A guide for executives, department heads, and people managers
AI for people leaders
The three Cs of AI adoption: compliance, competency, and cost. Set the expectations, help people learn, and fund the tools and support they need.
Give your team a workable way to use AI.
People may already be using AI to draft messages, summarize documents, analyze information, or write code. A leader needs to know which accounts they use, what information they share, and whether they can check the results. Those answers shape the policy, the training, and the purchasing decision.
RUDI organizes this work around three Cs: Compliance → Competency → Cost. First, establish the conditions for responsible use. Then build the skills to work within them. Use that understanding to choose and fund the right environment. Set a budget for the initial work, and revisit all three as you learn from a pilot.
01 · Compliance
Define how AI may be used.
Your AI policy needs to work at two levels. At the organization level, it sets responsibilities and requirements. At the employee level, it explains what someone should do when they open a tool or connect it to workplace information.
- Macro: the organization’s policy
- Identify who approves tools and use cases, which data may be processed, what contracts and controls are required, and who handles exceptions or incidents. Assign responsibility for access, retention, review, and changes to the system.
- Micro: the employee’s instructions
- Tell people which account to use, what information they may upload, which connectors are approved, what needs a human check, and how to disclose AI assistance. Show them where to get help when the policy does not answer their question.
Make the policy usable in a real task.
For a team preparing a client update, guidance could identify the approved workspace, the project folder it may access, the information to leave out, and the person who reviews the draft before it is sent. The employee should be able to follow that example without interpreting a procurement contract.
Build a shared resource with the policy, approved tools, examples, and a contact for questions. Keep it current when permissions, products, or workflows change. RUDI’s RESPECT framework helps teams check responsible ownership, equitable access, safety, privacy, effectiveness, control, and transparency.
Choose the account and controls that fit the work.
Review the actual plan, contract, settings, and connected services with your IT, security, and policy owners. Document requirements for identity and access, data handling, retention, audit records, and support. Compare those requirements with the available business, team, or enterprise offering.
Product names alone cannot establish whether a workflow meets your obligations. Enterprise controls still need configuration and an owner. Treat API access as another implementation to evaluate: its permissions, retention settings, application, and contractual terms all matter.
Leave this stage with: an approved-use policy, employee guidance, and a named person responsible for each AI workflow.
02 · Competency
Help people use AI within those expectations.
A policy becomes useful when employees can apply it. Training should help people choose appropriate work, provide context, check an answer, protect information, and recognize when a person needs to take over.
Start by asking how people already use AI and where they need help. Combine that baseline with a short, practical exercise using approved or synthetic material. Confidence and tool access give you part of the picture; observing someone complete and review a task shows what support they need.
- Understand the tools
- Explain what the approved systems can do, their limitations, and the difference between an employee assistant and a digital worker carrying out a defined task.
- Practice the work
- Use examples from the team’s responsibilities. Teach people to give context, evaluate evidence, correct mistakes, and decide whether the output is useful.
- Apply responsible-use rules
- Practice choosing permitted information, checking access, recognizing when approval is needed, and reporting an unexpected result or disclosure.
Build support around different starting points.
Give employees time to practice, accessible materials, and a way to ask questions. Some will need introductory sessions; others will need guidance on connectors, coding tools, or agent workflows. Involve both experienced users and people who need more support in the pilot so you can see how the process works across the team.
A shared library can hold reviewed examples, reusable instructions, short demonstrations, and common corrections. Record what a good result looks like and when human review is required. Update those resources as the team learns.
For a business unit, pair training with mapping one recurring workflow. Identify where AI assistance, an agent, an automation, or a small application would help. Test the change with the people who will use it and compare quality and total effort with the current process.
Leave this stage with: evidence of what people can do, a training and support plan, and a tested workflow with review responsibilities.
03 · Cost
Fund a compliant, competent workforce.
With requirements and training needs in view, ask what tools, access, and support each group needs. Start with the licenses you already hold. Identify any missing capability before adding another subscription.
Consider the workplace suite and department needs together.
For an organization using Microsoft 365 or Google Workspace, RUDI recommends evaluating the AI available in that environment first. Copilot and Gemini can bring AI into the places where employees already handle documents, messages, and meetings. The exact features and access depend on the edition, license, and configuration.
A department may also have a clear reason to use ChatGPT, Codex, Claude, or Claude Code: a development workflow, a particular document task, or an application built on a provider’s API. Give that additional tool a defined job, owner, and budget. ChatGPT and Claude also offer organization-wide plans; a suite-plus-department arrangement is one way to organize adoption, rather than a limit on what each vendor can support.
Scroll the table sideways to compare platforms and licensing.
| Platform | Where to evaluate it | Licensing and budget considerations |
|---|---|---|
| Microsoft Copilot | Work in the Microsoft 365 environment. Check the features available across the applications your people use. | Separate the Copilot Chat access available to your organization from paid Microsoft 365 Copilot licenses, suite bundles, and metered agent services. Confirm the underlying Microsoft 365 requirements. Microsoft plans and pricing. |
| Google Gemini | Work in Google Workspace. Gemini availability in Gmail, Docs, Drive, and other apps varies by edition. | Workspace includes AI features with different access across plans. Compare the full suite cost and any additional AI or cloud services. This is a different purchase from a personal Google AI plan. Google Workspace plans. |
| OpenAI ChatGPT & Codex | Employee assistance, department workflows, and coding work. Business and Enterprise support managed organizational use. | ChatGPT Business lists standard and premium seats. Enterprise pricing is negotiated and can include credit- or token-based arrangements. The API has separate billing. OpenAI business plans; billing details. |
| Anthropic Claude & Claude Code | Employee assistance, document and department workflows, and coding work. Compare Team and Enterprise controls. | Claude Team lists standard and premium seats. The published Enterprise offering combines a seat fee with usage charged at API rates. Confirm the terms of the offer you are buying. Claude plans and pricing. |
Distinguish seats from metered usage.
A seat subscription buys a person access to specified applications and allowances. As checked on September 22, 2026, both ChatGPT Business and Claude Team list standard seats at $20 per user per month and premium seats at $100 when billed annually. Monthly billing is $25 and $125 respectively. These are published US-dollar prices before tax; check current prices, limits, seat minimums, and contract terms when purchasing.
Microsoft and Google package their workplace offerings differently. Use their current plan pages to compare what is already included, what is an add-on, and what must be bought as a separate service.
For API-based applications, charges commonly depend on the model and the input and output tokens processed. A token is a unit used to measure model input or output. Additional charges can apply for tools, search, storage, or other services. Compare the relevant Claude API and Gemini API rate cards for the features your application uses.
Other systems meter work in credits or actions. Copilot Studio, for example, offers Copilot Credits and pay-as-you-go billing. The budget should reflect the actual billing unit, included allowance, and any overage rules for each service.
Budget for the work around the software.
Include training, workspace administration, connector setup, maintenance, and the time people spend reviewing results. Track usage and spend by team or workflow, set alerts or limits where the platform supports them, and remove unused access as people change roles.
Review whether the tool improves the task after checking and rework are counted. A cheaper seat that creates more manual repair may cost the team more overall. A higher-priced tool needs evidence that its additional capability is useful for that job.
Leave this stage with: a license plan, a metered-usage budget, and funded ownership for training, administration, and review.
Keep managing the AI environment.
AI adoption creates ongoing work. Someone needs to manage licenses, onboard and remove users, approve connectors, and review changes to the tools. A connector can make an assistant more useful by giving it relevant information; it also creates an access decision that needs an owner.
Before enabling one, identify the source, the people who may use it, and the actions it may take. Test access with representative roles, including information a user should not be able to reach. Review permissions again when employees move teams or the source system changes. Microsoft’s connector documentation explains the permission model for its supported connector types.
Every digital worker also needs a human manager who can inspect results, change its instructions, and pause its work. Make the route for questions and corrections visible to employees and anyone affected by the output.
RUDI helps organizations put these pieces together through AI workspace setup and policy support, team training, and agents and applications built around real work.
Start with one team and one workflow.
- Confirm the rules. Name the policy owner, inventory the accounts in use, and publish guidance for the selected task.
- Practice with the people doing the work. Assess their starting points, teach the workflow, and check what support they need.
- Choose and fund the environment. Match licenses, permissions, usage limits, and ongoing management to the tested workflow.
Review the results with the team before expanding. Return to compliance, competency, and cost whenever the work or technology changes.
Questions leaders ask
What are the three Cs of AI adoption?
RUDI uses Compliance, Competency, and Cost, in that order. Define how AI may be used, help people learn to work within those expectations, and fund the tools and ongoing support the work requires. Revisit all three as use grows.
Should we choose a team plan or an enterprise plan?
Compare the specific plan and contract with your requirements for identity, access, retention, audit records, support, and data handling. A team or business plan may fit one organization; another may need enterprise controls. Document the decision for the actual tools, data, and workflows involved.
Does a paid AI subscription include API usage?
Check the product and agreement. A seat subscription provides access to specified applications and usage allowances. API usage and other metered services can have separate billing. OpenAI, for example, bills the API separately from ChatGPT subscriptions.
Can we use more than one AI platform?
Yes. RUDI recommends a clear purpose and owner for each platform. An organization might use AI in its workplace suite and approve an additional tool for a department. Review the added value, access, training needs, and total cost before expanding that arrangement.
Vendor details checked September 22, 2026. The linked vendor pages describe current features and billing; RUDI’s suggested adoption sequence and workplace examples are our guidance for planning the work.
Work with RUDI
Plan your team’s AI adoption.
Bring your policy questions, current tools, and a workflow your people want to improve. We can help you connect responsible use, practical training, and ongoing AI management.
Discuss your team’s AI environment