Skip to content

Work / Evidence

Evidence that AI can become organizational capability.

Not a product shelf. These are inspected samples from organization fieldwork, enterprise AI systems, product experiments, and public writing.

01
Organization fieldwork

AB InBev, SNOWPLUS, Longfor: M&A integration, 0-to-1 building, business replication, and organization efficiency.

02
Enterprise AI systems

Whoos Solutions and related product work tested how AI enters workflow, knowledge, collaboration, and accountability.

03
Product samples

One Page, MARGIN, FileFlow, Fairmate, and SFA turn fuzzy intent into inspectable operating loops.

04
Public thinking

Essays and the AI Organization OS handbook make the method visible enough to inspect, reuse, and question.

Product samples

Each product proves a different operating pattern.

Focus loop

One Page

Live sample
Focus in one gesture.
Problem
Meetings, ideas, and tasks scatter across tools.
Agent role
Turn raw input into actions, context, and next steps.
Proof signal
A note moves from recorded to ready to act on.
Intent parser

MARGIN

Live sample
Natural language to action.
Problem
People express intent in language while software asks for fields.
Agent role
Parse action, object, and constraints into an executable workflow.
Proof signal
One sentence becomes a next move without teaching a new interface.
Safe automation

FileFlow

Live sample
Automation with boundaries.
Problem
File automation becomes dangerous when nobody can see what changed.
Agent role
Split bulk work into auditable steps with human approval at key boundaries.
Proof signal
Automation becomes inspectable, bounded, and reversible.
Rule-system case

Fairmate

Case study
A working rule system, and why I paused its productization.
Problem
Compensation, evidence, procedure, and communication boundaries often sit in different rule languages.
Agent role
Turn public rules, regional parameters, evidence checklists, and process steps into a self-check path.
Proof signal
The 33-city rule base, regional parameters, and evidence checklist prove that ER knowledge can be structured; C-side productization is paused, and the asset can later be reframed as employer-side compliance capability.
Agent team delivery

SFA

AI sales coach close to real business context.

Meeting recordings, business knowledge, and sales methodology were distilled into specs, then built through an agent team delivery loop.

52
development specs
2,716
passing tests
4 days
delivery window
Advisory bridge

If your AI work has demos but not operating capability, start with the evidence.

Send the context: where AI is already used, where the loop breaks, and whether the real blocker is in roles, workflows, knowledge, accountability, or governance.

Evidence