Worldmodel · Embedded operators
An autonomous squad in 30 days.
We embed in one of your engineering squads and leave you with a number: the same scope held with half the people, or twice as much shipped at the same headcount.
Book 30 min Where is your squad today?
Free self-assessment, four minutes, result on screen.
Six levels. We install the fourth.
The full ladder →- L1 AI engaged
- L2 AI activated Most enterprise squads
- L3 AI-native team
- L4 AI-native organization We take you here
- L5 Self-optimizing
- L6 Autonomous value network
Most enterprise squads
We take you here
Your teams have Copilot, Cursor and Claude Code. You see more pull requests, not more value shipped. That is Level 2, and it is where most enterprise squads sit. An agent opening PRs inside a process designed for humans creates a review bottleneck, not an acceleration.
What changes in the pipeline
All fourteen steps →Four phases, fourteen steps. Each block below is one step. Two phases go fully agent-run.
- Human
- Human + agent
- Agent
- Gone
1 · Discovery
calls · reviews · telemetry · priorities
Today
Signals: Human + agentData and monitoring: Human + agentPriorities: Human + agentAt L4
Signals: AgentData and monitoring: AgentPriorities: Agent3 of 3 steps stop needing a person
Agents watch every signal, rank the opportunities and start work within goals your people set.
2 · Conception
tickets · spec · design · handover
Today
Ticket: Human + agentSpec: Human + agentDesign: Human + agentHandover: Human + agentAt L4
Ticket: AgentSpec: AgentDesign: Human + agentHandover: Gone3 of 4 steps stop needing a person
Agents turn goals into executable specs and the separate handover disappears. Design keeps human judgment.
3 · Production
code · test · review · deploy
Today
Code: Human + agentTest: Human + agentReview: Human + agentDeploy: HumanAt L4
Code: AgentTest: AgentReview: AgentDeploy: Agent4 of 4 steps stop needing a person
Agents implement, test, review and release passing changes, restoring the known-good version on failure.
4 · Marketing
docs · launch assets · communication
Today
Documentation: Human + agentLaunch assets: Human + agentCustomer communication: Human + agentAt L4
Documentation: AgentLaunch assets: Human + agentCustomer communication: Human + agent1 of 3 steps stops needing a person
Agents publish docs and prepare launch material with each release. Humans approve brand and positioning.
People keep goals, accountability and exceptions. Design and customer communication keep human judgment, at every level, including the ones above this one.
And in the squad itself
Six specialist job titles converge into one. The product engineer owns the feature from the customer signal to the release note, and agents do the work in between. It is a bigger job than any of the roles it replaces, and the engineers who live it do not want to go back.
Today
8 dedicated people
At Level 4
6 dedicated people
A hypothesis, not a prerequisite. Staffing conclusions come after the measurement, never before.
The 30 days
-
01 Embed
We join the squad, prepare the repo, and measure the baseline: lead time, throughput per person, where the time goes.
-
02 Discovery
Live signals become ready work. Feedback, usage and incidents feed discovery agents, checked against your PM’s own decisions.
-
03 Production
Spec to tested release. Coding, testing, review and deployment agents, wired with your engineers. Real changes ship behind flags.
-
04 Level 4
The squad runs the loop, including overnight. Reviewed at 9am. Then we remeasure, and you decide the next step with data.
Two workstreams run in parallel the whole month. The tooling: an agent-ready repo, connected signals, reusable agents for discovery, implementation, testing, review and release, with scoped access and tested rollback. And the change management: working alongside your PM, designer and engineers, answering doubts with evidence, coaching the squad until it operates the workflow itself.
Three ways to start
5 days
Diagnostic
€8,000
An audit of one squad and a costed 30-day plan. To know where you stand and what it would yield.
Credited against the Pilot if you sign within 30 days.
30 days
Pilot
€30,000
One squad taken to Level 4, measured before and after.
The engagement most CTOs choose.
90 days
Factory
€120,000
Up to 8 squads. The internal platform, the playbook and the metrics, with your champions trained to continue without us.
Book 30 min Two senior operators on every engagement. No subcontracting, no junior bench. Prices exclude VAT.
We have done it ourselves.
Vizir
Founded in 2018, taken to €1M ARR without raising, sold to LumApps at the end of 2022.
LumApps
Ran engineering at the cutting edge, putting AI methods to work across every team. Took the AI product line from €1M to €10M ARR in three years, inside a scale-up of several hundred people.
Otto, and today
Otto runs on two people where it would have taken ten. We now bring that speed of execution to larger tech groups.
We do not sell a method we read about. We show you the repo.
Questions we get
Does it work on legacy code?
Yes. That is where the gain is largest. The context layer we build exists exactly for that.
Which tools?
The ones you already have, if that is Claude Code, Cursor or Copilot. Otherwise we recommend and you choose. We resell no licenses.
What about code security?
Nothing leaves your perimeter. We work on your accounts, your models, your SSO. If you already run an internal platform, we build on it rather than replace it.
How much of my team’s time does it take?
One tech lead at half time, and the squad working normally. No two-day workshop, no seminar.
What does autonomous mean, exactly?
Agents run the build, the tests and the first review end to end, and people keep the intent, the trade-offs and the final call. You can measure it: lead time, throughput per person, and the team size it takes to hold a scope.
Are you an agency?
No. Two founders who built and sold software companies, and who build with agents every day. We take one engagement at a time, two at most.
Thirty minutes to know if this is for you.
We look at one of your squads, tell you what we would do in 30 days, and what it would yield. If it is not the right moment, we will say so.