Get Figma access token
3. Get Atlassian API token
4. Configure MCP servers for Figma and Jira
5. Ask AI to create epics and stories based on Figma
6. Wait 10 minutes. Done.

(3/7)

YouTube video by The Product Compass Case study: How to Automate Figma → Jira (Epics, Stories) With MCP

First, the demo.
Kind of boring.

AI creates 6 epics and 30 user stories.
I'm watching: youtu.be/-J1T4dF8zqo

(2/7)

How to Figma → Jira epics and stories in 10 min. with AI and MCP:

(without touching the keyboard)

(1/7) 🧵

P.S. Enjoy it?

- Follow me @pawelhuryn.bsky.social to master AI PM together
- Share this thread with others: bsky.app/profile/pawe...

I appreciate it!

How to Build a RAG Chatbot Without Coding (AI PM Series) Learn about RAG, develop a better AI product intuition, and build an AI-powered chatbot for your PM portfolio. A step-by-step guide with ready-to-use templates.

(8/8) You can repeat it, to:
- Learn about RAG
- Develop a better intuition for managing AI products
- Build an AI-powered RAG chatbot for your portfolio
- Automate your work

Probably the only no-code, no-install tutorial.

How-to and templates: www.productcompass.pm/p/how-to-bui...

(7/8) In my new post, I present everything you need to create a functional chatbot using LLM and RAG with ready-to-use templates.

The process takes 15-45 minutes and doesn’t require coding or installing anything locally.

(6/8) 3. Tech stack

You can build it virtually for free:
- UI: Lovable(free version)
- Orchestration: n8n (free trial)
- LLM: GPT-4o-mini by OpenAI (less than $2)
- Embedding model: text-embedding-3-small
- Vector database: Pinecone (free tier)
- Documents: Google Drive

(5/8) 2. Runtime

When the user asks a question, the question is also converted into a vector and used to retrieve the most similar document chunks.

Finally, an LLM uses retrieved chunks and the original request to generate an answer.

(4/8) TL;DR: We can use those vectors to measure the similarity of text strings, e.g., when performing search, clustering similar data, detecting anomalies, or labeling data.

(3/8) 1. Preprocessing

When you use RAG, your data (e.g., documents) is not stored in the original format.

Instead, it's split into chunks (e.g., 500-1000 characters each), which are then converted into multi-dimensional vectors and stored in a vector database.

(2/8) Unlike an LLM, RAG can work with millions of documents.

But how does it work?

I built a RAG chatbot in 15 minutes. No coding.

A great way to learn, automate your work, or create a solution for a PM portfolio.

(That's probably the only tutorial to do it end-to-end without installing anything locally)

(1/8) 🧵

Have you ever considered your IQ is actually 110-120? Nothing wrong with that.

IQ is what IQ tests measure. Many use this circular definition on purpose as IQ doesn't account for all intelligence aspects.

I mean, you're publicly claiming to have an IQ higher than Einstein's (160).

(5/5) Hope that helps!

Enjoy this?

Follow me @pawelhuryn.bsky.social and repost this to help your friends: bsky.app/profile/pawe...

I appreciate it.

--

P.S. You can download my infographic (PDF) and 30+ others for free by subscribing to my newsletter: www.theproductcompass.tech/download-pm-...

(4/5) Finally, make sure your teams feel safe to experiment:

“A good failure is when the value of the lesson is greater than the cost of the lesson. A bad failure is when the value of the lesson is much less than the cost of the lesson.” - Alberto Savoia, author of The Right It

(3/5) Remember that you can’t eliminate risks.

The key to making decisions is taking calculated risks informed by data, qualitative insights, and intuition.

The ultimate test is users interacting with your product in the real world, using their own data.

(2/5) For example, a business model being copied.

Depending on the context, three other risk areas to consider:

- Go-to-market (I recommend separating it from viability for new products)
- Strategy and objectives
- Teams

Product Management risks

Product management is, at its heart, about managing risks.

Product teams usually focus on Value, Usability, Viability, and Feasibility.

But I’ve seen products and initiatives fail for reasons that didn't easily match those categories.

🧵

Deep Market Researcher AI Agent for Product Managers Things just got serious. This is probably the most specialized and fastest AI research agent for Product Managers.

𝐇𝐨𝐰 𝐜𝐚𝐧 𝐈 𝐮𝐬𝐞 𝐢𝐭?

Examples, best practices, and free access: www.productcompass.pm/p/deep-marke...

Hope that helps!

𝐖𝐡𝐚𝐭'𝐬 𝐧𝐞𝐱𝐭?

Deep Market Researcher will soon fuel other specialized agents that leverage best practices, templates, and custom prompts:

- Product Strategist
- PRD Generator
- Product Trio Ideation
- Assumptions Identifier
- Mock PM Interviewer

𝐂𝐚𝐧 𝐈 𝐮𝐬𝐞 𝐢𝐭 𝐟𝐨𝐫 𝐟𝐫𝐞𝐞?

Yes! I’m currently testing the platform to see what infrastructure costs it will generate, so there is a limit of 2 (everyone) and 5 (free subscribers) requests/per day.

There is no dedicated paid tier.

I'm attaching a market research report for Amazon AWS: drive.google.com/file/d/1bsBF...

- Researched: 60 websites
- Time: 30 sec
- Result: 22-page PDF report

No LLM gives such an answer in one go.

Just launched a free Deep Research AI Agent for PMs.

It can help you when working on product strategy, crafting a PRD, or preparing for a PM interview.

Enjoy this?

1. Follow me @pawelhuryn.bsky.social

2. Share this thread with your friends

I appreciate it!

bsky.app/profile/pawe...