An operator who ships. Not decks, not concepts. Working products with live demos you can touch, three decades of enterprise tech turned toward agentic AI. Everything below is a real process, running now.
An AI agent named Otto that places real phone calls, navigates payer IVRs, and turns each conversation into a structured, reusable Counterparty Graph. It predicts which calls will succeed, rewrites its own playbook mid-call when a payer changes the rules, and recovers denied revenue through appeals it knows will win. A working system you can watch make a live call.
An enterprise AI agent studio where every agent is governed by default. You give an agent instructions and grant it tools over MCP; Warden classifies every tool by risk, runs reads on their own, and holds anything that moves money or changes state for a human to approve. A policy engine adds spend caps, rate limits, and time windows. Every step lands on a hash-chained, tamper-evident audit log, with redacted telemetry and live cost metering. A working platform you can build an agent in and watch it run.
A solo-built SaaS that helps homebuyers understand a property before they make an offer, reading disclosures, flagging risks, and turning a stack of confusing documents into a clear picture of what you're actually buying. Shipped, in production, real users.
Enterprise IoT for indoor agriculture. An integrated software-and-hardware platform that moved indoor grow operations off manual analog sensors and onto automated environmental control, a cloud dashboard for temperature, humidity, lighting, pH, and irrigation across multi-room and multi-site facilities, with predictive analytics on top. Built on a fault-tolerant, redundant distributed architecture (patent-pending) for near-continuous uptime.
Proprietary work: fault-tolerant distributed systems and a Cloud Universal Controller for indoor agriculture, plus the hardware and UI/UX built for the grow market. I founded it, led it as CEO, built the dev team in India, and delivered an operation to an Oakland grow room.
Near real-time radio listener-data platform. AdEntify brought empirical measurement to terrestrial radio advertising, a market long starved of accurate data. Instead of small-sample panels reporting weekly, a mobile app captured brief, unobtrusive background audio samples, so data could be aggregated from potentially millions of listeners and made available immediately.
An identification engine matched uploaded samples against station broadcasts to determine exactly what was heard, when, and where. A listener incentive program rewarded users with points and location-based coupons, while stations and advertisers got web dashboards, a MatchMaker service, and a media-allocation tool to justify pricing, tune programming, and track ad ROI. As a founding member, I coded the audio-collection algorithms and applications behind the capture-and-match pipeline.
I design and build agents that take real actions in the world, not chatbots. CompletionLoop's agent places live phone calls, navigates payer IVRs, extracts structured data from the conversation, and drives an outcome. Built and running today.
The hard part of agents is state and judgment. I built a typed Counterparty Graph (the agent's memory of who it deals with), Self-Rewriting Playbooks that adapt when the rules change mid-call, and an Appeal Engine that reasons toward recovery. Design patterns, not demos.
At Cisco I drove agentic AI infrastructure for building Webex AI agents, and shipped the LLM inference layer beneath it at a ~50B tokens/month run rate. I know what it takes to run agents in production for large enterprises, not just in a notebook.
Models extract; humans decide what matters. I design where an agent acts autonomously versus where a person stays in the loop, the trust and safety boundaries that make agentic products shippable. That judgment is the difference between a demo and a product.
I'm Francis Kurupacheril, a product executive with 29 years in enterprise technology, from telephony and VoIP to collaboration platforms, and into AI since 2018. Founder and CEO of 453 Labs (formerly Piot Networks). I built on-device speech technology at BabbleLabs, acquired by Cisco, where I've shipped AI platforms at global scale.
Currently open to founding product roles, Founding CPO or VP Product, in agentic AI. The processes above are the proof: I build and ship, I don't just spec.
Named inventor on multiple granted U.S. patents in real-time communications, video, and collaboration, cited in 100+ later filings by Google, Amazon, Microsoft, Cisco, Zoom, Samsung, and IBM over more than a decade.