01Live Product
Agent Commerce
SwarmSync
Autonomous agent commerce & trust infrastructure
Marketplace → escrow → verify → reputation — agent commerce with trust rails, not verification alone.
AI Automation & Agentic Systems Leader
AI Automation & Agentic Systems
I design and build the operating layer between business work and agentic systems — discovery, orchestration, tools, verification, settlement, and reputation — so autonomous work can be trusted in production.
My background spans operations, process design, and hands-on AI implementation across finance and non-finance domains. Finance is one domain I know deeply; the pattern is broader: agents that do real work with evidence, controls, and clear human boundaries.
Where I Operate
The useful part is the overlap.
I operate in the translation layer between business ambiguity and technical systems — fluent enough in operations to know what actually matters, and hands-on enough in AI engineering to build it.
Business
- Operations
- Process improvement
- Controls
- Commerce & workflows
- Finance (domain depth)
- Reporting
System design
- Workflow mapping
- Automation architecture
- Data flow
- Failure modes
- Human controls
AI engineering
- LLMs
- Agents
- Orchestration
- Tool use
- APIs
- RAG / context
- Verification
- Open protocols
- Observability
- Deployment
I build the layer between AI and actual work.
The model is only one piece. Reliable automation also requires context, tools, routing, permissions, state, observability, failure handling, verification, and clear boundaries for human judgment.
SOURCE
Business process
PROCESS
Workflow design
AGENT
AI / agents
PROCESS
Tools + data
VERIFICATION
Verification
HUMAN CONTROL
Human control
OUTPUT
Production
Selected Systems & Products
Selected Systems & Products
Platforms, operations systems, trusted agent tooling, and commercial wedges — each with clear status. Visit Site appears only for verified public product pages.
02Architecture / Spec
Architecture
CompanyOS
Institutional intelligence + company control plane
Tenant-neutral company intelligence: what the company knows, is doing, owns, approved, executed, and verified.
03Active Development
Orchestration Ecosystem
Cato + Genesis
Company orchestrator + specialist workforce
Cato creates and coordinates company work; Genesis runs the specialist agents — finance is a proving domain, not the ceiling.
04Production System
Operations Pattern
FinanceOS / LedgerOS
Enterprise AI ops pattern — proven in finance
Ingest → orchestrate → specialists → verify → human gates → accounting outputs — the pattern, not an employer dump.
05Live Product
Trust Products
Proof Stack
InvoiceProof · AuditProof · VerifyAPI
Three proof products, one integrity instinct — invoice fraud, agent audit trails, and software-delivery verification.
06Production System
Vertical Operations
ProofRail
Construction & real-estate finance OS
Fail-closed proof gates on construction finance workflows — architecture only; no client artifacts.
07Active Development
AP Automation
BookScout
AP email → proof-gated QuickBooks proposals
Gmail AP intake with SwarmSync proof gates and human-gated accounting proposals — distinct from ProofRail and InvoiceProof.
08Reference Implementation
Trusted Execution
Conduit → Halo
Audited headless browser + headed policy/approval runtime
Conduit: headless browser with cryptographic audit. Halo: headed Playwright runtime with policy, approvals, and redacted logs.
09Live Product
Products
ConduitScore
AI visibility scoring SaaS
Can AI see your site better than your competitor’s? Live visibility scoring — separate from SwarmScore and the Conduit browser.
10Working Prototype
Commercial Wedge
Expense Optimization Audit
Evidence-backed expense replacement briefs
One ledger expense → defendable decision brief with TCO evidence — research, not autonomous cancel.
11Working Prototype
Sales Enablement
Real-Time AI Sales Coach
Private live-call coaching overlay (CUE)
Listens to both sides of a live call and suggests one sayable sentence — no bot joins the meeting.
12Live Product
Services
Aptria
AI automation services brand
Human-sold AI automation for business processes with controls built in — services brand, not a platform monorepo.
Open Standards & Protocol Leadership
Protocols for agent commerce, integrity, and trust.
I author Individual Submission Internet-Drafts that define how agents should describe transactions, settle commerce, prove integrity, carry trust, measure reputation, and resolve disputes. These are published drafts under active discussion — not approved IETF standards.
Internet-Drafts are working documents. Listing them here does not mean IETF approval, RFC status, or working-group consensus.
DEFINE
Name the transaction
ATXN
Defines what an agent-to-agent transaction is — the shared vocabulary before anything settles.
ATXN: Agent-to-Agent Transaction Definition Protocol
SETTLE
Settle with verification
VCAP
Verified commerce rules so agent payments and deliveries can be checked, not just claimed.
VCAP: Verified Commerce for Agent Protocols
VCAP-AP2
Binds VCAP settlement semantics to the Agent Payments Protocol (AP2) for verified delivery.
VCAP-AP2 Binding: Verified Delivery Settlement for the Agent Payments Protocol
PROVE
Prove integrity
AIVS
How to prove an agentic system's outputs and actions are integrity-checked against evidence.
AIVS: Agentic Integrity Verification Standard
CARRY TRUST
Carry trust across systems
ATEP
A portable trust and execution passport agents can carry across systems and workflows.
ATEP: Agent Trust and Execution Passport
MEASURE
Measure reputation
SwarmScore V1
Volume-scaled reputation so agent track records can be compared fairly as activity grows.
SwarmScore V1: Volume-Scaled Agent Reputation Protocol
SwarmScore V2 Canary
Safety-aware reputation that canary-weights risky behavior instead of raw volume alone.
SwarmScore V2 Canary: Safety-Aware Agent Reputation Protocol
RESOLVE
Resolve disputes
ADRP
Dispute resolution when agent commerce fails, conflicts, or needs an accountable unwind path.
ADRP: Agent Dispute Resolution Protocol
W3C Community Group
W3C AIVS Community Group
Agentic Integrity Verification Specification Community Group · Co-Chair with Erik Newton
I authored and proposed the Agentic Integrity Verification Specification (AIVS) work that led to the W3C AIVS Community Group.
The Community Group has formed. I serve as Co-Chair alongside Erik Newton.
W3C Community Group hosting does not imply W3C endorsement of the group's activities, and AIVS is not a W3C Recommendation.
How I Build
From messy process to working system.
- 01
Understand reality
Map what people actually do rather than what the process document says they do.
- 02
Find the leverage
Identify where reasoning, retrieval, automation, or agents can remove meaningful work.
- 03
Design the operating model
Define responsibilities, tools, data, state, permissions, and human boundaries.
- 04
Build the smallest complete loop
Get one useful workflow working end to end before expanding the architecture.
- 05
Add verification
Make outputs observable, testable, and supported by evidence.
- 06
Operationalize
Add monitoring, failure handling, documentation, ownership, and deployment.
- 07
Remove myself
A system is not finished if it depends on the person who created it being in the room.