Operating Model Rituals
Rituals are recurring management events through which the AI function synchronizes goals, initiatives, products, resources, risks, and impact.
Their job is not to add bureaucracy, but to ensure managed movement of the AI agenda: from strategy and business goals to specific initiatives, pilots, adoptions, and confirmed impact.
Without rituals, AI adoption quickly becomes scattered pilots: RAG here, an LLM assistant there, a dev agent elsewhere — but the company does not know what actually helps, where bottlenecks are, or which decisions to scale.
Why rituals are needed
Rituals regularly answer five management questions:
- What is priority now? — which business goals, directions, and initiatives need the AI function's attention.
- Which AI initiatives are moving and which are stuck? — where help, decisions, resources, or escalation are needed.
- Which AI products to grow, pilot, or stop? — which platform and application products the business actually needs.
- Where are the risks? — data, security, architecture, operations, quality, legal constraints, impact.
- How much value are we really creating with AI? — which initiatives delivered impact, which only promise it, and which should be closed.
Map of AI function rituals
1. AI strategy committee
Goal: link the AI agenda to company goals and make key decisions on priorities, resources, and scaling.
Participants: director of the AI function; heads of key business lines; head of product portfolio; head of project office; head of infrastructure; representatives of IT, security, architecture, data, finance, and other adjacent functions.
Cadence: monthly or quarterly.
Topics: AI adoption strategy goals; key initiative portfolio; status of major pilots and adoptions; confirmed and expected impact; major blockers; scaling, stop, or resource reallocation decisions; current processes and needed improvements; need for new AI products or infrastructure capabilities.
Outputs: approved priorities; resource decisions; scaling decisions; escalation list; updated AI function focus.
2. AI initiative portfolio committee
Goal: manage the business funnel of AI ideas and initiatives — from incoming requests to completion, rejection, or handover to operations.
Participants: head of project office; project / initiative leads; business representatives; head of product portfolio; head of infrastructure when needed; data, IT, security, and architecture experts when needed.
Cadence: weekly or biweekly.
Topics: new AI ideas; initiatives in evaluation; initiatives in delivery; initiatives awaiting impact; stalled initiatives; closure candidates; resource conflicts; routing to a specific AI product.
Outputs: decision to evaluate, defer, or reject an idea; initiative owner assignment; next step; business funnel status update; blockers and owners to remove them.
3. AI product council
Goal: manage the AI product portfolio — which products the company needs, how they evolve, which pilots run, and how products are reused in initiatives.
Participants: head of product portfolio; AI product owners; head of infrastructure; head of project office; user representatives when needed; IT, security, architecture, and data when needed.
Cadence: biweekly or monthly.
Topics: product status; new business needs; pilot results; user feedback; capability gaps; product roadmaps; improvement prioritization; products as standard functional blocks for initiatives.
Outputs: updated product roadmap; pilot decisions; scaling, stop, or replacement decisions; improvement list; clarity on products available for new initiatives.
4. Delivery sync
Goal: ensure regular movement of initiatives already in implementation.
Participants: head of project office; initiative leads; product owners of involved AI products; technical executors; business representatives; IT / data / security when needed.
Cadence: weekly.
Topics: delivery status; detailing, prototype, pilot, adoption stages; blockers; dependencies on IT, data, security, architecture; readiness for the next stage gate; timeline or impact risks.
Outputs: updated statuses; blocker list; movement decisions; stage gate preparation; initiatives needing escalation.
5. Stage gate review
Goal: decide whether an initiative or product moves to the next lifecycle stage.
This is a control point, not a status meeting: the team must prove sufficient readiness to move forward.
Participants: initiative owner; project lead; business representative; AI product lead; head of project office and infrastructure when needed; IT / security / architecture / data when needed.
Cadence: event-driven, when ready to transition.
Checks: business problem clarity; impact owner; AI product or delivery track; technical feasibility; data and integrations; risks; pilot or adoption plan; success criteria; impact measurement approach.
Possible decisions: move forward; return for refinement; change delivery track or product; freeze; reject; hand over to operations.
6. Impact confirmation ritual
Goal: separate real results from declarations and identify which initiatives truly delivered benefit.
Participants: business impact owner; initiative lead; head of project office; finance representative when needed; AI product lead; analysts or metric owners.
Cadence: after pilot completion or an impact observation period.
Outputs: confirmed or unconfirmed impact; scale, refine, or close decision; impact registry update.
7. Operations and support ritual
Goal: ensure adopted AI solutions are not orphaned after pilot or launch.
Participants: AI product lead; business process owner; operations team; head of infrastructure; IT; security and user support when needed.
Cadence: monthly or per SLA.
Outputs: support decision; improvement list; incident log; SLA / process updates; assessment of ongoing usefulness.
8. AI idea mining ritual
Goal: regularly find new AI application opportunities in business departments.
Participants: head of AI training and promotion; AI champions; business department representatives; head of project office; head of product portfolio when needed.
Cadence: monthly or in waves within departments.
Outputs: new AI idea list; initial potential assessment; business funnel candidates; understanding of useful AI products for the department.
9. AI culture training and adoption ritual
Goal: raise employee maturity and help the business apply AI deliberately, not just "order AI."
Formats: product demos; LLM training; task framing for AI; idea workshops; case reviews; open office hours; champion training.
Outputs: higher awareness; better AI ideas; fewer chaotic requests; more reuse of existing products; AI champion network.
10. Internal AI function ritual
Goal: synchronize the AI function team — products, initiatives, infrastructure, training, and management.
Cadence: weekly.
Outputs: shared position; weekly priorities; management decisions; preparation for external rituals; less internal chaos.
Minimum ritual set
Early on, five rituals are enough:
| Ritual | Purpose |
|---|---|
| AI strategy committee | Link AI to company goals |
| Initiative portfolio committee | Manage incoming ideas and priorities |
| Product council | Manage AI products and their evolution |
| Delivery sync | Move initiatives in implementation |
| Stage gate review | Control stage transitions |
Other rituals are added as maturity grows.
How rituals connect
- Strategy committee sets goals and priorities.
- Idea mining finds new opportunities.
- Portfolio committee turns ideas into managed initiatives.
- Product council ensures the right AI products exist.
- Delivery sync moves initiatives through implementation.
- Stage gate review controls transitions.
- Impact confirmation verifies value was created.
- Operations ritual keeps solutions running.
- Training and champions expand AI application capability.
- Internal ritual ties it all into one managed system.
Ritual design principle
Core principle: every ritual makes decisions, not just collects statuses.
Bad ritual: "Let's each share what we're doing."
Good ritual: "Which initiatives move forward, which stall, which close, where do we need resources, where is there risk, and where is impact confirmed?"
Typical mistakes
- Everything on one big meeting — separate strategy, initiative portfolio, product portfolio, delivery, impact, operations.
- Rituals without decisions — every meeting needs a concrete output.
- No decision owners — each ritual needs: owner, participants, inputs, outputs, decision types.
- No link between products and initiatives — initiatives show demand, products create reusable capability.
- Impact discussed too late — record the impact hypothesis at evaluation, revisit at stage gate and after pilot.
Summary
Operating model rituals are the management frame of the AI function.
They let the company manage the full AI agenda as a system: from goals and ideas to products, implementation, operations, and confirmed value.
The main outcome is transparency: the company knows which initiatives exist, why they matter, which products implement them, where they stand, who owns them, which decisions are needed, and which impact is confirmed.