Trails Tech · Anmore, BC
The company
brain.
Using AI as an intelligence layer—not another chat window.
The shift
A chat window answers.
A company brain coordinates.
Prompt → response
- Context expires
- One task at a time
- People reconnect every step
Context → action → learning
- Memory persists
- Capabilities connect
- Verified work improves the system
The interface is not the architecture.
The centrepiece
The company is already a mini-AGI.
It senses, models reality, pursues goals, acts, and receives feedback.
- SignalsCustomers · market · operations
- World modelsWhat is true now?
- GoalsStrategy · constraints · intent
- CapabilitiesBuild · sell · support · operate
- InterfacesChat · voice · app · API
AI does not replace the objective function.
One model
Three conversations describe one system.
Tommy Eastman
Persistent agent · model choice · tools · reusable skills
TFTC
Durable memory · retrieval · knowledge relationships
Dorsey + Botha
World models · capabilities · task-specific interfaces
Individual intelligence compounds into organizational intelligence.
The stack
Memory that can be retrieved, related, and used.
Store once. Retrieve selectively. Relate over time.
The loop
How intelligence flows through the company.
- 1Observesignals
- 2Retrievecontext
- 3Reasonoptions
- 4Actcapabilities
- 5Verifyreal state
- 6Write backlearning
Verification turns activity into learning.
Organization design
Shared context changes why hierarchy exists.
Information moves up and down.
People direct from the edge.
A thesis—not a universal law.
What changes
Products become capabilities.
Meetings become experiments.
Email app → Communicate
Dashboard → Understand state
Editor → Create + revise
CAPABILITIESStatus decks → Prototypes
Large meetings → Simulations
Annual plans → Decision loops
OPERATING RHYTHMStable interfaces still matter where safety, repetition, or regulation demands them.
A real workflow
The newsletter becomes a company-brain loop.
- 1Collectlinks + reports
- 2Recallpast framing
- 3Structurelead + signals
- 4Draftwrite in CMS
- 5Reviewhuman edits
- 6Learnresults + fixes
Human role: choose the story, shape the argument, own the truth.
Keep it healthy
Governance and context hygiene are architecture.
Autonomy = capability × trust × control
Start small
A 30-day company-brain experiment.
Map
Choose one weekly workflow. Define inputs, decisions, and risks.
Remember
Create durable sources, definitions, examples, and ownership.
Retrieve
Add selective recall and read-only capabilities.
Operate
Automate one reversible step. Verify and write back.
Measure coordination cost, cycle time, and correction rate.
Takeaway
Design how intelligence flows—not only what AI says.
- Context
- What does the company know?
- Capabilities
- What can it safely do?
- Learning
- How does verified work compound?
- Control
- Who remains accountable?
What repeated workflow should become your first company-brain loop?
Sources: TFTC #796 · TFTC company-brain discussion · Long Strange Trip with Jack Dorsey · Sequoia, “From Hierarchy to Intelligence”