Layer · VK / TOOLS

The tool overview

Living document. Concrete suggestions for tools for product leaders and product organizations that want to work AI-first.

Half of this has probably been swapped out in six months. That’s the point.

How the list is organized

Ten categories. The first five match the minimum stack from Layer 5. The last five are product-management specific.

For each category: 3–5 tools I either use myself, have seen work well, or think are worth a look. Short descriptions. No affiliate links, no recommendations I don’t stand behind.

Three principles for choosing

If you’re facing a blank choice:

  1. Pick what your team is already close to. An okay tool the team uses 100% beats a perfect tool the team uses 30%.
  2. Pick open standards over closed ecosystems. MCP beats proprietary plugins. API-based integrations beat native-only.
  3. Pick what you can switch away from. If it takes you a day to migrate to an alternative, you’re in a good position. If it takes three months, you’re locked in.

What the list doesn’t cover

I’ve deliberately written the short version. There are hundreds of tools in every category. But more names rarely help, they just increase choice anxiety.

If you think an obvious tool is missing, write to me. I update the list continuously based on input.

Last updated 01.05.2026 Living document
Claude Code
Terminal-based agent from Anthropic. Strongest for longer, agentic tasks. My primary stack.
Code →
Cursor
IDE-based. Good for people who want a visual environment. The switch to Claude models has raised the bar.
Code →
Windsurf
Cursor competitor. Slightly more agentic. Worth trying.
Code →
GitHub Copilot
Still the most widespread in enterprise. Less aggressive than the others.
Code →
Replit Agent
Web-based. Good for quick prototypes without a local setup.
Code →
Lovable
Prompt-to-app. Good for quick MVPs. Limits on complexity.
No-code →
v0 (Vercel)
Components and UI flows. Stronger on the React ecosystem.
No-code →
Bolt.new
Stackblitz-based. Runs directly in the browser.
No-code →
Replit
Broader than the three above. Backend too.
No-code →
MCP (Model Context Protocol)
Anthropic’s open standard. This is where the stack should converge over the next few years.
Context →
Notion MCP
Direct access to your workspace. I use it intensively myself.
Context →
LangChain / LlamaIndex
More dev-heavy frameworks. Relevant if you are building custom RAG.
Context →
Custom MCP servers
Build your own for internal systems. It is easier than it sounds.
Context →
LangSmith
Tracing, evals, debugging. The LangChain ecosystem.
Evals →
Braintrust
Eval-focused. Good if you want to test prompts systematically.
Evals →
Helicone
LLM observability. Logging, latency, cost tracking.
Evals →
PostHog
Broader product analytics, but has AI features. I use it myself.
Evals →
Cloudflare AI Gateway
Rate limiting, logging, a security layer between app and model.
Governance →
Lakera
Prompt injection and data leak detection.
Governance →
Credo AI
Governance platform for AI policies across the organization.
Governance →
Snyk
Secrets and vulnerability scanning. Has expanded into AI-generated code.
Governance →
Dovetail
Classic research platform with AI features for transcription and themes.
Discovery →
Maze
Usability testing with AI summarization.
Discovery →
Notably
AI-first research. Younger and more agentic than Dovetail.
Discovery →
Looppanel
Interview analysis, very fast.
Discovery →
Granola
My favorite. Takes notes live, formats them after the meeting. Privacy focus.
Meetings →
Fathom
Broader meeting recorder. Good for Zoom/Meet.
Meetings →
Tldv
Recorder + AI summary. Strong search across meetings.
Meetings →
Otter
The veteran. Still good at transcription, less strong on AI summaries.
Meetings →
Linear
Issues, specs, roadmap. Builds AI directly into the flow, the new standard for modern product teams.
Communication →
Slack
De facto team chat. The MCP server makes it interesting for agent flows, a lot of context lives here.
Communication →
Loom
Async video. Saves meetings. AI summaries make it useful without people watching the whole clip.
Communication →
Notion
The shared workbench: docs, databases, specs. Covered above as a context layer; here it is the collaboration angle.
Communication →
Cowork
Anthropic’s browser agent for non-developers. I use it myself for LinkedIn scheduling and DNS work.
Browser →
Browser Use
Open source. More dev-heavy setup.
Browser →
OpenAI Operator
ChatGPT-integrated. The most mature right now.
Browser →
Claude (web/app)
My primary tool for longer writing and structuring.
Writing →
ChatGPT
Good for quick drafts, especially with Custom GPTs.
Writing →
Loops
Email platform with a good API. I use it for all my products.
Writing →
Cursor Tab / Apple Intelligence
For micro-writing at the system level.
Writing →