I bring AI agents into the messy reality of business — at the application layer, not the lab. Currently building NForce, an enterprise platform for deploying conversational AI across email, voice, WhatsApp, and webchat. Previously bootstrapped and sold SoundStage.fm to Audius.
I work at the application layer — taking frontier models and turning them into systems that businesses can actually deploy, govern, and trust.
Twenty years of full-stack engineering, product design, and founder reps means I see the whole picture — from prompt to pipeline to P&L.
Twenty years of full-stack consulting and shipping enterprise products. PoliSuite for mid-market insurance carriers, AMPSocial for marketing teams, CloudPal for SaaS subscription management. Learned every layer of the stack — and the business behind it — the hard way.
Built SoundStage.fm — a 3D metaverse for live music with WebRTC videoconferencing and spatial audio, where artists and fans gathered in custom-branded virtual venues. Bootstrapped from zero, acquired by Audius in 2022.
All-in on enterprise AI. Founded NForce — a multi-tenant platform for deploying conversational agents across email, voice, WhatsApp, and webchat. Not a researcher. The person who turns models into deployed, governed, revenue-bearing product.
A multi-tenant SaaS platform for deploying conversational AI agents across every channel a business actually uses — governed by structured playbooks, grounded in real knowledge, and integrated with the systems where work gets done.
Most AI products bolt a chatbot onto a SaaS. NForce is the inverse — a governance-first platform where every agent action is shaped by versioned playbooks, executed through audited tool calls, and synced back to the systems of record. Built for enterprises that need to deploy AI without losing control.
Open-source CLI agent for sales & RevOps. Free, terminal-native, MCP-compatible.
Below is the toolkit I've assembled to take frontier models from experimental to deployed.
Multi-agent systems with governance, not vibes. Stateless inference engines, in-memory guideline matching, dependency/priority/entailment relationship graphs, deterministic playbook resolution.
Ingest from anywhere — Drive, Notion, websites, commerce catalogs, uploads — parse, chunk, enrich, embed, and serve with hybrid retrieval (vector + lexical + reranking) that actually surfaces the right chunk.
Agents that do work, not just talk. Tool-calling bridges, callback-based execution, schema-driven definitions, 20+ production toolkits — CRM operations, calendar, storage, escalation, web browsing, custom integrations.
Agents that escalate cleanly when they shouldn't act alone. Live takeover, supervised hand-off, intervention rules, and a real-time event bus that surfaces anything outside the playbook to operators.
Smart-tag system that detects situations, sentiment, and outcomes in conversations — automatically record KPIs, fire alerts on patterns, and surface anything that needs attention. No one has to read every transcript.
Securing the agents themselves, not just the infrastructure. LLM guards against prompt injection and unsafe outputs, end-user authentication, and access controls that let you scope agents private, public, or to specific user groups.
Agent performance metrics and financial-impact dashboards. Conversation KPIs, resolution rates, time saved, revenue attribution — the layer that proves AI is earning its keep, not just running.
Production AI without the panic. Version-controlled playbooks (preview/released), automated test suites with AI-generated scenarios, smart-tag analytics, intervention/takeover, content safety filtering.
Whether you want to deploy NForce, need a fractional partner on a hard build, or just want to compare notes — let's talk.
