[KongchangAI]
Person

Jukka Blomberg

A developer known for operating a fleet of 13 AI agents and managing 14 startup companies, and an advocate of the agent management philosophy of narrow gatekeeping with broad autonomy.

Timeline (last 90 days)

Oct 14

Developer Jukka Blomberg runs 14 startups using 1 human plus 13 AI agents, of which 12 already have launched products.

Unverified50%
Oct 14

Over the past 90 days, Jukka's fleet of agents completed 1,235 outputs, all recorded in an append-only ledger.

Unverified50%
Oct 14

The five categories of operations Jukka set as not autonomously executable by AI agents include: operations involving real money, authentication and key management, sending and publishing externally, creation and scheduling of timed automation tasks, and destructive or irreversible changes.

Unverified50%
Oct 14

Jukka's core management logic is summarized as 'narrow gates, broad autonomy'.

Unverified50%
Oct 14

The envelope mechanism contains four elements: the specific scenario covered (only this one), hard limits (amount, frequency, number of people reached), stop-loss rules (stopping conditions), and how each automated action is marked in the log.

Unverified50%
Oct 14

Before an agent runs, there must be a contract specifying the person responsible, the task scope, the operating rhythm, the data it can read, the data it absolutely cannot see, the write locations, the applicable gates, the escalation path, and the termination conditions.

Unverified50%
Oct 14

The record format of the append-only ledger is: date | entry ID | source | project | status | notes, the status has only four kinds—BUILT, PARTIAL, ESCALATED, KILLED—and once a line is written it cannot be modified.

Unverified50%
Oct 14

Jukka's agent governance template is released free under the CC BY 4.0 license and includes principles, a gate matrix, agent contracts, the ledger format, the envelope mechanism, measurement methods, and how teams deploy it mapped onto Claude, Copilot, or OpenAI.

Unverified50%
Oct 14

Human-in-the-Loop means the agent must wait for human approval at every key action; Human-on-the-Loop means humans only monitor and act as a fallback while the agent runs autonomously within established boundaries; Jukka's five gates are a human-in-the-loop design and are triggered only for high-risk actions.

Unverified50%
Oct 14

Jukka's agent drafted a post for a company page and tried to publish it, and the safety check blocked the click.

Unverified50%

All Facts (10)

Unverified

Developer Jukka Blomberg runs 14 startups using 1 human plus 13 AI agents, of which 12 already have launched products.

50%
Unverified

Over the past 90 days, Jukka's fleet of agents completed 1,235 outputs, all recorded in an append-only ledger.

50%
Unverified

The five categories of operations Jukka set as not autonomously executable by AI agents include: operations involving real money, authentication and key management, sending and publishing externally, creation and scheduling of timed automation tasks, and destructive or irreversible changes.

50%
Unverified

Jukka's core management logic is summarized as 'narrow gates, broad autonomy'.

50%
Unverified

The envelope mechanism contains four elements: the specific scenario covered (only this one), hard limits (amount, frequency, number of people reached), stop-loss rules (stopping conditions), and how each automated action is marked in the log.

50%
Unverified

Before an agent runs, there must be a contract specifying the person responsible, the task scope, the operating rhythm, the data it can read, the data it absolutely cannot see, the write locations, the applicable gates, the escalation path, and the termination conditions.

50%
Unverified

The record format of the append-only ledger is: date | entry ID | source | project | status | notes, the status has only four kinds—BUILT, PARTIAL, ESCALATED, KILLED—and once a line is written it cannot be modified.

50%
Unverified

Human-in-the-Loop means the agent must wait for human approval at every key action; Human-on-the-Loop means humans only monitor and act as a fallback while the agent runs autonomously within established boundaries; Jukka's five gates are a human-in-the-loop design and are triggered only for high-risk actions.

50%
Unverified

Jukka's agent drafted a post for a company page and tried to publish it, and the safety check blocked the click.

50%
Unverified

Jukka's agent governance template is released free under the CC BY 4.0 license and includes principles, a gate matrix, agent contracts, the ledger format, the envelope mechanism, measurement methods, and how teams deploy it mapped onto Claude, Copilot, or OpenAI.

50%

Source Articles