The work grows. You remain outside.
Each box is a decision you cannot yet explain. Red means opacity, not an error in the work.Involut · Pinguinista
Understand what you build.
Grow through the work.
AI can produce a work before you understand it. We want working with it to create opportunities to learn, question and decide, too.
Augmented Mastery
To develop understanding, judgment and the ability to intervene while working with AI. Delegating execution can coexist with building your own mastery.
The manifesto expresses a choice about how we want to work and learn. It does not depend on AI reaching a ceiling or promise to protect a job.
Read the English manifesto ↗Open text · CC BY-SA 4.0
Auctor
Use decisions from your own work to get oriented, take a position, check evidence and revise your understanding.
The starting point depends on what you know about that work. Agreeing with AI does not demonstrate mastery: your reasons, ability to intervene and transfer to another case matter.
Read the Auctor draft · Spanish ↗Draft · Parts 2–6 pending · CC BY-SA 4.0
Involut
A first application of Auctor to agent-produced code. Decision cards, questions before revealing, access to original material and notes that return to your agent.
The tool already lets you explore changes and discuss them. Full adaptation to Auctor and its evaluation are in development. It does not award a certification or mastery score.
Read the tool's contract ↗Software · AGPLv3 · repository private for now
An everyday decision
A working answer is not the end of the conversation.
This example illustrates an Auctor practice. It is not a demo of the viewer or an assessment of your ability.
Your agent added a cache to the user profile to reduce database queries.
What should happen when the profile changes?
Before looking at the solution, consider which data could become stale and where invalidation should happen. If you need more context, that is a useful starting question.
This note stays on this page and is lost on reload. It is not sent or graded.
Open for learning. Clear about what we share.
You can share and adapt the manifesto and Auctor with attribution and share-alike terms. Involut has a separate software license. The texts are published; the product repository remains private during development.