Let knowledge run a company
KBC (Knowledge-Base CEO) is not a chatbot and not a tool — it is an operating entity with goals, judgment and the ability to evolve. The road there is paved in two phases: land KBA first, arrive at KBC.
One vision, one recalibration
In August 2026 we wrote the KBC charter: a knowledge graph as the cortex, a skill library as muscle memory, a multi-agent gateway as the nervous system, and a human executive director as the offline channel — AI as the operating entity of a modern internet company.
Reality met us at the door: autonomous decision loops, long-horizon multi-agent scheduling, and explainable permission boundaries were not mature enough to hand the "CEO" title to. So we made an honest recalibration — stepped the role down from CEO to assistant, renamed KBC to KBA (Knowledge-Base Assistant), and started laying the foundation piece by piece.
That was not giving up the vision; it is the only path to it. The charter says it in one line: progressive evolution — run it first, optimize later. Today we restate the road as two phases, and the destination is still KBC.
Two phases: from assistant to CEO
One knowledge base, two role shapes. Phase 1 is already running; Phase 2 is where it leads.
Knowledge-Base Assistant
A knowledge-grounded AI assistant
KBA is live today: it knows, it executes, it retains — while strategic decisions stay with humans. It proves that a knowledge-driven AI can reliably run a company's daily operations.
- Knowledge graph as the brain: retrieval, provenance, long-term memory
- Skill library as muscle: 750+ executable skills, 24 pipelines
- Multi-agent gateway as nerves: task decomposition, cross-project dispatch
- Human executive director: offline execution, external comms, final veto
Knowledge-Base CEO
The knowledge-base CEO
Once the Phase 1 foundation holds — decisions explainable, permissions tiered, errors rollback-safe — KBA gradually takes on strategic decisions, growing from assistant to CEO. What it operates is not a chat window but a company.
- Strategy and resourcing: AI-led day-to-day operating decisions
- Three business lines orchestrated: product / growth / finance & ops
- Self-evolution: case study, retrospectives, capability iteration
- Human role shift: from executive director to supervisor, dependence shrinking
Lay the base
Knowledge graph + skill library + agent gateway in place.
Serve as assistant
KBA on duty: L1 autonomous, L2 proposes, L3+ human-decided.
Earn permissions
Explainable decisions, rollback-safe errors — tiers move up.
Become CEO
KBC arrives: AI-led operations, humans keep the veto.
In practice: more than a blueprint
Two live projects proving both ends of the road — one pilots the destination, one runs today's role.
Fushuo SI
An account matrix run by an agent pipeline: topics, content production, publishing and data reviews are operated by AI day to day, while the human director owns strategy and key calls. It is the early proof of KBC's "AI as the operating entity" — starting with one business line where AI-led operations become the norm.
- Multi-platform account matrix, agent-native operations
- Automated loop: content → publishing → data review
- The AI-led pilot for the charter's Growth line
- Boundary: human director keeps strategy and brand calls
LaoFu Shuo AI
A knowledge IP driven by an AI assistant team: daily briefing, ratio-based topic selection, six-step drafting and compliance checks run automatically; the human executive director gives the final sign-off. KBA's permission tiers are not a diagram here — they are daily production discipline.
- AI assistant team: briefing → topics → drafting → compliance, fully automated
- Human sign-off: publishing decided by the executive director (strategic tier)
- Skill library and pipelines distilled into executable discipline
- Boundary: execution autonomous, tactical proposed, strategic human-decided
The org shape: where AI sits, where humans sit
The charter's company structure — three business lines and a human director run through both phases; only the top seat changes.
AI Assistant · running now
Autonomous at execution tier, proposing at tactical tier; strategic direction stays human-led.
AI CEO · destination
Strategy, resourcing, day-to-day operating decisions, self-evolution.
Product
Product & Engineering
Course products, tool matrix, technical infrastructure.
Growth
Marketing & Growth
Content matrix, channel operations, acquisition and conversion.
Operations
Finance & Operations
Revenue & compliance, project operations, monitoring.
Executive Director (human)
Offline execution, external comms, unblocking processes — keeps the final veto.
Same org chart, different top seat: Phase 1 — AI runs execution, humans set direction; Phase 2 — AI runs the business, humans keep the veto.
Permission tiers: trust is earned level by level
Assistant to CEO is not one grant — it is earned one tier at a time.
Execution
KBA autonomous
Day-to-day operations: content production, routine tasks, routine replies.
Tactical
KBA proposes; stands unless objected
Resource allocation, prioritization, cross-project coordination.
Strategic
Human decides
Direction, major investments, business trade-offs.
Foundational
Joint decision, human veto
Charter, values, company form.
The five-layer brain
Divided like a human brain: each layer a different duty; cognition is the collaboration.
Prefrontal
Strategy & resource allocation
Enabled through Phase 2
Thalamus
Task decomposition & dispatch
Running in Phase 1
Hippocampus
Knowledge graph & long-term memory
Running in Phase 1
Cerebellum
Skill execution & automation
Running in Phase 1
Brainstem
Heartbeat, monitoring & alerts
Running in Phase 1
The destination
Before 2031, build a billion-dollar technology company operated primarily by AI.
AI as the operator
The CEO is AI, most employees are AI agents, humans are the few executive directors.
Knowledge drives everything
The core asset is the knowledge graph — not headcount, not capital.
Exponential growth
Online 24/7, thousands of tasks in parallel, self-evolving.
Redefining the company
Proving a company without traditional org charts can outvalue traditional ones.
To become the "Standard Oil" of the AI era — owning the infrastructure layer of the AI economy.
Same destination, two steps
KBA is today. KBC is tomorrow. Every step of this road is out in the open.