I'm Head of Product at DualEntry, an AI-native ERP for finance teams. I run product management, design, product marketing, and education. Khosla Ventures and Lightspeed co-led our $90M Series A. Before DualEntry, I spent 8.5 years in product at Cockroach Labs, from pre-revenue through $100M+ ARR.
I build product with agents doing the repetitive work. When we lose a deal, an agent scores the loss from the call recordings, I talk to the buyer, and agents turn that conversation into Linear projects the team can act on. I write here about how I build products with AI, from prototypes to production.
.jpg)
DualEntry's docs, MCP server, and in-product AI Accounting intelligence all run on the same content, so finance teams get accurate answers in minutes instead of support tickets.
Three-phase AI workflow that separates strategic design from tactical execution. First-time success rates jumped from 45% to 82%, implementation time dropped 62%, and post-deployment bugs fell 80%.
Strategic AI model selection: using Claude Opus for architecture, Claude Code for implementation, and Claude Haiku for validation. Multi-model orchestration with 9,574x ROI.
So you were worried about a failure, and it happened. Despite all of the planning, something went wrong, and now the goal is to minimize the impact of that failure.
CockroachDB is built to prevent systemic breakdowns by isolating failures, automating recovery, and maintaining consistency across environments.
In this 3-part series, "Surviving Failures with CockroachDB," we'll explore the most common failure scenarios affecting modern applications and databases.
Reduced prices in compute for CockroachDB Cloud Standard and Advanced plans. Only pay for what you use in data transfer, backups, and changefeeds.