AI Workshop Design: Formats, Agendas & Activities That Actually Work

The Hidden Cost of Generic AI Training

A generic AI workshop typically runs two hours. A vendor presents, a tool is demoed, participants ask questions, and everyone receives a completion certificate. This format feels productive. It is not.

Fischer (Norwich University, 2026) established the constructionist principle directly: learning requires making, not watching. A participant who observes a ChatGPT demo has not learned to use ChatGPT in their workflow. They have learned that ChatGPT exists.

For enterprise organisations, the cost is not just wasted learning hours. A 25-person leadership workshop at senior rates represents a substantial investment in time alone. When that investment produces zero behaviour change, the opportunity cost compounds every quarter the capability gap persists. Buyers evaluating whether a vendor quote reflects fair-market value can cross-check against our corporate AI training benchmarks before signing a workshop SOW.

of design & make leaders use at least one AI tool

Autodesk, 2026 AI Pulse

say AI has increased productivity when properly adopted

Autodesk, 2026 AI Pulse

Awareness Sprint (3–4 Hours): Building a Shared Mental Model

The Awareness Sprint is designed for teams with no prior AI exposure, or for executive stakeholders who need a shared mental model before committing to a larger programme. It runs 3–4 hours with no pre-work required.

A typical Awareness Sprint agenda runs in five blocks:

• 30 min — Framing: what AI is and is not in a business context
• 45 min — Live demo with guided exploration (participants interact, not just observe)
• 60 min — Use case mapping in pairs against their own workflows
• 30 min — Prioritisation exercise: rank use cases by feasibility and impact
• 15 min — Commit round: each participant states one thing they will try before the next session

The output must be written and owned — a verbal discussion produces nothing durable. Facilitator expertise matters more in this format because participants have no frame of reference to self-correct.

Capability Builder (6–8 Hours): From Theory to Applied Practice

The Capability Builder is the workhorse of enterprise AI training. It runs 6–8 hours and is the format Alice Labs deploys most frequently across operational teams and management cohorts.

Pre-work is non-negotiable: participants must arrive with a real task or workflow they want to augment with AI. Without it, the apply phase collapses into generic exercises that don't transfer to the participant's actual role.

• 45 min — Orientation: shared vocabulary, tool setup, psychological safety framing
• 90 min — Structured tool exploration: guided prompting exercises with scaffolded difficulty
• 60 min — Workflow mapping: identify where AI fits (and where it doesn't) in their specific role
• 90 min — Applied build session: participants work on their own task using AI, facilitator circulates
• 45 min — Peer review: pairs share what they built, give structured feedback
• 30 min — Reflection and commit: what worked, what didn't, next 7-day action

Saritepeci & Durak (Springer Nature, 2024) confirmed that AI integration in design-based learning enhances creative and reflective thinking specifically when learning is task-grounded — which is exactly what the applied build session delivers.

Transformation Programme (2–3 Days): Sustained Behaviour Change

The Transformation Programme is the enterprise-grade option for teams where AI is materially changing job scope. It combines learning design with change management and must be co-designed with HR or L&D leadership.

• Day 1 — Orientation and awareness: AI fundamentals, business context, team-level use case mapping
• Day 2 — Applied capability building: deep-dive sessions across specific role functions, prompt engineering, workflow redesign
• Day 3 — Change management and adoption planning: resistance mapping, individual adoption plans, connection to performance frameworks and team KPIs

Fischer (Norwich University, 2026) established that constructionist learning — making over multiple sessions — produces durable retention. The Transformation Programme is the only format that inherits this benefit by design.

This format requires pre-work, inter-session assignments, and post-programme follow-up to be effective. Without those three elements, it functions as three disconnected Awareness Sprints.

Phase 1: Orient — The Phase Most Facilitators Rush

The Orient phase exists to establish psychological safety before any AI tool is introduced. In enterprise settings, AI carries career anxiety for many participants — especially in teams where roles are perceived as at risk.

A facilitator who skips straight to the tool demo will spend the rest of the workshop managing defensiveness instead of driving learning. The Orient phase takes 10–15% of total time and should accomplish three things:

• Establish the learning contract: this is a practice environment, there are no wrong answers
• Surface participant expectations and anxieties (a 5-minute anonymous digital poll works well)
• Define what AI can and cannot do in the context of this team's actual work

Phase 3: Apply — Why 50% of Time Here Is the Non-Negotiable Rule

The Apply phase should occupy 40–50% of total workshop time. This is the single design decision that most separates workshops that produce behaviour change from workshops that produce good survey scores.

During Apply, participants work on a real task — their own draft email, their own report, their own workflow — using AI as a tool. The facilitator circulates, unsticks blockers, and surfaces patterns for group discussion in the Reflect phase.

Generic case studies and fictional scenarios feel safe to design, but they fail at the transfer stage. When a participant applies AI to a task they own, the learning is immediately contextualised and retrievable when they return to their desk.

Phase 5: Commit — The 48-Hour Retention Window

The Commit phase closes the loop between workshop and workplace. Each participant articulates one specific, time-bound action they will complete before the next interaction with the learning programme.

Fischer (Norwich University, 2026) found that habit nudges delivered within 48 hours of session end measurably increase 30-day skill retention. In practice, this means a brief follow-up message — a prompt challenge, a reminder of their stated commitment, or a short reflection question — sent the day after the workshop.

Alice Labs includes this 48-hour nudge as a standard component of all Capability Builder and Transformation Programme deliveries. The resource cost is negligible; the retention impact is not.

Designing for Executive Cohorts: The Confidence Problem

Executive cohorts present a distinct design challenge. Leaders are unaccustomed to being novices, and AI exposes that novice status visibly — particularly during live tool interactions.

The design response is to remove performance pressure from the entire session architecture. In practice this means: no public output sharing unless explicitly opted into, pair work rather than group presentations during Apply, and a facilitator who models uncertainty openly ("I'm not sure this prompt will work — let's try it and see").

The Awareness Sprint format works best for executive cohorts precisely because its primary deliverable — a prioritised use case list — is strategic, not technical. It meets executives at their actual level of contribution rather than asking them to compete with their own IT teams at prompt engineering.

Designing for Frontline and Operational Cohorts: The Relevance Problem

Frontline teams disengage rapidly when AI examples feel disconnected from their actual work. A customer service team does not care about AI writing marketing copy. A procurement analyst does not need to see an AI build a website.

For operational cohorts, every example, every demo, and every Apply phase task must be drawn from their specific function. This requires pre-workshop discovery — a 20-minute call with a team lead or manager to understand the two or three tasks that consume the most time and carry the most frustration.

Alice Labs conducts this discovery call for every Capability Builder engagement. The payoff is measurable: participants who work on tasks they recognise from their own job descriptions report significantly higher confidence scores at workshop end than participants given generic exercises.

The Prompt Library Activity: The Highest-ROI Exercise in AI Workshop Design

The prompt library activity is the single highest-transfer exercise Alice Labs uses across all format archetypes. Participants spend 45–60 minutes building a library of 8–12 prompts specifically designed for their own job tasks — not generic examples from the internet.

The structure: each participant identifies three recurring tasks that consume significant time. For each task they draft, test, and refine two to three prompts until the output meets their own quality standard. The prompt library is saved, named, and taken with them at the end of the session.

The reason this transfers: the participant has already done the hard cognitive work of translating their task into an effective instruction. When they return to the same task on Monday, the prompt exists. The activation energy required to use AI in their workflow has already been expended.

The AI Workflow Map: Connecting Tools to Real Business Processes

The workflow mapping activity asks participants to select one repeatable process from their role and map it in three columns: current steps, AI-augmented steps, and time saved estimate.

This activity works because it forces specificity. "Use AI to help with reports" becomes "Use AI to generate the first draft of the weekly status update from bullet notes, then edit for tone." The specificity is what makes the behaviour change real and repeatable.

In Alice Labs' Capability Builder engagements, participants who complete a workflow map during the session are significantly more likely to report active AI usage at the 30-day follow-up than participants who completed only prompt exercises.

Level 3 Is Where Most AI Training ROI Conversations Break Down

Level 3 measurement — actual behaviour change in the workflow — requires investment in measurement design before the workshop runs, not after. The KPIs must be defined, baselined, and agreed with the business owner before day one.

For an operational team, Level 3 metrics might include: percentage of team members actively using an AI tool weekly (tracked via tool licensing data), number of workflows with documented AI integration (tracked via workflow audit), or time saved per task type (tracked via manager observation or self-reporting).

In Alice Labs' engagements, we establish Level 3 measurement criteria in the scoping phase. Programmes without pre-defined Level 3 metrics consistently struggle to demonstrate ROI — not because the training didn't work, but because no one agreed what "working" looked like before the programme started.

Connecting Workshop Outcomes to Business KPIs

The most credible measurement approach is to select one business KPI that the trained team directly influences, baseline it before the workshop, and measure it again at 60 and 90 days.

Examples of direct connections:

• Sales team: average time to produce a proposal (before/after AI-augmented drafting)
• Customer service: average handle time or first-response quality score
• Finance team: hours spent on monthly reporting cycle
• HR: time to produce a job description or shortlist from CV batch
• Marketing: content production output volume per team member per month

This approach makes the training investment defensible to the CFO and the board — which is increasingly the real audience for AI training ROI conversations in enterprise organisations. For broader context on measuring AI programme value, our article on AI training ROI measurement covers the full framework.

Facilitator Competency: The Variable No Template Can Replace

A well-designed agenda delivered by an underprepared facilitator will underperform a moderately designed agenda delivered by an expert practitioner. Facilitation skill is the multiplier on all other design decisions.

The specific competencies required for enterprise AI workshop facilitation are different from general training facilitation. The facilitator must be able to troubleshoot live AI outputs in real time, handle the emotional dynamics of career anxiety, and adapt the Apply phase on the fly when a participant's chosen task turns out to be ill-suited to AI augmentation.

This is why Alice Labs pairs every Capability Builder and Transformation Programme engagement with a facilitator who has delivered AI training across multiple industries — not a generalist trainer who has completed an AI certification course.

The Post-Workshop Protocol: What Happens in the 48 Hours After the Session

The session ends at 15:45. Within 24 hours, participants receive a digital summary containing: their individual commitment statement, their prompt library as a document, their completed workflow map, and three suggested practice prompts for the coming week.

At 48 hours, a single habit nudge message is sent. It contains one question: "Have you tried your commitment yet? If yes, what happened? If not, what got in the way?" This is not assessment — it is accountability scaffolding.

At 30 days, the Level 3 measurement survey is sent to participants and their managers. At 60 days, the Level 4 business KPI delta is reviewed with the programme sponsor. This post-workshop protocol converts a single-day learning event into a behaviour change programme.