Featured systems

Three systems sit at the center of the brand.

They define the public story: agent receipts, web evidence, and the operating system that keeps one person compounding with taste intact.

Public evidence systemWeb evidence audit

ShadowBuyer Web Data Audit Agent

A cash-flow oriented web-data agent that captures public pages, generates evidence cards, compares differences, and outputs a concise audit report.

Bright DataEvidence cardsAudit report

Problem: Public pages contain price, inventory, policy, and fee signals, but buyers still need reviewable evidence.
Protocol: A small audit agent can turn messy public web data into evidence cards and a one-page report.
Trace: Capture the page, extract fields, compare differences, preserve screenshots, and write claim boundaries.
Evidence: A cash-flow oriented web-data agent that captures public pages, generates evidence cards, compares differences, and outputs a concise audit report.
Next: Package a tighter demo and a clearer paid pilot entry.

Protocol buildAgent receipt protocol

A Mantle-facing AI DevTools project that turns agent claims, missing evidence, review gates, and fixes into structured work receipts.

SACPAgentOps DoctorMantle

Problem: Agents can confidently claim work is done while missing evidence, review, and responsibility boundaries.
Protocol: Agent output should leave a receipt: claim, evidence status, required fix, and review gate.
Trace: SACP receipt schema, unsupported-claim detection, human review gates, and hash anchoring.
Evidence: A Mantle-facing AI DevTools project that turns agent claims, missing evidence, review gates, and fixes into structured work receipts.
Next: Turn the protocol into a clearer demo and developer-facing README.

Public GitHubSolo-company operating system

A Markdown operating system for coordinating AI workers with founder decisions, durable worklogs, handoffs, dashboards, and reusable skills.

MarkdownAI workersHandoffsMemory

Problem: One person using many AI workers needs memory, decisions, handoffs, and worklogs that survive context drift.
Protocol: A Markdown operating system can make solo work more durable than a pile of chats.
Trace: Founder decisions, AI employee handoffs, dashboards, worklogs, reusable skills, and audit receipts.
Evidence: A Markdown operating system for coordinating AI workers with founder decisions, durable worklogs, handoffs, dashboards, and reusable skills.
Next: Publish a cleaner public map and starter guide.

Private systemLearning operating system

A knowledge system for turning reading into structured reflection, reusable insight, and long-term founder learning loops.

ReflectionLearning loopsKnowledge base

Problem: Reading often disappears into notes without becoming judgment or action.
Protocol: Reflection should become reusable memory, not just a feeling after reading.
Trace: Structured reflection, reusable insight cards, and long-term founder learning loops.
Evidence: A knowledge system for turning reading into structured reflection, reusable insight, and long-term founder learning loops.
Next: Convert private notes into a sanitized case preview.

Live productAI learning tool

A live IELTS study assistant for reading review, writing coaching, OCR-assisted material processing, and review-card based learning.

Web appOCRWriting coach

Problem: Language learning needs review loops, OCR-assisted material handling, and writing feedback.
Protocol: A small learning assistant can turn practice into reviewable cards and targeted coaching.
Trace: Reading review, writing coaching, OCR-assisted material processing, and review-card learning.
Evidence: A live IELTS study assistant for reading review, writing coaching, OCR-assisted material processing, and review-card based learning.
Next: Connect the learning evidence back into the personal OS story.

Research systemDeep learning system

An evidence-constrained thinking coach that helps users reason better, ground claims in source material, and preserve useful learning traces.

FastAPIReactGraphRAGEvidence ledger

Problem: Learning tools often produce answers without forcing better reasoning.
Protocol: A thinking coach should ground claims in source material and preserve useful learning traces.
Trace: Evidence-constrained reasoning, source grounding, learning traces, and GraphRAG experiments.
Evidence: An evidence-constrained thinking coach that helps users reason better, ground claims in source material, and preserve useful learning traces.
Next: Prepare a public case note without exposing private research data.