I build the brain
for organizations.
AI is smart but blind.
I build the memory, reasoning, and continuity that turn agents into intelligence.
Startup CTO. Sold to private equity. Three years operating inside. Now I run engineering with a team of twenty agents.
Operating principle
Architecture before tooling. DDD. CQRS. Event Sourcing. Applied to humans, software, and agent teams.
Origin
Built a startup.
Sold it to private equity.
Then I broke the model.
I was a startup CTO. We sold to private equity, and I stayed on for three years operating inside — building engineering teams and adopting AI inside portfolio companies. I saw the dashboards lie. I saw the Jira motion. I saw the gap between code and reality widen every quarter.
Every CTO I sat across from was drowning in activity and starving for understanding. The reporting layer was theater. The ledger was somewhere else.
Jira tracks the expense report. GitHub holds the general ledger. Nobody was reading the ledger.
That became an obsession. Not another dashboard. Not another ticketing layer. An intelligence layer that connects code, strategy, meetings, finance, contributors, and outcomes into one execution graph.
I left to build it. InteliG is the result. Today I operate a one-person engineering org running ~20 agents — same architectural mindset I brought to PE, applied to humans, software, and agent teams.
Operating principles
What I believe,
and what I build accordingly.
Architecture before tooling.
Tools change every six months. Domain boundaries, aggregates, and event flows do not. I design the system, then choose the tools.
Memory + reasoning + continuity.
An LLM without memory is a stranger who forgets your domain every conversation. Intelligence is what persists across turns, sessions, and systems.
GitHub is the ledger.
Project management artifacts are accounting entries about the work. Code is the work. Truth lives in the repo, not the ticket.
Intelligence as the operating layer.
Not a chatbot bolted onto a dashboard. A reasoning system that observes execution, remembers context, and acts on intent.
Reasoning collapses on dirty data.
Intelligence systems are only as honest as the ledger feeding them. I run standing agents that curate the inputs continuously: repairing gaps, closing stale items, and treating code as the canonical source. Clean data is the prerequisite for reasoning, not a downstream nice-to-have.
Agents are teammates, not macros.
A real agent reasons. It has memory, judgment, and a place in the org chart. If your agent is a workflow with an LLM step, you have automation, not intelligence.
Build systems that tell the truth, or they will lie to you.
How I operate
I run engineering
with a team of twenty agents.
Same architectural mindset I brought to private equity. Applied to humans, software, and now agent teams. Each agent has scope, memory, and a place in the org. None of them are workflows with LLMs attached.
Read the operating model →Coding agents
Claude, Codex, Cursor, custom harnesses. Vertical slice execution, PRD-driven, memory-backed.
Personal cognition
I dogfood the same intelligence pattern InteliG ships at organizational scale, scoped locally to myself. Persistent memory across notes, meetings, and conversation.
Product agents
Cognis inside InteliG. Reasoning over execution graphs across code, strategy, meetings, and finance.
Ops agents
Outreach, content, scheduling, lead pipelines. Each with scoped memory and a defined role in the org.
Engagements
How I get hired.
Three productized tracks. Scoped, deliverable, finite. No fractional-CTO treadmill. Designed to make a decision, ship a system, or unblock the next phase.
Pre-acquisition
Technical Diligence Sprint
2 weeks+
A code-first read of the target. Architecture, team, AI readiness, exit-blockers. Written for the investment committee, not the CTO.
For: PE Operating Partners, Investment Teams
Deliverables
- Architecture and codebase truth assessment
- Engineering org map and bus-factor analysis
- AI readiness scorecard
- Risk surface and remediation cost estimate
Post-close
AI-Native Value Creation
90 days+
Install AI-native operating layer. Replace dashboards with reasoning. Ship cost takeout while raising velocity. Operator embed, not slideware.
For: PE-backed portfolio CTOs, Operating Partners
Deliverables
- Execution intelligence layer installed
- Agent-team operating model designed
- Engineering ROI baseline + 90-day delta
- Handoff playbook for the in-seat CTO
Team enablement
AI Coaching for Engineering Teams
4 weeks+
Three files per feature: PRD, log, README. Design, frontend, and backend specs live in a workitem folder next to the code. Single backlog across the repo. Concern-based separation, minimal cognitive collision. This is the operating discipline that turns AI into a real force multiplier instead of a 1.2x productivity rumor.
For: CTOs, engineering leaders, PE Operating Partners installing AI uplift across a portfolio
Deliverables
- PRD, log, and README scaffolded per feature
- workitem/ folder with action-items, design, frontend, backend specs as needed
- Single repo-wide backlog replaces Jira
- PMs onboarded to push specs the way engineers push code
- Coding agent harnesses installed and team-trained
Pre-sale or rescue
Exit Readiness & System Rescue
4 weeks+
Clean the codebase story for the next buyer. Or untangle a SaaS bending under scale and bad decisions. DDD/CQRS strategy to unblock the next phase.
For: PE firms approaching exit, founders in trouble
Deliverables
- Architectural diagnosis with prioritized fixes
- Pre-diligence remediation plan
- Documented system invariants and risks
- Founder/board-ready narrative
Bespoke
Custom Engagement
Scoped to fit
Your situation does not fit one of the three above. That is fine. We talk, scope a fixed deliverable together, and run it the same way: scoped, finite, no treadmill.
For: Founders, CTOs, or operators with a specific problem
Deliverables
- Scope defined on a 30-minute discovery call
- Fixed timeline, single named deliverable
- Same operating principles as the standard tracks
- Same finite end-point, no rolling retainer
Not sure which fits?
Short intro call. 20 minutes. Tell me the situation. I will tell you which engagement applies, or if none of them do.
What I build
Operator proof.
I do not just advise. I ship. These are the systems I have built or am building right now, applying the same architecture and agent operating model I bring to engagements.
Flagship — Execution Intelligence
InteliG
The brain for engineering organizations. AI-native intelligence layer over the execution graph: code, strategy, meetings, finance, contributors, decisions. Cognis reasons across all of it.
Subscription billing platform
Chargematic
Event-driven billing engine competing with Chargebee and Recurly. Strong DDD and CQRS. Plug-and-play integrations.
Java · Angular · Postgres
AI-first vertical SaaS
TreeInventory.ai ↗
AI tree inventory for arborists. Photo capture → AI extraction → property report. Walk the property, hand them the report before leaving the driveway.
React Native · Next.js · Supabase
Non-custodial automation
AI Trading Platform
Custom execution engine for automated futures trading strategies. Strong risk controls. AI-enhanced execution. No custody of client funds.
Java · custom engine
Writing
The thesis,
in long form.
Where the operating principles get unpacked. Where the contrarian takes get sharpened. No hedging.
Cognis
As a CTO, I Had One Question. Cognis Answered It.
Every CTO has asked the same question: what are my developers actually working on? Cognis moves beyond dashboards to evidence-backed answers.
Knowledge
Confluence Is Where Knowledge Goes to Die
Confluence. Notion. Google Docs. Internal wikis. All graveyards nobody reads. The future of documentation lives in the repo.
Method
The Signal Method
Most products are built on amnesia. The Signal Method connects every meeting, decision, and line of code into a single traceable system.
Code Intelligence
Your Jira Says 47 Bugs. Your Code Says 112.
Labels are manual, inconsistent, and useless at scale. Commit Intelligence reads the ledger directly.
Intelligence
Dashboards Are Not Intelligence. Stop Pretending They Are.
The entire category of dev analytics is stuck in the dashboard paradigm. CTOs do not need more charts. They need a system that thinks.