October is four months away. That sounds like a comfortable buffer. It isn’t.
D365 Wave 2 2026 deploys in October, and it’s not a standard maintenance release. Wave 2 will build directly on the agentic foundations Microsoft established in Wave 1: expanded Finance agent capabilities, deeper Business Central AI agent workflows, broader MCP integration in Finance and Supply Chain, and more autonomous Copilot features across Sales and Service. Every scripted test suite that wasn’t updated for Wave 1’s agent changes will face compounding exposure in Wave 2.
The pattern that plays out every October for teams without wave-ready coverage is familiar: the update deploys, recordings break, developers are pulled from other work, a regression sprint begins, the release goes live 2–3 weeks late with coverage gaps. That cycle is not inevitable. It is the product of starting preparation in September rather than June.
This guide is the June plan: what Wave 2 is bringing to each D365 module, the 8-question audit to identify your coverage gaps right now, how to build agent-based coverage that self-heals through Wave 2 changes, and the month-by-month preparation timeline from June to October go-live.
If you want the full picture of Dynamics 365 test automation before diving into Wave 2 specifics, the complete guide to D365 test automation covers the landscape. The Beyond RSAT guide explains the structural reason scripted tools repeat this cycle every wave.
1. Why June Is the Right Time to Start Wave 2 Preparation
The argument for June is straightforward: it is the only month that gives you enough runway to do this properly.
Microsoft typically opens D365 Wave 2 early access in August–September. That preview window is 6–8 weeks long. It is the only time available to validate your test suite against Wave 2 changes before they reach production. For teams with scripted test suites, those 6–8 weeks are entirely consumed by identifying which scripts broke and repairing them. For teams with AI agent coverage built before August, those 6–8 weeks are available for genuine validation, confirming that every high-risk D365 process produces the correct outcome in the Wave 2 environment.
June gives you July to build coverage and August to validate it. That is the sequence that eliminates the October regression sprint. Every month you delay compresses that window.
There is a second reason June specifically matters for Wave 2 2026. Wave 1 (April–September 2026) introduced genuinely new capabilities, not just UI updates. Business Central now has an Agent Designer and autonomous payables and sales agents. D365 Finance has expanded MCP support for custom autonomous agents. These are new process layers that most D365 test suites have not been updated to cover. Wave 2 will change these capabilities further. Teams that have not validated against Wave 1 agent changes are already carrying a gap that Wave 2 will deepen.
“The teams that have no regression sprint in October are not the ones who worked harder in September. They are the ones who made a different decision in June.”
2. What Wave 2 2026 Is Bringing to D365, and What Your Test Suite Must Cover
Microsoft has not yet published the Wave 2 2026 release plan, that typically releases in late July or August. What we know from Wave 1’s stated trajectory, Microsoft’s published roadmap priorities, and the direction of each D365 module is where Wave 2 will push the product. The table below maps each module to the Wave 2 developments that will build on Wave 1 and the test coverage that will be required as a result.
| D365 Module | What Wave 2 2026 is building on | What your test suite must cover |
| D365 Finance | Deeper agentic finance close, expanded MCP agent capabilities for AP/AR, faster GL automation, enhanced cross-entity consolidation | Finance agent output validation (GL account, dimensions, period) · AP agent decision accuracy · MCP workflow outcome assertions · intercompany balance at consolidation |
| D365 Supply Chain | Enhanced AI demand forecasting, supplier communication agent maturation, improved quality management for sample handling | Demand plan accuracy · supplier agent transaction output · procurement-to-pay end-to-end cycle · quality inspection workflow outcomes |
| D365 Business Central | AI agent expansion (payables, sales orders, custom agents), Agent Designer maturation, deeper Power Platform and Power Automate integration | BC AI agent decision accuracy · AL extension regression after agent framework updates · Power Automate flow behaviour · Agent Designer workflow outcomes |
| D365 Sales & Customer Service | Deeper Copilot integration across CRM, automated case routing agent, enhanced scheduling and field service intelligence | Copilot suggestion accuracy against CRM data · case routing decision validation · quote-to-order flow integrity · lead scoring output correctness |
| Finance & Ops cross-app | Expanded MCP for custom autonomous agents, natural language ERP queries, asynchronous dual-write operations improvements | MCP agent output validation · dual-write sync accuracy after Wave update · NL query result correctness · financial dimension mapping across apps |
The Finance and Business Central rows carry the most testing weight for most organisations. Wave 1 established autonomous AI agents inside both modules. Wave 2 will deepen their capabilities, which means any team that deployed Wave 1 agents in production without automated outcome validation is operating with an untested process layer. Wave 2 changes will land on top of that unvalidated foundation.
For Finance specifically: the distinction between UI validation and financial outcome validation is not a technical preference, it is a compliance requirement. A Finance agent that posts an invoice to the wrong GL account will return a green UI status. Only field-level assertion against the GL entry, financial dimensions, and posting period reveals the error. The D365 Finance testing guide covers this in detail.
For Business Central: Wave 1’s Agent Designer means BC teams can now build custom agents that execute business processes autonomously inside the platform. Testing those agents requires understanding what outcome they were supposed to produce, not just whether the Designer completed successfully. The Business Central test automation guide covers Wave 1 2026 AI agent testing for BC specifically.
3. The Wave 2 Readiness Audit: 8 Questions to Answer About Your D365 Test Suite Now
Before building a Wave 2 preparation plan, you need to know where your current D365 test suite is exposed. These 8 questions map directly to the Wave 2 changes that are coming and the coverage gaps most commonly found in D365 test environments going into a Wave release.
| # | Question | Why it matters for Wave 2 |
| 1 | Is your test suite validating AI agent outputs, or only UI interactions? | Wave 1 introduced autonomous agents inside D365. Wave 2 matures them. If your tests confirm that a form submitted but not that the agent made a correct decision, you have a gap Wave 2 will expose. |
| 2 | How many tests broke when Wave 1 (April 2026) deployed? | If the number exceeded 20% of your suite, you have maintenance debt that Wave 2 will compound. Every scripted tool has the same structural exposure. |
| 3 | Do you have cross-module coverage for Finance period close and the P2P cycle? | These are the two highest-risk D365 workflows and the most likely to be affected by Wave 2 Finance and Supply Chain changes. Manual-only or isolated module testing is insufficient. |
| 4 | Does your Business Central test suite cover Wave 1 AI agent workflows? | Wave 1 introduced BC AI agents and the Agent Designer. Wave 2 matures both. If BC testing was not updated for Wave 1 agent changes, Wave 2 will compound that gap. |
| 5 | How long did your Wave 1 regression sprint take? | If the answer is more than 5 business days, you are in the maintenance-debt zone. Wave 2 will produce the same sprint unless the architecture changes before October. |
| 6 | Do you have test coverage for MCP workflows in D365 Finance and Supply Chain? | Wave 1 introduced MCP for custom autonomous agents in AP and procurement. Wave 2 expands this. MCP-enabled processes require outcome validation, not UI replay. |
| 7 | Does your D365 Finance testing produce field-level evidence of correctness? | For audit and compliance, GL account, financial dimension, and posting period must be asserted explicitly. Pass/fail status is not evidence, it is confidence. |
| 8 | Have you registered for the Wave 2 2026 early access preview? | Microsoft typically opens Wave 2 early access in August. Teams registered for preview have 6–8 weeks to validate before production deployment. Unregistered teams have zero. |
The most important questions for Wave 2 2026 specifically are 1, 4, and 6. These three map directly to the new capabilities Wave 1 introduced that Wave 2 will build on: AI agent outputs (question 1), Business Central Agent Designer workflows (question 4), and MCP autonomous agent processes (question 6). If your test suite was built before Wave 1, or if it was not updated to cover Wave 1’s agent changes, these three gaps are guaranteed exposures when Wave 2 arrives.
Question 7, whether your Finance testing produces field-level evidence, is the compliance question that separates test suites that satisfy auditors from those that don’t. If your D365 Finance testing returns pass/fail status without GL-level assertions, that is not a testing limitation, it is an audit risk. The D365 multi-entity and intercompany testing guide covers what field-level financial evidence looks like and why it matters.
| A useful framing for each of the 8 questions: If your answer to question N is “no,” what will the first week of October look like when Wave 2 deploys? For most teams, questions 1, 4, and 6 produce the same answer: a developer sprint. |
4. Building Self-Healing D365 Coverage That Survives Wave 2
The architectural reason scripted D365 test suites break on every wave is that they are coupled to UI state. They record which button to click and which field to check. When Wave 2 moves the button, renames the field, or adds a step to the process, the recording breaks. It doesn’t matter whether the tool is RSAT, a Selenium script, an RPA bot, or a low-code recorder, the coupling point is the same, and the failure mode is the same.
Sofy’s D365 agents are coupled to process outcomes, not UI state. When Wave 2 changes how a Finance period close is initiated, the agent does not break, it detects the change, adapts its path through the new UI, and validates that the period close produced the correct trial balance, correctly closed the sub-ledgers, and correctly locked the period. The test continues to pass. A self-healing event is logged showing what changed and how the agent adapted. No developer required.
The three coverage areas to build before the Wave 2 preview opens
Given the four months between June and October, three areas of D365 coverage should be built first, in this order:
- First: Finance and Supply Chain AI agent outcome validation. The AP agent, MCP-enabled procurement workflows, and Finance autonomous close capabilities introduced in Wave 1 are the areas where Wave 2 will change the most. Build Sofy agent coverage that validates the financial outcomes these agents produce, not the UI path they take.
- Second: Business Central AI agent and AL extension regression. Wave 1’s Agent Designer and autonomous BC agents need test coverage that validates their decision outputs. Additionally, every custom AL extension in your BC environment needs process-level regression coverage built before Wave 2, because Wave 2 will update the agent framework that extensions interact with.
- Third: Cross-module E2E workflows for Finance period close and P2P. The D365 Finance period close sequence and the procurement-to-pay cycle are the two highest-risk multi-step workflows in any D365 environment. They are also the two most commonly under-tested. Building Sofy agent coverage for both before Wave 2 eliminates the scenarios most likely to produce silent errors in a production environment.
The self-healing testing framework, and specifically the difference between Level 1 selector retry, Level 2 element re-identification, and Level 3 workflow-level adaptation, is covered in detail in the self-healing test automation guide. Level 3 is what makes Wave 2 preparation sustainable rather than a once-per-wave rebuild.
5. Your June–October D365 Preparation Timeline
Five months. Five phases. One objective: arrive at October’s Wave 2 go-live with a test suite that has already passed against the preview environment and requires no emergency sprint when production deploys.
| Month | Phase | What to do |
| June 2026 | Audit & Gap Map | Run the 8-question audit above. Map every D365 process changed by Wave 1 against your current test coverage. Identify the gaps, Finance agent workflows, BC AI scenarios, MCP processes, cross-module E2E. Prioritise by financial risk and change frequency. |
| July 2026 | Build Agent Coverage | Build Sofy agent coverage for all high-risk Wave 2 target processes. Start with Finance period close, AP agent workflows, and BC AI agent scenarios. Run your first suite against the current Wave 1 D365 sandbox to confirm baseline accuracy. |
| August 2026 | Preview Registration | Register for D365 Wave 2 2026 early access (typically opens late July–August). Connect Sofy to your Wave 2 preview environment. Trigger your full agent suite. Review self-healing event logs, agents adapt to preview changes automatically. |
| September 2026 | Preview Validation | Complete your Wave 2 preview validation cycle. Every high-risk process should have passed the agent suite against the preview environment by end of September. Expand coverage to secondary modules (Sales, Customer Service, Field Service) as capacity allows. |
| October 2026 | Wave 2 Go-Live | Wave 2 reaches production. Your Sofy agent suite has already validated against the preview. The October go-live is a confirmation run, not a regression sprint. No developer emergency. No delayed release. Full coverage from day one of production deployment. |
Two things make this timeline work in practice. The first is the August preview registration, this is a hard date that determines whether you have 6–8 weeks of preview validation time or zero. Register early at the Microsoft 365 admin centre or through your D365 partner before the end of July.
The second is the quality of the coverage built in July. Agents built quickly in July will produce shallow assertion coverage, they will confirm that processes ran but not that they produced correct financial outcomes. Agents built with field-level assertions, GL account, financial dimensions, posting period, sub-ledger balance, produce the kind of evidence that passes both the Wave 2 preview validation and any subsequent audit review.
For teams with D365 Supply Chain coverage in scope, the D365 Supply Chain testing guide maps the procurement, inventory, and warehouse workflows to automate first. For Finance period close across multiple legal entities, the D365 multi-entity testing guide covers the intercompany and consolidation scenarios specific to Wave preparation.
Build your Wave 2 2026 D365 coverage before the preview window opens.
Sofy’s D365 AI agents cover Finance, Supply Chain, Sales, and Business Central, self-healing through every Wave release, every agent update, and every MCP workflow change. No scripting. No regression sprints.
Choose your platform:
Frequently Asked Questions
D365 Wave 2 2026 covers October 2026 through March 2027. Microsoft typically opens the early access preview environment in August–September, giving teams 6–8 weeks to validate before production deployment begins in October. Early access registration usually opens through the Microsoft 365 admin centre in late July.
Wave 2 2026 builds directly on the agentic foundations established in Wave 1. Wave 1 introduced autonomous AI agents inside Business Central, expanded MCP capabilities in Finance and Supply Chain, and began the transition toward agentic ERP processes. Wave 2 matures these capabilities, deeper Finance agent integration, more sophisticated BC agent workflows, expanded cross-app MCP support. Teams that didn’t update test coverage for Wave 1’s agent changes face compounding exposure in Wave 2.
Most D365 test suites, RSAT recordings, RPA bots, script-based automation, are coupled to specific UI states. They record which buttons to click and which fields to check. When a wave update changes a form layout, renames a field, adds a required step, or introduces a new process path, those recordings break because the UI state they were built on no longer exists. Sofy’s agents are coupled to process outcomes rather than UI state, which is why they adapt when waves change the interface rather than breaking.
Run the 8-question readiness audit in this guide and answer honestly. Identify how many tests broke in Wave 1 and how long the repair sprint took. Identify whether your test suite validates AI agent outputs or only UI interactions. Identify whether you have cross-module E2E coverage for Finance period close and P2P. Those three gaps, questions 1, 3, and 4, are the Wave 2 exposures most likely to produce a regression sprint in October if not addressed in June and July.
