Training · OrchestratorMoveemergingverified

Seed the Veterans

also known as Embed Pilot Power-Users, AI Pioneer Embed, AI Seeding, Power-User Deployment

Take people who already use AI well — early adopters from a pilot or power-user cohort — and physically embed them in new teams for a fixed sprint. The team learns by watching, asking, and copying, not by sitting through a course.

How the learner advances

flowchart TD A[Pilot creates veteran pool 4-8 week cohort with genuine AI depth] --> B[Assign veteran to receiving team 2-4 week resident expert role] B --> C[Veteran works visibly shows prompts + outputs + iterations on real work] C --> D[Daily open-problem sessions team brings real problems, veteran solves live] D --> E[Team tries variants immediately capability transfers through observation] E --> F[Document emerging use patterns prompt templates + pitfalls co-owned by team] F --> G{Embed window over?} G -- No --> C G -- Yes --> H[Veteran rotates to next team] F --> I[More capable team + team AI use pattern document]

Intent. Transfer working AI capability to new teams through direct peer observation and co-working rather than through any form of instruction.

When to apply. Apply after a pilot has created a pool of genuinely capable AI users and a new wave of teams is starting to receive access. The gap this pattern closes is the one between 'I have the tool and I completed the training' and 'I use the tool fluently in my daily work.' Embed veterans in teams where daily work is sufficiently similar to the veteran's background for the modeling to be meaningful.

Threshold — earns the next step. The team can run a daily open-problem session without the veteran, using AI to solve a real problem from their work, and can extend or update the team AI use pattern document without external help.

Masterpiece — the artifact that proves it. A team AI use pattern document — built collaboratively with the veteran during the embed — containing at minimum five prompt templates for team-specific tasks, three workflow integration points, and two known pitfalls with workarounds, maintained and updated by the team after the veteran has left.

Example scenario

A media company's content production team has just been given access to an AI writing assistant after a successful pilot in the technology team. The content team has 18 people — writers, editors, and social media managers. They completed mandatory AI literacy training at 100%, but three weeks after tool access, usage logs show only three people used the tool more than five times. The head of content asks for a veteran from the technology team pilot. A senior developer who used AI daily during the pilot moves into a dedicated desk in the content bullpen for three weeks. On day one she opens the AI tool and writes a real blog post draft in front of the team, narrating every step: the brief she gave it, why the first draft was too generic, how she refined the prompt, what she kept and what she rewrote. Day two she runs the first open-problem session: the social media manager brings a real brief for a campaign announcement. The veteran solves it live in 14 minutes. The social manager immediately tries a variant for a different campaign and produces a usable draft in 20 minutes. By week two the team has produced nine prompt templates for their specific content types. The veteran documents them in a shared Google Doc before leaving. At the end of the three weeks, the team's weekly AI use rate is at 81%. The masterpiece is the shared prompt library: nine templates, four pitfalls noted with workarounds, and three workflow integration points — all built by the team, maintained by the team, and still in use six months after the veteran left.

Facets

Container — embedded expert
Mode — peer modelinghands-on coachinguse-case discovery
Reach — team
Persona — AI pioneerearly adopterembedded expert
Craft (AI Fluency) — FluencyFlow
Guardrail — veterans must have genuine AI depth, not just enthusiasmavoid creating a dependency that blocks team self-sufficiency

Inputs

Pool of verified AI-capable veterans — People who have come through a pilot or extended use period and demonstrably do real work with AI — not just enthusiasts, but people who can show finished work they produced with AI assistance and explain their process.
Receiving teams with genuine work to do — Teams that have active, non-trivial work underway — not teams set up for a training exercise. The veteran's modeling is most powerful when the team can immediately try what they just watched on something that actually matters.

Outputs

More capable team — A team that has moved from tool access to genuine daily use through direct observation of a peer doing real work with AI, not through instruction.
Team AI use pattern — The masterpiece: a documented set of AI use patterns specific to the team's actual work — prompt templates, workflow integration points, known pitfalls — built by the veteran and the team together during the embed, and retained by the team after the veteran leaves.

Steps (6)

Run a pilot to create the veteran pool
Before any embedding, run a 4–8 week pilot with a motivated cohort to produce people with genuine AI depth. The pilot's output is the veteran pool; without it, embedding spreads enthusiasm rather than capability.
Assign one veteran to each receiving team
Place one veteran in each new team for a 2–4 week resident expert period. The assignment is a role: not to teach, but to work alongside the team and show AI in action on real work. The veteran uses AI visibly, explains their reasoning out loud, and invites the team to try the same move immediately.
Demonstrate with real work, not slides
The veteran's daily practice is the curriculum. Show the prompt, show the output, show the iteration, show the failure. Narrate the thinking: 'I am asking it to do this because… this output is not quite right because… I would change the prompt by…' This is the move the course could not deliver.
Hold daily open-problem sessions
Schedule a 20-minute daily slot where any team member brings a real current problem and the veteran solves it with AI in real time, narrating every step. The team watches and immediately tries variants. The problem must be real; synthetic exercises kill the modeling effect.
Document the team's emerging AI use patterns
By week two, start capturing what is working in the team's context — prompt templates for their specific task types, workflow integration points, pitfalls discovered. This documentation is co-owned by the team, not just the veteran. Before the veteran leaves, the team should be able to maintain and extend it.
Rotate veterans across teams
After the embed period, move the veteran to the next team. Do not extend the embed beyond its window — a team that retains a veteran indefinitely never develops self-sufficiency. Rotation also prevents veteran burnout and ensures skill spreads broadly.

Principles

Showing real work is the mechanism — observing a peer solve a real problem with AI transfers more capability than any hour of instruction because the learner sees the full process, including failures and corrections.
Self-sufficiency is the exit condition — the embed is successful when the team can extend the patterns the veteran left without needing the veteran's presence.
Rotation is mandatory — a veteran who stays too long creates a dependency and stops spreading capability to teams still waiting.

Unlocks methodologies (1)

A learner who completes this pattern is equipped to execute these methodology families:

Deployment & Operations

Known uses (2)

Known failure modes (2)

[veteran-as-permanent-dependency]
Anti-pattern: the veteran embed extends indefinitely because the team finds it convenient to keep asking the veteran rather than developing their own fluency. The team becomes dependent rather than capable, and the veteran cannot move to the next team. Hard embed-end dates prevent this.
[embed-without-documentation]
Anti-pattern: the veteran models excellent AI use for the team but nothing is captured in writing during the embed. When the veteran leaves, the team reverts to pre-embed behaviour within weeks because there is no artefact to anchor the learning.

Related trainings (3)

Sources (3)

Provenance

Ecosystem: enterprise
Added to catalog: 2026-05-27
Last updated: 2026-05-27
Verification status: verified

How the learner advances

Steps (6)

Run a pilot to create the veteran pool

Assign one veteran to each receiving team

Demonstrate with real work, not slides

Hold daily open-problem sessions

Document the team's emerging AI use patterns

Rotate veterans across teams