Jack Lau
Jack Lau is an individual software engineer, computer-science and artificial-intelligence student, and solo developer/creator—not an organization. He builds software and open-source projects, including coding.sh, a local-first observability and evaluation tool for AI agents, and publishes technical writing on projects such as Ratchet.
At a Glance
- Individual developers and users of coding agents
- AI/ML engineers building agent applications
- Teams evaluating prompts, models, coding-agent configurations, and agent workflows
- Organizations needing self-hosted or local-first AI observability and evaluation
- +1 more
AI Tools by Jack Lau
(1)coding.sh
Local First AI Agent Observability
Discussions
No discussions yet
Be the first to start a discussion about Jack Lau
Latest News
coding.sh v0.1.7 released with CLI star prompts, expanded repository checks, doctor diagnostics, SDK weight-edit telemetry, and related session/eval support.
coding.sh v0.1.6 released with privacy controls, a trace-format JSON Schema, forwarding/sinks, serverless support, and expanded eval/tool-eval capabilities.
coding.sh v0.1.5 released with session safety flags, refined evaluation workflows, and additional CLI/session analysis capabilities.
coding.sh v0.1.4 released as part of the project's rapid initial release sequence.
Products & Services
Free, open-source, local-first observability and evaluation platform for AI agents. It records prompts, model calls, tool calls, file changes, costs, sessions, subagents, and traces; provides a local viewer and CLI; and supports evals, experiments, benchmarks, gateways, SDKs, OpenTelemetry, and team-server deployments.
Open-source Rust SPI-flash programming and hardware automation tool for CH341A and CH347 USB programmers. It can detect, read, back up, write, verify, erase, and analyze supported chips, retries failed blocks, and includes a built-in MCP server with 31 tools for agent-driven hardware work.
Open-source, self-hosted AI web agent built with Go and React, with multi-provider LLM support, GitHub OAuth, sandboxed Docker tool execution, AES-256 encrypted API-key storage, and a React chat interface.
On-device voice-to-text application for macOS using Swift, TCA, WhisperKit, and Parakeet TDT v3; a global hotkey records speech and transcribes it locally for pasting into any application.
Market Position
Jack Lau's coding.sh is positioned as a free, open-source, local-first alternative in the AI-agent observability/evaluation category. Its differentiation is the coding-agent-first workflow (hooks, session timelines, tool/file diffs, failure flags, eval replay, benchmarks, and cost analysis) combined with local storage and optional SDK/OpenTelemetry production ingestion. Comparable products in the broader category include Langfuse, LangSmith, and AgentOps; coding.sh emphasizes local-first operation and coding-agent workflows rather than only hosted LLM tracing.
Founding Story
Jack Lau's public profile identifies him as the founder of SpeechStream, a real-time speech-processing platform started in February 2021 and acquired in March 2025. His later open-source work follows the same pattern of building practical developer tools: coding.sh was created to record and inspect AI-agent sessions, costs, tool calls, file changes, evaluations, and experiments locally.
Business Model
Revenue Model
For coding.sh, the public site says the software is free and open source and that model spend remains the user's existing spend. The project supports local use, optional team-server deployments, SDKs, gateways, and self-hosted infrastructure; no paid pricing model is stated on the reviewed public pages.
Target Markets
- Individual developers and users of coding agents
- AI/ML engineers building agent applications
- Teams evaluating prompts, models, coding-agent configurations, and agent workflows
- Organizations needing self-hosted or local-first AI observability and evaluation
- Hardware developers and operators automating SPI, BIOS, and embedded-device workflows
- Debugging failed or long-running coding-agent sessions
- Tracking token usage, latency, and model spend
- Testing prompts, models, skills, and AGENTS.md changes
- Turning failed agent runs into regression evaluations
- Benchmarking agents and comparing configurations
- Observing production AI applications through SDKs or OpenTelemetry