Solutions & capabilities
Systems built and operated for real work
Each entry is something I have built, run, and can defend with evidence: live URLs, write-ups, or open source, with a real last-updated date. The complete inventory is date-ordered within each grouping.
Earlier products and experiments live separately in Earlier products and experiments, with what each one taught.
AI & Agent Systems
Systems that keep agents running, measured, and honest about cost.
Web Automation Microbench
Head-to-head benchmarks of 33 browser automation tools on one identical task, with a second real-work ranking that reorders them.
Agent Benchmark Matrix
A provenance-first coverage matrix of agent, computer-use and tool-use benchmarks, with harness separation and honest gaps.
LLM Pareto Frontier
A live quality-versus-cost map of major language models, with the efficient frontier calculated in the browser.
Agent Operations Control Plane
Launchd-managed workers, an ops board, alerts, and runbooks for a personal agent fleet.
Agent Routing and Lifecycle System
Role-aware routing from the cheap workhorse to frontier help, with A2A as a discovery capability.
Coding Agent Observatory
Token, cost, latency and session telemetry across every coding agent I run.
Model Routing Performance Lab
Benchmarked provider routing against measured latency, cost and quality.
Local LLM Lab
A measured field guide to every local model installed on Apple Silicon.
Model Intelligence Maintainer
A workbook and guide comparing model quality, price and provider coverage.
Agent Orchestra
A demonstrator for orchestrated multi-agent interaction patterns.
Martech & Measurement
Governance, QA, and reconciliation that make marketing data trustworthy.
AI Discovery Intelligence
A source-backed view of how consumer AI surfaces find, retrieve, cite and recommend brands — and which marketing actions the evidence actually supports.
Ad Platform Capability Explorer
Qualified planner answers for advertising-platform capabilities: supported / conditional / unknown, with conditions, evidence basis and a verification date, over a searchable master table of the published corpus.
Global Measurement Governance System
Taxonomy, QA, and reconciliation for measurement across markets and vendors.
Media QA and Attribution Reconciliation Toolkit
Tag QA, pixel checks, and attribution reconciliation that survive consent and real browsers.
Products & Operational Tools
Working products that support real operations, not demos.
AI-Assisted Product Definition System
Turns messy commercial context into PRDs, plans and scoped decisions.
Creative Observatory
A source-aware workbench for reviewing public ad-library evidence.
Hackathon Voting App
A production-ready single-screen judging app built for a live hackathon room.