Core architecture
Capability Intelligence
Persistent, evidence-based understanding of capability, context, goals and outcomes. External AI remains replaceable; evidence and continuity persist.
Architecture in developmentIdeas & products
Experiments in capability intelligence.
LecXie is a family of developing concepts exploring how persistent evidence and increasingly capable AI could change learning, careers, organisational decision-making and workforce management.
These are deliberately labelled by maturity. Each status shows how far the idea has developed and what remains to be explored.
Career
Career Intelligence
Opportunity discovery, career memory, evidence-based matching, trajectory and application intelligence for complex careers.
Prototype / conceptCareer Hub · Work in progress
Your career,
quietly moving forward.
Tell LecXie naturally.
Let the system do the trawling.
Spend attention where it counts.
Explore the public Career Hub preview. Access to the private beta requires approval.
Explore the Career Hub →Learning
Learning & Capability
Adaptive journeys built from the smallest useful intervention, tied to real capability needs and evidence of application.
In developmentDeliberation
Conflab
Structured AI-supported reasoning using multiple perspectives, evidence, challenge, uncertainty and progressively validated conclusions.
GPT demo / developing concept
Try the early demo
Explore a topic with AI personas and a facilitator. Join the conversation, debate a motion or watch different perspectives unfold.
Try the Conflab GPT demo →Opens in ChatGPT. You may need to sign in.
Conflab · Illustrative conversation
YouCould an AI agent take on this decision?
ConflabLet’s examine the evidence, the consequences and who remains accountable.
Explore the conversation →
Delivery perspectiveWhich part of the task is repeatable, and what would a successful result look like?
Challenge perspectiveWhat happens if the agent is wrong? Which decisions require human approval?
YouIt could prepare a recommendation, but a person should approve the decision.
Working conclusionStart with recommendations under human review. Compare results against agreed criteria before considering more authority.
Still uncertainHow reliably will it handle unfamiliar cases? Test that explicitly.
A scripted example of the concept, not a recording of a working product.
Workforce
Mixed Workforce
Organisation design for humans, AI assistants, autonomous agents and systems sharing work, authority and accountability.
Research propositionAssurance
Agent Competency Assurance
Capability ≠ competence ≠ authorisation. Exploring continuous evidence of agent performance under defined operating conditions.
Research propositionDesign principle
“Better external AI should make LecXie better, not obsolete.”
The durable layer is longitudinal evidence, decisions, outcomes, provenance and context — not a bespoke general-purpose model.
Follow the thinking
From idea to evidence.
Explore the developing Agent Competency Assurance framework, the seed-to-forest worked example, and the Papers programme.
These are current research artefacts and concept descriptions; the Papers page identifies work still planned.