The demonstration

The example began with a quote in Dynamics 365 where a discount above a set threshold triggered approval. The flow retrieved customer and seller information, presented decision context and returned the answer. An extended version used an adaptive card in Teams and allowed a defined group to respond.

What this means for the organisation

The webinar also showed why filtering belongs in the trigger, how previous runs can support testing and how versions are published. AI supplemented the information, while the decision remained with an accountable approver.

The discount threshold made the demonstration testable: the same business rule triggered the flow whenever its condition was met. Customer and seller data supplied context, while the Teams card moved the response into the approver’s workspace. Run history and versioning also showed how a faulty flow can be traced and corrected.

Microsoft’s 2016 Flow launch records the product’s original role in connecting services, data and automated actions. It explains the platform background, while the webinar shows a later, concrete approval flow. The product source describes the capability; the recording shows selected rules, data and a human decision in one example.

A production approval flow needs ownership of thresholds, recipient groups and error handling. Testing should include missing customer data, rejection, absence and parallel responses. Run history can then reveal whether a case stopped during retrieval, routing or human review, and whether the rule needs changing.

Sources

Primary source: ProsessPilotene, “Automatiser godkjenningsprosesser med Power Automate, AI og Teams!,” 15 September 2025 (Norwegian).

Microsoft, “Announcing Microsoft Flow General Availability,” 31 October 2016.

Further reading

How do you create your first flow?

Join the discussion

What would you investigate first before using this? Share your questions or experience in the discussion.