Competency DAG
Roots → branches → leaves with prerequisite edges. Same schema that powers prognosis and training paths.
Progress Intelligence
We map how learning actually happens.
A live competency graph that sees where every learner stalls, lets AI generate the fix, and stands up an entire exam vertical in days.
daysto launch a vertical
216competencies AI-drafted
100%AI outputs supervised
Selected: Perfekt · stall point — amber
The moat
Learning as a process you can see.
Nodes are competencies. Edges are prerequisites. Color is mastery. Amber is a stall. Purple is an AI intervention — drafted, tagged, and waiting for human bless. This is not a LMS dashboard; it is a living Progress Intelligence graph.
Selected: Perfekt · stall point — amber
Bottleneck detection
Where learners stall shows up as amber — the graph tells you what to fix before the exam date does.
AI intervention
Generate the drill, tag the leaf, send it through the supervision gate. Premium, not hallucinated.
Vertical Launch Engine
Paste an exam outline → AI drafts the competency model and content in days.
The outcome is a knowledge base you can launch: a reviewed competency graph plus tagged content — not a blank LMS. Watch % ready climb from outline to launchable.
01 · Outline
Paste the exam
Drop the content outline. The engine starts drafting the knowledge base immediately.
02 · Model
Competency graph
AI drafts the full competency model — structured for expert review before anything launches.
03 · Content
Knowledge base
Questions and study drafts land tagged to competencies — supervised before learners see them.
04 · Ready
Climb the gauge
Coverage, quality, and depth climb together until % ready hits launch.
Fresh verticals start at 0%. The Launch Copilot makes the climb to a launchable knowledge base visible.
The Supervising Agent
AI you can trust, because you can audit it.
Every generation is an interaction. Every review scores accuracy, tone, and hallucination. Gold examples calibrate the next draft. Premium — not slop.
Accuracy & tone
Feature-level quality dashboards. Red features raise Supervisor findings automatically.
Review → calibrate
Accept / revise / reject. Gold rows become few-shots for the next generation cycle.
Gate before trainees
AI outputs start in review — nothing ships unapproved.
generate→review→calibrate→improve
Durable learning
We measure retention, not cramming.
Score-on-the-day is vanity. The flywheel is whether knowledge survived — longitudinal proficiency with spike-then-fade detection on the graph.
Innovation Lab
The innovation lab.
mypath is where new AI learning modalities are invented and proven with real learners before they graduate into the CogMira platform.
AI avatar patient tutors
Conversational clinical partners that rehearse judgment under pressure — before the wards.
Real-time speaking evaluation
Live scoring of fluency, accuracy, and discourse — feedback in the moment, not after the session.
Self-supervising AI quality
Modalities that audit themselves — so only proven patterns graduate into the CogMira stack.
One engine, every exam
Partners as verticals on CogMira — not a marketplace.
Libelle, Lerntopia, T314M / Progress Intelligence, NextRecruitment — each a vertical running the same graph + engine + supervisor.
Libelle
Language vertical on the shared competency engine
Lerntopia
Program delivery powered by the launch stack
T314M / Progress Intelligence
Enterprise progress × learning graph
NextRecruitment
Role-competency verticals for hiring paths
Proof
The German taxonomy proof.
216
Leaves in the Deutsch A1–B1 competency tree — AI-draftable schema, human-blessed, with prerequisite DAG and CEFR/skill anchors. The first vertical that proved the engine.
Live verticals
Operator layers shipping: taxonomy, launch, supervisor.
Retention flywheel
Longitudinal data accrues as cohorts run — compounding moat.
Talk to us about your vertical.
Bring an exam outline. We’ll show the graph, the climb to a launchable knowledge base, and the supervision loop — on your domain.