What's inside this template

Who it's for

AI Operations Leads, programme leads and transformation leads reporting AI programme progress to a steering group, executive committee or board

When to use

Monthly by default (fortnightly if the programme is in an early build phase or decisions are blocking); starting from your first steering group and continuing through the programme lifecycle

Key benefit

An update structure that leads with measured outcomes, separates claimed from verified value, scores opportunities by ICE (Impact × Confidence × Ease), treats stalled work as reportable, and leaves room for your own judgement, so the group sees delivery confidence rather than activity theatre

Sections included

Header: period, author, overall status and why in one line
Outcomes: what changed in the business
Opportunities and deliverables pipeline
Top opportunities by ICE score
Stalled items and asks
Decisions needed
Risks and security
Adoption
Next month
My view
Glossary: key terms

Complete template content

Claude skill ai-steering-group-update

This page comes with an installable skill. Install it, tell Claude what you are working on, and it works through the method on this page with you and produces the output.

Claude desktop or Claude.ai

Download the .zip, then go to Customize > Skills, upload the file, and enable the skill.

Claude Code

Unzip it into ~/.claude/skills/ for every project, or .claude/skills/ for one. Then run /ai-steering-group-update.

Download the skill (.zip)

Prefer Google Docs? Open this template in Google Docs to make a copy and customise it directly. Open in Google Docs →

NOTE: To use this template, copy the content using the “Copy page” button above, then fill in each section for your period. The free Claude skill above drafts it from your notes, a tracker or the Kowalah platform (when connected), leaving OPEN placeholders for missing baselines.

Cadence: Monthly is the default. If your programme is in an early build phase, or the group needs to unblock decisions quickly, run it fortnightly. Keep the same structure, shorten “Next month” to “Next two weeks”, and update outcomes only when there’s a new measurement.

The one-page AI steering group update

This update carries your name: it’s how the value you’ve created gets seen and credited. Fill it in the week before your steering group. Keep it to one page. Lead with what changed in the business, not with activity. Every number should say how it was measured and whether someone outside the AI team has verified it.

Field	Entry
Period	[e.g. September 2026]
Author	[Name, role]
Audience	[Steering group members]
Overall status	On track / At risk / Off track
Why, in one line	[The single reason for that status]

1. Outcomes: what changed in the business

Only processes with a baseline belong here. No baseline? It goes in the pipeline section.

Process	Business owner	Outcome type	Baseline → Now	Period	Unit	Basis	Claimed or verified	Verified by	Evidence link
[Process]	[Name]	Revenue / Cost / Productivity / Adoption / Other	[x → y]	One-off / Weekly / Monthly / Quarterly / Annual	[hours/week, days, £, %]	Estimated / Sampled / Measured	Claimed / Verified	[Name, role]	[Optional link]

One-line summary of value this period: [ ]

2. Opportunities and deliverables pipeline

These are leading indicators, not results. Say so.

Opportunities

Status	Count now	Change since last update
Captured
Shortlisted
Promoted
Parked
Rejected

Deliverables

Status	Count now	Change since last update
In progress
Done

Next three deliverables to complete:

[Deliverable]: owner, target date
[Deliverable]: owner, target date
[Deliverable]: owner, target date

3. Top opportunities by ICE score

ICE = Impact × Confidence × Ease (each 1-10, composite out of 1000). Shows what’s worth promoting next.

Opportunity	Team	ICE score	Impact rationale
[Opportunity name]	[Team]	[I×C×E = score]	Business / Productivity / Adoption: [one line]

4. Stalled items and asks

Stopping is a result. An honest update builds more credibility than an inflated one.

Item	Why it’s stalled	What unblocks it	Who can	By when

Stopped this period, and what we learned: [ ]

5. Decisions needed

Decision	Options	My recommendation	If deferred	Needed by

6. Risks and security

Top risk	Owner	Mitigation
1.
2.
3.

Tool, connector and plugin approvals pending: [ ]
Incidents this period: [None / details]

7. Adoption

Taken from your AI tools’ admin analytics. Adoption shows reach, not value.

Team	Licensed seats	Active seats	Trend
			↑ / → / ↓

8. Next month

Three to five things that will be true by the next update.

9. My view

Three lines of your own judgement. Only you can write this part. It’s why the update carries your name.

Glossary

Opportunity: A potential AI use case or process improvement that has been captured but not yet built. Opportunities move from Captured (identified) to Shortlisted (scored and prioritised) to Promoted (approved for delivery). They can also be Parked (on hold) or Rejected.

ICE score: Impact × Confidence × Ease. Each factor is scored 1-10, giving a composite score out of 1000. Impact is the business value, Confidence is how sure you are it will work, Ease is how quickly you can deliver it. Used to rank opportunities objectively.

Deliverable: An approved piece of work (an agent, workflow, process redesign or rollout) that is being built or has been completed. Deliverables move from In progress to Done.

Outcome: A measured change in a business process that has a baseline, a current value, and verification from the business owner. Outcomes are what the programme is judged on.

Process: An end-to-end business workflow (e.g. lead triage, month-end close, claims handling) that an AI deliverable redesigns or improves.

Template by Kowalah, UK Anthropic Implementation Specialist. kowalah.com

Get started

How to use this template

01

Copy the template, or use the skill

Use the 'Copy page' button and fill in each section, or download the Claude skill and build the update with Claude from your notes, a tracker or the Kowalah platform (when connected)

02

Fill it in the week before the steering group

Lead with what changed in the business (Outcomes section), not with activity. Every number should say how it was measured and whether someone outside the AI team has verified it

03

Keep it to one page

Prune detail so the whole update prints on one page. Length is not credibility. Attach a pipeline tracker or risk register separately if needed

04

Report stalls before wins

List stalled items and what unblocks them (Stalled items and asks section) before celebrating delivery. An honest update builds more credibility than an inflated one

Questions

Frequently asked questions

Common questions about steering group updates

What should an AI steering group update include?

Outcomes with baselines first (what changed in the business, with outcome type and period), then the opportunities and deliverables pipeline (Captured/Shortlisted/Promoted for opportunities, In progress/Done for deliverables, plus Parked/Rejected counts), top opportunities by ICE score, stalled items and asks (with what unblocks them, who can, and by when), decisions needed, risks and security (the top three risks with owner and mitigation, plus approvals pending and incidents), adoption (active versus licensed seats, with the trend), next month (the three to five things that will be true by the next update), and your own view (three lines of your judgement, which only you can write).

How long should it be, and how often?

One page. Monthly by default; fortnightly in early build phases or when decisions are blocking. Keep the same structure; shorten 'Next month' to 'Next two weeks' and update outcomes only when there's a new measurement.

How do I report AI ROI when the results aren't in yet?

Report the opportunities and deliverables pipeline as leading indicators, label estimates as estimates, and only move a process to Outcomes once it has a baseline. Show the pipeline count by stage (Captured/Shortlisted/Promoted for opportunities, In progress/Done for deliverables, plus Parked/Rejected counts) and the change since last period. The Top opportunities by ICE score section lists the highest-scoring opportunities with their impact rationale.

How do I report stalled or failed use cases?

List them in the Stalled items and asks section before wins, with why they're stalled, what unblocks them, who can, and by when. Anything stopped goes in with what was learned. Stopping is a result, and an honest update builds more credibility than an inflated one. Parked and Rejected opportunities appear as counts in the pipeline.

What's the difference between adoption and value?

Adoption is reach: active versus licensed seats by team, and the trend (up, flat, down). It shows who has arrived, not what they've achieved. Value is a measured change in a business process (Outcomes section), with a baseline, a target, and verification. A release can have high adoption and no measured value yet, or vice versa.

Who should verify the outcomes?

The accountable business owner of the process, not the AI team. If a lead triage agent claims to have cut response time from a day to minutes, the head of sales verifies it, because they own the domain and will carry the result on their scorecard. Verification moves a claimed benefit from aspiration to evidence.

Can Claude write my steering group update?

Yes: the free ai-steering-group-update skill (download it above) drafts the update from your notes, a tracker, or Kowalah platform data when connected. It reads opportunity status, ICE scores and outcome data directly. It asks before inventing anything, leaves OPEN placeholders for missing baselines, keeps claimed and verified apart, lists stalled items before wins, and never fills in 'My view' (the final section), because that's the part only you can write.

What should the steering group decide each month?

Whatever is in the Decisions needed section: unblocks, scale/redirect/stop calls for live releases, promoting opportunities into deliverables, and approvals. The section should list the decision, the options, your recommendation, what happens if it's deferred, and the date it's needed by. If nothing needs a decision, say so and move on.

More templates

Related templates.

Ready when you are

Draft your update from your live programme, free

The Kowalah Platform logs your use cases, processes and outcomes, and the in-chat programme review drafts your steering update from live data. Free to start, and no share link or automatic adoption data until you're ready.

Start for free

The One-Page AI Steering Group Update