Tom Sullivan

I use applied AI and product thinking to streamline operations and unlock new value. I bridge strategy and engineering to deliver agents, analytics, and experiences that stick. My work blends requirements analysis, pragmatic builds, and clear documentation so teams can adopt and sustain the tools we ship.

I help teams with:

• AI agents & automation (Voiceflow, Twilio, Agent.AI, ChatGPT) to reduce manual work and speed up decisions
• Systems & process analysis: stakeholder interviews, functional/technical requirements, and flow mapping
• Data & analytics: instrumentation, attribution/tracking fixes, dashboards, and SQL-backed reporting
• Product delivery: scoping, iterative build/test, documentation, and handoff that drives adoption

Recent applied projects:

• Wildlife Triage Agent (Voiceflow + Twilio): Guides callers with triage logic, FAQs, patient status, and smart call routing; built via structured requirements and iterative testing.
• Poker Player Profiler (Agent.AI): Extracts Hendon Mob data and produces structured scouting reports; documented process and automation steps end-to-end.
• Poker Simulator (Python + PokerKit): Simulates play and hand outcomes with an emphasis analysis and player improvement.

How I work:

• Discovery: Clarify goals, constraints, stakeholders, and current workflows so we’re solving the right problem.
• Design hypothesis: Propose an approach, expected outcomes, and success criteria; outline the data and system touchpoints.
• Iteration & testing: Build a focused version, test with real cases, and refine based on feedback and results.
• Documentation & training: Create concise docs and handoff guides; provide brief training so the team can own the solution.
• Monitoring & analysis: Set up tracking and reporting, review outcomes on a regular cadence, and identify enhancements for the next cycle.

Current focus:

• LLM agents: natural-language conversation; user-centric experiences that improve satisfaction and outcomes; scaling service offerings (quantity and quality) efficiently and affordably; prompt and context engineering to optimize performance; and reliable structured outputs when needed.
• Structured agent evaluation, enhancement ideation, and refinement: define success measures, capture interaction data, review transcripts and failure cases, run small experiments, and ship improvements on a steady cadence.
• Practical integrations that boost efficiency, productivity, and impact: apply AI to intake/triage, knowledge access, workflow automation, and analytics/reporting.

If you need a builder who can translate goals into working agents, data flows, and interfaces—let’s talk.