MY AI WORKFLOW
CARTEL
Carter’s Autonomous Research Team & Executive Lab
First find the method, then work by it.
When the work ends, the experience stays. In CITADEL, connected AIs solve unfamiliar problems together; this system remembers what worked and reuses it from the next job on.
The more I work,
the more ways of working pile up.
B runs a set pipeline.
Generalist
Changing the team and method to fit an unfamiliar problem
ADAPTIVE TEAM Flexible roles and routes
Astra / Fable decides, and Luna / Flash widen the search. Roles and tools are picked again based on the results. It is a general-purpose team that looks for the very method of solving an unfamiliar problem.
A proposal in a field I have never handled.
What should be researched, and how should it be checked?
C / A → B
Locking in what worked
Setups that worked again and again, turned into a pipeline
researcher · writer · reviewer
✓research → write → verify
✓sources · templates · evidence criteria
✓C is not a third team;
it is the process that turns explored experience into a specialist team.
Specialist
Running a set pipeline again and again
DEFINED PIPELINE roles · order · tools fixed
Proposals 01 · 02 · 03
↓- 01ResearchResearcher
Collecting sources
- 02WriteWriter
Proposal template
- 03VerifyReviewer
Evidence criteria
Verified output and the next job with the same setup
In a field where the method is already known, roles, order, tools and verification criteria are fixed in advance. It is a specialist team that runs new input through the same pipeline efficiently.
Reuse the proven proposal team.
In a familiar field, start from B.
Results and feedback are remembered and fed into the next round of exploring and running.
Work together
The CLIs connected in CITADEL split roles and work. Unfamiliar problems get an exploring team; familiar ones get a proven specialist team.
Remember what worked
Not just the output: which roles, order, tools and checks worked well are kept too. The reasons for revisions and human feedback become context for the next job.
Make the next job better
The accumulated methods are reused and refined again with what running them teaches. It is a loop where one result becomes the next team’s starting point.