We're the team behind
Symbench.
Symbench is building modular, operator-first systems that help teams shape and trust AI-native operations.
Operator-first
Built for the people running the network, not just the systems around it.
Modular
Start with one workflow, then extend into a broader orchestration layer.
Adaptive
Digital twins should learn with the operation instead of freezing into stale models.
Human + AI
AI should amplify operational judgment, not bury teams behind opaque logic.
Build the orchestration layer where planners and AI work together.
Make real-time, AI-native operations usable without giving up oversight.
Top-down digital twins kept failing. So we built bottom-up.
After 35+ years running and advising supply chains at companies like Kimberly-Clark, Newell Brands, and Timken, Paul kept watching the same pattern repeat: large, top-down digital twin initiatives that looked impressive in a pitch and fell apart the moment someone needed to model a real scenario. Modeling the future meant weeks of manual data manipulation, in spreadsheets nobody could fully validate, with no way to track what changed between versions. Worse, the "tribal knowledge" held by the people who actually ran the network — the planners and operators closest to the work — never made it into the model at all.
Paul had been carrying the idea for two years before recent advances in AI made it buildable: instead of forcing one unwieldy model to represent an entire network end to end, connect many smaller models that are each fully validated. He found co-founder Surya through Y Combinator's founder matching platform, and together with Rajib they've spent the last six months turning that idea into Symbench.