The short answer
The questions are consistent: is our data safe, who is accountable, what will it cost, how do we know it works, and what if it gets it wrong. This guide answers each from the way agents are built in Dynamics 365 and the way we deliver them.
Is our data safe?
Agents run inside your Microsoft tenant with the access you grant, and they act through the same permissions as your users. We agree the data an agent may read before it is built, and we do not add data sources you have not approved. Review the data protection rules that apply to you, such as the Data Protection Act in Kenya or POPIA in South Africa, with your legal team before connecting personal data.
Who is accountable when an agent gets it wrong?
A named person in your organisation. Each agent has an owner, an approver for the steps that matter, and an audit trail. The agent does the work. Your team keeps the accountability.
Will it replace people?
The agents in Dynamics 365 take over repetitive steps such as keying, matching, chasing and first drafts. People move to review, exceptions, customer conversations and decisions. How you use the time that frees up is a management decision, and the best results come from deciding it before the pilot.
What does it cost?
There are three costs: licences or consumption for the agent capability, implementation to configure and connect it, and the run cost of monitoring and improving it. Pricing and licensing differ by agent and change between release waves, so we give a figure after the readiness assessment, not before it.
How do we know it works?
We agree measures and a baseline before the build, run the agent in review mode against real cases, and compare. Typical measures are cycle time, cost per transaction, error rate, hours saved and satisfaction.
How do we calculate the return?
Take the hours of the process now, multiply by the loaded cost of the people, subtract the run cost of the agent and add any measurable gain such as fewer late fees, captured discounts or faster billing. You can use our return estimator as a starting point. Treat early figures as a range until the pilot gives you your own numbers.
Do we need perfect data first?
No, but you need data that is good enough for the chosen process. Part of the readiness assessment is deciding what has to be fixed before an agent goes live and what can be fixed alongside.
Frequently asked questions
Is our data safe with AI agents?
Agents run in your Microsoft tenant with the permissions you grant. We agree what an agent may read before it is built and review the data protection rules that apply to you.
What does an AI agent project cost?
Licences or consumption, implementation and run cost. Pricing differs by agent and changes between release waves, so we quote after the readiness assessment.
How do we measure return?
Compare cycle time, cost per transaction, error rate and hours against a baseline agreed before the build, and subtract the run cost.