It seems to be progressing in the Sales Department, but what exactly should we be deciding as a company?
This was a line in a Slack message sent to members of the Corporate Planning team after an executive meeting by Mr Sato (a pseudonym), who is responsible for driving digital transformation at Manufacturing Company A. Six months earlier, the company’s Sales, Administrative and Information Systems departments had each been experimenting with generative AI. However, the state of AI adoption had not become a formal agenda item at executive meetings, and both results and issues remained confined within individual departments. Operational managers felt that AI was useful but had not been standardised, while senior management believed there was insufficient evidence on which to base further investment decisions.
The company now holds a monthly AI review and uses a single-page dashboard to examine usage, reductions in operational workload, risks and priorities for the following month. After reviewing five departments and 18 use cases over the past three months, the meeting can now classify each initiative as one to continue, stop or investigate further.
This article is intended for corporate planning teams, digital transformation leaders and departmental managers seeking to scale AI across their organisations. It explains the agenda for an AI governance meeting, how to present operational AI in management reports and how to design reviews that lead to investment decisions. The goal is to create an environment in which management and frontline teams can improve departmental experiments while looking at the same information. However, merely setting up a monthly review meeting will not embed AI adoption. A repeatable operating model only begins to emerge when the meeting is combined with named owners, frontline training, data management and a mechanism for consolidating usage information. If your organisation faces similar challenges, use this article as a reference and adapt the ideas to your own governance structure.
The further AI adoption progresses by department, the harder company-wide decision-making becomes
In the early stages of introducing generative AI, it is effective for each department to start with small experiments. Suitable applications gradually become clear: the Sales Department may use AI to produce first drafts of proposals, Marketing to create article structures and advertising copy, Administrative teams to summarise meeting minutes and internal FAQs, and the Information Systems Department to improve the efficiency of help-desk support.
However, as the number of experiments grows, a different set of problems arises from a company-wide perspective.
- Which initiatives are producing results?
- Which departments should receive additional investment?
- Who is overseeing information-management risks and incorrect answers?
- Are several departments duplicating similar trials?
If these questions cannot be answered, AI remains a convenient tool for frontline staff and is unlikely to lead to business transformation at management level.
To scale AI across the organisation, a forum is needed that converts small frontline results and issues into management decisions. The monthly AI review meeting fulfils this role.
What is a monthly AI review meeting?
A monthly AI review meeting brings together information on how each department is using AI once a month and reviews results, issues, risks and priorities for the following month.
It is not merely a reporting session. Its purpose is to organise AI adoption not around whether people have used AI, but as information that can support operational improvement and investment decisions.
For example, the meeting may make the following decisions:
- Roll out successful use cases to other departments.
- Add guidelines for activities that are widely used but carry high risks.
- Pause initiatives whose benefits are not apparent.
- Identify departments that need additional training data or staff development.
- Organise the key issues that should be escalated to the executive meeting.
The monthly AI review is an intermediate governance forum connecting frontline usage with executive-level AI decision-making. Departmental meetings tend to optimise locally, while executive meetings may find individual use cases too detailed. Placing a company-wide review meeting between the two makes it easier to run a continuous improvement cycle for AI adoption.
Public guidelines also emphasise the importance of senior-management leadership, risk management and embedding AI governance within the organisational structure.
Topics to cover in a monthly review meeting
A monthly AI review can become a box-ticking exercise if it covers too many topics. At first, it is easier to operate if the meeting focuses on five areas: usage, results, issues, risks and priorities for the following month.
Usage
The first point to confirm is which departments are using AI, for which tasks and to what extent.
Example items include the number of participating departments, total users, active users, covered processes, frequency of use, principal purposes, new use cases and the number of suspended or dormant use cases.
It is important not to treat the number of uses alone as a result. Even if the number of AI chat sessions is increasing, this is weak evidence for company-wide expansion unless it is reducing working time or improving quality. Conversely, a use case with relatively few interactions may still contribute significantly to standardising a critical process.
Usage should therefore be viewed only as the starting point for assessing outcomes.
Results
Next, confirm what has changed as a result of using AI.
Outcome measures differ by process. For meeting minutes, possible measures include drafting time, number of revisions and time to distribution. For internal enquiries, they may include the first-response rate, staff handling time and number of unresolved cases. For sales materials, useful measures include time to first draft, proposal preparation time and the number of review comments.
When presenting figures, clearly state the period, scope and conditions.
For example, rather than writing “Meeting-minute preparation time was reduced”, it is more useful for management reporting to write: “Across 12 regular Sales meetings held in April 2026, the average time required to produce a first draft of the minutes fell from 35 minutes to 12 minutes.”
Not every result can be quantified. In such cases, record comments from staff and changes in behaviour.
- Members of the team can now produce weekly reports in a consistent format that previously only the department head could create.
- Variation between staff members’ responses to enquiries has decreased.
- Key issues are organised before meetings, allowing meeting time to be used for decision-making.
These qualitative changes are also important evidence in a monthly review.
Issues
Issues should always be reviewed alongside successful results. If the meeting presents only positive examples, it will quickly become a reporting session. It is more important to identify underperforming use cases early.
Common issues include:
- Only a small number of knowledgeable employees use the tool.
- Prompts depend on individual expertise.
- Output quality is inconsistent.
- Frontline teams do not fully trust AI-generated answers.
- Training data is out of date.
- Usage rules are unclear, leaving staff unsure what information they may enter.
- No outcome-measurement criteria have been agreed.
- Several departments are duplicating similar initiatives.
It is important not to dismiss these issues as a lack of individual skill. If a department is not using AI effectively, the cause may not be poor personal literacy but the absence of suitable templates or clearly defined usage scenarios.
In the monthly review, organise issues around operations, data, training and allocation of responsibility rather than around individual people.
Risks
Risk review is essential for an AI governance meeting. As adoption expands across the organisation, risks expand with it.
The four principal areas to review are shown below.
| Risk area | What to review |
|---|---|
| Information management | Whether personal data, confidential information or customer information has been entered inappropriately |
| Answer quality | Whether incorrect, outdated or unsupported answers are being used without verification |
| Access control | Whether more people than necessary can access highly confidential information |
| Operational accountability | Whether the final reviewer, approver and accountable user of AI output are clearly identified |
AI risk management will not become embedded simply by adding more prohibitions. What matters is clearly defining which information may be handled, in which processes and to what extent.
For example, when using an operational AI platform that can separate users, data and applications by project, projects can be designed by department or process so that only the necessary people can handle the relevant data. Kanata can also organise AI chat, AI summarisation and e-learning by project, making it one option when operations need to be separated by department or purpose. However, an organisation’s existing groupware, knowledge-management tools or BI tools may already provide sufficient control. The important point is not the name of the tool, but whether the scope of responsibility and the location of data are clear.
Priorities for the following month
Finally, decide the priorities for the following month. A common failure in AI meetings is to stop at reporting.
- It is being used in Sales.
- Administrative teams are also trialling it.
- The Information Systems Department is testing it.
Statements such as these do not lead to any change after the meeting.
At the end of the monthly review, classifying each use case into the following four categories makes decision-making easier.
| Category | Decision |
|---|---|
| Continue | Maintain current operations and continue tracking results |
| Improve | Review prompts, training data, education, access design or other elements |
| Scale | Roll the use case out to other departments |
| Stop | Pause the initiative because its benefits are limited or its risks are high |
When scaling across the organisation, expanding every initiative is not necessarily the right answer. Being able to stop an initiative quickly is also important for improving the quality of investment decisions.
How to choose participants for the monthly review meeting
If too many people attend a monthly AI review, it becomes a reporting session; if too few attend, decisions cannot be made.
The following basic structure provides a useful starting point.
| Role | Typical responsibility |
|---|---|
| Meeting owner | Corporate Planning or the Digital Transformation team |
| Company-wide programme owner | CDO, Head of Digital Transformation, Head of Corporate Planning or equivalent |
| Technology and security owner | Information Systems or Information Security |
| Departmental managers | Sales, Marketing, Administration, Customer Support and other departments |
| Use-case owners | Frontline leaders responsible for each process |
| Secretariat | Digital Transformation, Corporate Planning or project-management staff |
The key is to separate the people who report from the people who decide.
Use-case owners report what is happening at operational level. Departmental managers decide whether an initiative should continue, expand or stop within their department. The company-wide programme owner and Corporate Planning set cross-departmental priorities and make investment decisions.
Not every stakeholder needs to attend every meeting. A regular meeting may involve Digital Transformation, Corporate Planning, the main departmental managers and Information Systems. Legal or security specialists can attend only when the agenda includes information management, contracts or personal data.
Dividing attendees into permanent members and ad hoc participants according to the meeting’s purpose makes the forum easier to sustain.
A standard agenda for the monthly AI review
A practical length for a monthly review meeting is 60 to 90 minutes. Starting with a long meeting increases preparation effort and makes the process harder to sustain. Begin with 60 minutes and extend it to 90 minutes only when the amount of discussion increases.
| Time | Agenda item | Content |
|---|---|---|
| 5 minutes | Review previous decisions | Confirm whether actions from the previous meeting have been completed |
| 10 minutes | Review the company-wide dashboard | Check usage, covered processes and key measures |
| 15 minutes | Review priority use cases | Examine use cases producing strong results and those facing major issues |
| 10 minutes | Risk and governance review | Check information management, answer quality, permissions and any rule breaches |
| 10 minutes | Investment decisions and prioritisation | Decide whether to continue, improve, scale or stop each initiative |
| 10 minutes | Confirm next month’s actions | Agree owners, deadlines and reporting methods |
The important feature of this agenda is that it establishes the overall picture first and then examines only priority use cases in depth. There is not enough time to give every use case equal attention. The monthly review should provide a broad view while concentrating discussion on matters requiring a decision.
What to include in the dashboard
Using a one-page dashboard as the basis of the monthly AI review helps discussion move forward.
The dashboard should not contain every detailed log. It should allow participants to understand the current position and required decisions quickly.
Company-wide summary
First, show the overall state of AI adoption across the organisation.
| Item | Description |
|---|---|
| Number of departments in scope | Departments actively pursuing AI adoption |
| Number of use cases in scope | Operational AI use cases under management |
| Active users | People who used the service during the period |
| New use cases | Initiatives added during the month |
| Candidates for scaling | Initiatives that may be suitable for other departments |
| Candidates for suspension | Initiatives being considered for suspension because of limited benefits or risks |
It is useful to design the company-wide summary so that it can also be reused in executive reporting.
Use-case list
Next, list the status of each use case.
| Item | Example entry |
|---|---|
| Use-case name | Drafting sales proposals |
| Department | Sales |
| Owner | Sales Planning Manager |
| Objective | Reduce first-draft preparation time |
| Status | Testing, live, improving or candidate for suspension |
| Outcome measures | First-draft time and number of reviews |
| This month’s result | Average preparation time reduced from 45 minutes to 20 minutes |
| Issue | Adjusting wording for individual customers remains time-consuming |
| Next action | Develop industry-specific prompts |
| Decision | Continue with improvements |
Risk register
Manage risks separately from results.
| Risk item | What to check | Example response |
|---|---|---|
| Entry of personal data | Whether data containing personal information has been entered | Communicate masking rules |
| Confidential information | Whether customer or contractual information is being handled appropriately | Review project permissions |
| Incorrect answers | Whether AI output is being used without human review | Assign a reviewer |
| Outdated data | Whether training data has been updated | Review the library monthly |
| Dependence on individuals | Whether only particular employees can use the system | Share and standardise prompts |
This risk register is important if the meeting is to fulfil its role as an AI governance forum. NIST’s AI Risk Management Framework also sets out an organisational approach to managing AI risk.
How to turn the review into an operational AI management report
The information organised in the monthly review may be too detailed to present directly at an executive meeting. For senior management, convert frontline logs into the information required for decisions.
An operational AI management report is easier to understand when it is limited to the following four points.
This month’s company-wide position
First, show how far AI adoption has progressed across the organisation.
As at April 2026, five departments and 18 use cases are included in the monthly review. Of these, nine are live, six are being tested and three are candidates for suspension or redesign.
State the period, number of departments, number of use cases and status clearly.
Main results
Next, present the initiatives that are producing results.
Across 20 proposals prepared by the Sales Department in April 2026, average first-draft preparation time fell from 45 minutes to 22 minutes. However, human review is still required for customer-specific customisation.
Include limitations when presenting results. A report containing only positive figures can create excessive expectations. For management decisions, it is necessary to clarify what AI can handle and where human judgement remains necessary.
This month’s issues and risks
Report issues as well as results.
The Administrative Department’s internal FAQ initiative shows potential to reduce enquiries. However, policy-data update dates differ by department, creating a risk that outdated information may be referenced. Next month, we will establish rules for updating training data.
Reporting issues in this way makes it easier to explain the need for additional investment or organisational measures.
Decisions required from management
Finally, make the matters requiring management decisions explicit.
- May the Sales proposal-drafting AI be rolled out to Marketing?
- May an additional 20 staff-hours per month be allocated to improving the Administrative Department’s FAQ data?
- May company-wide training on the AI usage guidelines become mandatory from next month?
- May two use cases with no visible benefits be stopped?
An executive AI meeting should request decisions rather than merely share information.
The information platform needed to operate monthly reviews
To sustain monthly reviews of AI adoption, an organisation needs a place to gather the relevant information.
It is possible to begin with Excel or a spreadsheet. However, as the number of participating departments and use cases grows, prompts, training data, chat histories, summarised outputs and training content tend to become fragmented.
At this stage, consider the following criteria when selecting an information platform.
- Can access permissions be separated by department or process?
- Can prompts and training data be reused?
- Is it easy to review usage of AI chat, summarisation, training and related functions?
- Is the platform easy for frontline employees to use in their daily work?
- Can administrators easily review outdated data and permissions?
Kanata, provided by our company, brings AI chat, AI summarisation and e-learning together in one environment and allows users, data and applications to be organised by project. For example, separate projects can be created for Sales, Administration, Marketing and Information Systems, with AI chat or summarisation functions placed within each business process.
On the other hand, companies that already have an internal portal, BI tools, knowledge-management tools or ticket-management tools may achieve better adoption by using those existing systems. The purpose of the monthly review is not to introduce a particular tool, but to keep results, issues, risks and required decisions visible over time.
How to prevent the monthly review meeting from becoming a formality
A monthly AI review will not become embedded simply because it has been created. Several points require attention if it is to continue effectively.
Do not turn it into a reporting session
The most common failure is to finish after hearing reports from each department. Reporting is necessary, but the purpose of the meeting is decision-making.
For every use case, always make one of the following decisions at the end:
- Continue
- Improve
- Scale to other departments
- Stop
- Carry out additional validation by the following month
If a meeting produces no decisions, participants will give it lower priority each time it is held.
Do not discuss only success stories
When AI is scaled across an organisation, there is often a strong emphasis on sharing success stories. Sharing what worked is valuable, but success stories alone do not reveal risks or opportunities for improvement.
The monthly review should also cover examples such as:
- Use cases where adoption did not increase
- Use cases that frontline staff found difficult to use
- Cases in which incorrect answers occurred
- Cases that saved less time than expected
- Cases where data preparation was insufficient
Do not blame failed initiatives. Treat them as opportunities to learn before wider deployment.
Do not use too many measures
When attempting to make AI adoption visible, it is tempting to add more measures. However, too many measures increase the time needed for data entry and aggregation, placing a greater burden on frontline teams.
At first, the following three types are sufficient.
- Usage measures
- Who used AI, for which process and how much
- Outcome measures
- How time, quality, volume, satisfaction or other results changed
- Risk measures
- Whether there were problems involving incorrect answers, information management, permissions or unreviewed use
Add more detailed measures only for priority use cases.
Share post-meeting actions within 48 hours
The impact of a monthly review is determined after the meeting, not during it.
As a guide, share the decisions, use cases to continue, improve, scale or stop, owners, deadlines and points to confirm before the next meeting within 48 hours.
Sharing only with meeting participants is not enough. Relevant information should also be communicated to the employees using AI in their daily work.
- Why is this initiative continuing?
- Why is this usage being stopped?
- What will the organisation focus on next month?
When these points are understood, AI adoption is more likely to progress with the support of frontline teams.
Monthly AI review meeting templates
Finally, the following templates can be used in the meeting. Adjust them to suit your meeting length and management fields.
Meeting overview template
Monthly AI Review Meeting
Reporting month:
Meeting date:
Participants:
Meeting owner:
Minute-taker:
Purpose:
Review company-wide AI adoption and decide on results, issues, risks and priorities for the following month.
Decisions required today:
1.
2.
3.Use-case report template
Use-case name:
Department:
Owner:
Status: Testing / Live / Improving / Candidate for suspension
Objective:
Process to be improved through this AI use case:
Usage this month:
Period:
Number of users:
Number of uses:
Number of processes or cases covered:
Results:
Quantitative results:
Qualitative results:
Issues:
1.
2.
3.
Risks:
Information management:
Answer quality:
Access control:
Operational accountability:
Actions for next month:
Owner:
Deadline:
Support required:
Decision:
Continue / Improve / Scale / Stop / Additional validationManagement report template
Operational AI Management Report
Reporting period:
Departments in scope:
Number of use cases in scope:
Company-wide summary:
- Live:
- Testing:
- Improving:
- Candidates for suspension:
Main results:
-
-
-
Main issues:
-
-
-
Risk and governance matters:
-
-
-
Matters requiring management decisions:
1.
2.
3.
Priority themes for next month:
-
-
-Summary: scaling AI requires a decision-making system, not merely another meeting
There is nothing inherently wrong with AI adoption progressing independently within each department. In the early stages, experimentation by frontline teams is necessary. Sales, Marketing, Administrative teams and Information Systems differ in the work they want AI to perform, the way they assess results and the types of risk they face.
However, if left as it is, adoption remains locally optimised. To scale AI across the organisation, departmental initiatives must be consolidated so that results, issues, risks and investment decisions can be considered in the same forum.
The monthly AI review meeting provides this mechanism.
The important point is not to create more meetings. It is to create a state in which management and frontline teams look at the same dashboard and can decide whether to continue, improve, scale or stop each initiative.
This requires the agenda, assignment of owners, dashboard, operational AI management report, risk review and post-meeting actions to be designed as one operating model.
However, a monthly review meeting is not a universal solution. Creating a governance forum alone will not embed AI if frontline usage scenarios remain unclear. AI becomes an organisational capability only when management’s decision criteria, the technology function’s governance and the business departments’ accountability for results are aligned.
Q&A: Frequently asked questions about monthly AI review meetings
Should the monthly AI review meeting be separate from the executive meeting?
In many cases, a separate meeting is easier to operate. Individual use cases are often too detailed for an executive meeting. A practical approach is to organise results, issues and risks in the monthly review and escalate only the matters requiring executive decisions.
Should every company-wide use case be managed from the outset?
Trying to manage everything from the beginning creates a heavy data-entry and aggregation burden. Start with processes that are used frequently, carry high risks or have a significant management impact. More use cases can be added after the governance forum has become established.
Which outcome measures should be used?
Possible measures include working time, volume processed, quality, number of rework cycles, number of enquiries and qualitative comments from users. The important point is to make comparison possible before and after AI is introduced. When using figures, always include the period, sample size and measurement method.
What are the minimum risks that an AI governance meeting should review?
At a minimum, review information management, answer quality, access control and operational accountability. In particular, regularly check whether personal or customer information has been entered, whether AI output has been submitted externally without review and whether unnecessary personnel can access highly confidential data.
What should be reviewed if the monthly meeting is becoming a formality?
First, check whether decisions are being made at the end of the meeting. If the meeting finishes with reporting alone, change the process so that every use case receives one of five decisions: continue, improve, scale, stop or conduct additional validation. Next, review whether decisions and owners are being shared within 48 hours, as this makes execution more likely.
