A broad plan for business applications and agents

Microsoft brought together product direction for sales, customer service, field service, finance, supply chain, HR and commerce in the first release wave of 2026. For Power Platform, the plan included modernised model-driven apps, better mobile and offline use, search improvements, Dataverse capabilities, and additional security and cost governance.

Copilot Studio and role-based agents had a prominent place. Microsoft described agents that could combine CRM data with email, meetings and other sources to support work in sales and service. This was still a plan for capabilities scheduled between April and September. Each item could be planned, in preview or progressing towards general availability.

The market expected stronger governance and fewer loose experiments

Gartner grouped its strategic technology trends for 2026 around building, orchestrating and protecting digital value. Forrester described 2026 as a year for moving AI from hype into practical work and emphasised skills and accountability in regulated environments. Both perspectives supported reading the product plan with governance and operations carrying as much weight as new features.

OpenAI’s practical guide recommended agents for complex workflows involving unstructured information and difficult rule sets, and described models, tools and instructions as core components. Anthropic advised starting with simple patterns and adding complexity only when simpler approaches fell short. Together they offered a useful correction to a long feature list: not every process required an agent.

Turning a product plan into an organisational plan

An organisation first needed to identify planned changes that affected existing processes, integrations or training. Each relevant item then required an owner who tracked real availability in the release planner and the tenant. Budget and capacity could be reserved for testing, while dependencies on planned features had to remain explicit risks.

For agent capabilities, data and action boundaries were central. Teams needed to know which Dataverse tables, documents and messages an agent could use and which actions required approval. Cost per completed task, response time, error rate and human rework were more informative than the number of features enabled.

The release plan gave management time to prioritise before rollout. It was most useful as a working document for testing, architecture and change management. By checking status continuously and connecting each feature to a measurable need, organisations could avoid planning around capabilities whose dates changed or that did not fit the real process.

Change management needed to be planned at the same time. New interfaces and agent capabilities affected roles, routines and expectations even when technical activation was small. A responsible product owner therefore needed to connect the release plan to communication, training, support and pilot exit criteria.

Sources

Microsoft: 2026 release wave 1 plans, 18 March 2026

Gartner: Top strategic technology trends for 2026

Forrester: Predictions 2026 — AI moves from hype to hard-hat work

OpenAI: A practical guide to building agents, 2025

Anthropic: Building effective agents, 19 December 2024

Further reading

Read more: What is Customer Insights Journeys?

Join the discussion

Which experience from your own organisation is most relevant to this story? Share what you learned.