Skip to main content
Discith logo
Discith · the fluency module

Fluency, not fear.

Tools don't create capability — people who can direct and verify them do. Discith builds AI fluency as a standing capability: measured on one ladder, refreshed continuously, carried by community.

Built for organizations where knowledge is the work — starting with Dutch higher education (WO & HBO).

The problem

Training theatre doesn't build fluency.

Most AI adoption stalls on the same mistake: deploy the tools and assume people will figure them out, or run a one-off training that is stale by the next quarter. Meanwhile the capability moves in months — and fluency has no "completed" state.

Common failure modes
  • One-off trainings, stale by the next quarter
  • Ungoverned shadow use fills the gap
  • No shared language between leadership and the floor
What it does

Fluency as a standing capability.

People who can direct and verify agents — not just use them

One shared language for AI across the whole organization

A self-sustaining community that outlives any single course

How it works

Five mechanisms, one system.

Discith is delivered as software, as method, and as an operating model. The five mechanisms interlock: the ladder gives the language, the pipeline produces the curriculum, the arena trains the judgment, the library and champions make it spread and stick.

The scale

Fluency Ladder

Five levels — Aware, User, Critic, Director, Orchestrator — applied per audience tier. Leadership targets Critic as a minimum: leaders first, never a diluted summary.

The engine

Fluency Pipeline

Six human-gated stages derive target levels and a living curriculum from your own documents. AI drafts, humans decide — every row carries provenance.

The training

Practice Arena

Train judgment against an AI configured to question, counter and expose weak reasoning — and never hand over the answer. Every session ends with a scored debrief.

The memory

Use-case Library

A curated catalogue of what colleagues actually delegated to AI, how they verified it, and what it yielded. Every entry is a worked example of the ladder in action.

The carriers

Champions network

Named per faculty or department, with an explicit, revocable curation mandate — a respected practitioner can lead the community without being a manager.

The Fluency Ladder

Five levels, one shared language.

Each level presupposes the one before it. Targets are set per audience tier — leadership, practitioners, process owners, everyone — and leadership goes first, at level 3 minimum.

1
Aware

Knows what generative AI and agents can and cannot do, and the basic risks.

2
User

Uses AI effectively for bounded tasks; recognizes unusable output.

3
Critic

Evaluates AI output systematically — judges how it was produced, not only the result.

4
Director

Directs AI on real work: delegates deliberately, guards quality, owns the outcome.

5
Orchestrator

Designs and governs combinations of agents and people — and transfers fluency to others.

The Fluency Pipeline

Six stages, human gates on every one.

The pipeline derives fluency targets and a living curriculum from your organization's own documents — not a generic course catalogue. AI drafts, humans decide: every stage ends in an explicit approval gate, and re-running a stage clears everything downstream, so derived content never lingers stale.

01
Intake
Gate: is this really us?

Exit profiles and role inventory extracted from your own documents, each with confidence scores.

02
Future scenarios
Gate: plausible — no hype?

A credible picture of each audience's work at the horizon year: which tasks become agent-assisted, which skills stop being core.

03
Delta analysis
Gate: the changes that matter?

Per audience: what to keep and deepen, what shifts, what is new, what degrades.

04
Fluency map
Gate: targets right — leaders first?

Named competencies with target and current levels per audience, each gap anchored in the delta analysis.

05
Curriculum
Gate: would we actually run this?

Learning lines and small, replaceable modules — sized in hours, not semesters — covering every competency.

06
Assessment
Gate: tests judgment, not artifacts?

Process-over-product assessment forms per learning outcome, with anti-pattern flags.

An independent, advisory AI critic flags hype, ungrounded claims and implausible levels on every stage's output. It informs the human gate — it never blocks. People do.

Where do you stand

From ad-hoc to self-sustaining.

Five maturity stages describe how far fluency has become a system rather than an event. A one-hour QuickScan locates your organization on the scale — and names the failure mode keeping you there.

M1
Ad-hoc

Individual experimentation; fluency invisible and unmanaged.

M2
Aware

Fluency named as a capability; one-off trainings; no targets, no community.

M3
Practicing

Ladder adopted, targets set, structured practice underway, first champions.

M4
Embedded

Living curriculum on cadence; library curated; fluency feeds role design.

M5
Self-sustaining

The community carries the system — the organization learns faster than the tech changes.

Part of the same house

The Learn stage of the journey.

Discith is the fluency module of the Orgith suite. Role Assessor surfaces which tasks shift to agents; Discith builds the fluency people need to direct and verify them; Crescith redesigns the roles around what both produce.

See the whole suite

Discith connects:

  • Role Assessor — the tasks people must learn to direct
  • Crescith — fluency levels feed the role redesign

Governance guardrails become curriculum: people learn the rules of the road as they learn to drive.

See Role Assessor in action.

Book a demo with Wiemer — one conversation is enough to see whether it fits your organisation.

Book a demo