Needed a Product Cofounder, So I Built One
Why I built the second brain that AI builders actually need
The Problem I Couldn't Ignore
I was building TextGuardian, a cybersecurity app to protect against scams. The technical challenges were hard but solvable. The product decisions were what kept me up at night.
Should I build this feature first or that one? Would users actually pay for X? What if the market doesn't care about Y? I had many product conversations with GPT-4. Seemed insightful, but lurking behind every conversational turn was "You're Absolutely Right!!!" Happy talk. Endless optimism. No logic about tradeoffs or constraints or risk.
I needed something that understood my context. Something that would challenge my assumptions, not just validate them. Something that would tell me why an idea might fail before I spent weeks building it. I needed Candid AI.
That's when I realized: we have AI for coding, but not for product judgment.
So I built Cutline. Not as another chatbot, but as a product second brain—a system that validates ideas, stress-tests assumptions, and develops your product taste over time. Read the full story of why I built this and the insights that shaped Cutline's approach.
Background
Cornell University - Graduate studies in Cognitive Science and Machine Learning. Focused on how humans make decisions under uncertainty and how AI can augment (not replace) human judgment.
Consultant, Wells Fargo - Built anti-fraud machine learning systems. Learned that the hardest part isn't the ML—it's understanding what problem you're actually solving and whether the solution will work in production.
- eBay - Built ML systems to detect fraud, abuse, and policy violations at massive scale
- LivePerson - Developed AI to protect customer conversations and prevent bad actors
- Upwork - Created systems to ensure platform safety and trust for millions of freelancers
Why My Background Matters for Cutline
Building AI for Trust and Safety taught me something crucial: AI systems need to predict what will go wrong, not just what will go right.
When you're protecting millions of users, you can't wait for problems to emerge. You need to anticipate risks, surface hidden assumptions, and validate your approach before deploying. That's exactly what a pre-mortem does for product development.
Cognitive science gave me the frameworks for understanding how people actually make decisions (spoiler: not rationally). Machine learning gave me the tools to augment those decisions with data and pattern recognition.
Leading AI teams at scale taught me the difference between building impressive technology and building technology that actually solves real problems. The latter is much harder—and it's what Cutline is designed for.
Product Philosophy
I believe the future of product development isn't about building faster—it's about building better. AI coding agents have made the "building" part ridiculously fast. But nobody's solved the "what to build" part.
Cutline isn't trying to replace product managers or founders. It's trying to make them better. To compress years of hard-won product intuition into weeks. To surface the failure modes you can't see on your own.
The best founders don't just ship fast. They ship the right things fast.
That's what Cutline enables. Not another AI chatbot that tells you what you want to hear, but a product second brain that tells you what you need to know.
What We're Building
Cutline is in active development, with a focus on three core capabilities:
We're also building deep integrations with AI coding tools via Model Context Protocol (MCP), so validation happens in your flow—not in a separate dashboard you forget to check.
Want to Chat?
I'm always happy to talk about product validation, AI product taste, or building in the vibecoding era.