Standing

Record · Method and standing · October 2026

How the VUDU WULF agent system is governed, and an estimate of how it compares.

VUDU WULF is run by one person working with a team of named AI agents. The agents do the work; authority stays with the human. This page documents how that is set up and checked.

Method

The rule the whole system is built on: Responsibility may be delegated. Authority may not be inferred.

  • A versioned constitution

    A written, versioned constitution with its own amendment procedure. Changes need a deliberate human step; a chat message is never an amendment.

  • An authority gate and log

    Nothing goes ahead without a typed, single-use human approval. Every decision is written to a tamper-evident, append-only log. The KO‑OS spine passes 109 of 109 checks.

  • Tested in 4 languages

    The same gate contract is implemented and tested in Python, Go, Rust and TypeScript.

  • 13 agents, zero authority

    13 named agents, each with a defined role, mission, behaviour and limit. None of them can approve, execute or change the constitution. Decisions stay with the human.

Standing (estimate)

Among people worldwide running a comparable setup (a team of 10 or more named AI agents, a written per-agent constitution, and a human-authority rule), this one is estimated at about #2,150 of roughly 12,000, around the top fifth.

Current central estimate · 9 Oct ~12:00 AEDT · up from ~#2,250 at 09:56~#2,150 of ~12,000Likely band #1,000 to #3,800
Position in groupAbout the top 18%Band spans roughly the top 8% to top 32%

Where ~#2,150 sits among ~12,000

Each dot is one person in the estimated group, ordered from #1 at top left.

  • Estimated position, ~#2,150
  • Estimated ahead (#1 to #2,149)
  • Likely band, #1,000 to #3,800
  • Rest of the group
#13,0006,0009,00012,000
The group size itself is uncertain: anywhere from about 1,000 to 150,000 people.

Weighting used: governance design 40%, day-to-day operation 20%, results 25%, published reach 15%. Governance alone would rank higher; results and reach pull the overall figure down.

Still growing

It's early. The system is a few weeks old, the method isn't published yet, and commercial results are still to come.

How the estimate was made

A step-by-step estimate. Each step after the first is a proxy, taking the middle of a low–high range.

  1. 8.2B people
  2. ~1B use generative AI
  3. ~2.2M build or run their own agents
  4. ~71k run a team of 10+ named agents
  5. ~12k with a per-agent constitution and human-held authority

Sources

  1. Microsoft AI Economy Institute, Global AI Diffusion Report, Q2 2026 (opens in a new tab)
  2. Stack Overflow, Developer Survey 2025: AI (opens in a new tab)
  3. Gartner, AI agent deployment poll (June 2025) (opens in a new tab) and agent governance survey, as reported by IDM (opens in a new tab)
  4. Deloitte, State of AI in the Enterprise 2026 (opens in a new tab)
  5. LangChain, State of Agent Engineering (opens in a new tab)
  6. Pan et al., Measuring Agents in Production (arXiv, 2025) (opens in a new tab)

Recorded 3 October 2026. Updated as results come in.