Automaton I: The Philosophy
A case for machines that can compute, reason, and propose without confusing intelligence for authority.
Emmanuel Sogelola
Security engineer, founder, and builder of intelligence that reasons, then waits.
Based in Brooklyn, New York. Working at Twitch and building Munk.
I build at the edge of trust and autonomy.
My work spans incident response, financial systems, and AI products. Across each field, trust must be earned, decisions must be legible, and authority must remain human.
Selected writing
Financial Advisory at Scale: In the Field
A continuation of the Cansbridge thesis, exploring what encoding empathy into financial advice looks like in practice through real advisory workflows.
Financial Advisory at Scale: Can Empathy Be Encoded?
A Cansbridge Scholar thesis on improving consumer financial literacy through personalized, human-centered financial advice and management.
Machines That Wait: Why Intelligence Should Stop Before It Acts
On provenance, permission, and building systems that can reason without confusing fluency for authority.
Experience
- Security Engineer, Twitch / Amazon
- Software Engineer, Wealthsimple
- Software Engineer, Zendesk
- Software Engineer, SurveyMonkey / Momentive
Selected projects
Munk Financial
An AI-native financial intelligence platform that turns personal financial data into clear, source-linked guidance, making expert-grade advice accessible to everyday people.
Echo-Echo (e-e)
A terminal copilot for markets: dual-feed monitoring across exchanges and prediction markets, health checks, and session logging. Phase 1 is observe-only — data flowing, no trading yet.
Solomon Platform
A multi-agent platform with local computer-use workers and a web control plane on one Go core. Every delegated task remains visible through its graph, output, and provenance trail.
Automaton
A provenance-first Python library for agents that reason, verify, and propose, but cannot act until a human decision gate grants scoped authority.