Product Manager · 0→1 · B2B · Platform
I build platforms that make
developers 10x faster.
8 yrs PM · 14 yrs building · GenAI + B2B + Platform
Selected Work
PM TOOL · 2025—PRESENT
PM tool for the AI-coding era. Connects customer feedback, decisions, and roadmaps to AI coding agents through a native 27-tool MCP server — so agents build against approved specs, not guesswork.
27
MCP tools
0→1
built from scratch
Free beta
live at ultipad.vercel.app
Read case study →
ENTERPRISE RAG · 2026
Self-hosted, explainable multimodal RAG platform. Every answer cites its source with visual highlighting — no black box, no SaaS lock-in, no per-seat pricing.
0.96
faithfulness score
~$0.001
per grounded answer
Self-hosted
full pipeline control
Read case study →
E-COMMERCE PLATFORM · 2024—2026
Operating system for direct response operators. 13 modules replacing 8 stitched-together tools — funnels, multi-CRM routing, fraud protection, and affiliate tracking in one console.
13
modules, 1 console
4
native CRM integrations
99.95%
checkout uptime SLA
Read case study →
How I Think
Start with the problem that has no owner, not the feature that has a champion.
When I scope a 0→1 platform, the first question is always: who is in pain right now, and what are they doing instead? The workaround is the product. If developers are copying API specs into Notion docs and pasting them into prompts, that's not a workflow — that's a gap in the platform I should build.
I define the non-negotiable in week one: the one thing the platform must do for a developer to have a better day. Everything else is deferred. I've learned that 0→1 dies from scope, not from lack of features.
The metric that proves the platform worked is almost never adoption — it's the behaviour change. Did the developer stop doing the manual step? Did the team stop asking the same question twice? That's the only signal worth optimising for at the start.