Winter ’27 Release Series
Salesforce Winter ’27 release brings a substantial number of changes across the platform. For most teams, the challenge isn’t finding out what’s new. It’s working out which updates could affect your Salesforce environment, which ones need planning, and which can wait.
In this session, Salesforce MVP Keir Bowden joins Ziipline’s Head of Technology Stu Barber to look at Winter ’27 from a practical delivery perspective. Drawing on real project experience, they focus on the changes worth knowing about and what they could mean for your architecture, integrations, automation and roadmap over the months ahead.
What this session covers
Platform changes that need attention
Some Winter ’27 changes are worth putting straight onto your review list.
The OAuth username-password flow for connected apps is being retired on 20 February, following a delay from the original Winter ’27 timing. Teams using this approach will need to move to alternatives such as client credentials or web server flow with PKCE.
The SOAP API login() call is also moving towards full retirement. It has already been retired from the Winter ’26 API, but organisations relying on older API versions have until Summer ’27 to migrate before it stops working across supported versions.
We also look at earlier Platform API versions moving towards retirement and Salesforce beginning to retire older Apex API versions. For teams managing older integrations or Apex code, these changes are worth investigating before they become a delivery problem.
There are also smaller but useful platform changes, including the new View Setup Audit Trail permission, generally available field history tracking for users, more flexible inline editing in list views, and increased Apex heap limits.
Salesforce’s AI direction is evolving
Winter ’27 arrives alongside a broader change in how Salesforce is approaching AI.
Following Dreamforce, the session looks at AIforce, Salesforce’s new direction for exposing Salesforce data and capabilities to AI tools. We also discuss Claudeforce and the Salesforce and Anthropic partnership, including the initial Salesforce in Claude offering.
Another key announcement is Koa, Salesforce’s specialised reasoning model for agentic CRM work. Unlike general-purpose frontier models, Koa has been developed specifically around CRM tasks and is designed to remain under Salesforce’s control, with the option of keeping data within the Salesforce environment.
The webinar also looks at how Setup with Agentforce is developing. It can now work with multiple users, explain access granted by permission sets and profiles, create and update roles, and help identify privileged users who still need phishing-resistant MFA.
For teams considering AI on the Salesforce platform, these developments are useful indicators of where the ecosystem is heading. They also raise practical questions around permissions, governance, data and how much autonomy to give AI tools.
Automation and development get more control
Winter ’27 brings several improvements that make Salesforce automation easier to build, test and maintain.
Flow element groups allow teams to organise several elements into a single conceptual unit, making complex flows easier to understand without changing their underlying behaviour.
Flow Builder also provides greater visibility of Agentforce metadata, while new validation catches certain errors when a flow is saved rather than waiting until runtime. Record-lock errors can also trigger an automatic retry after a short delay.
Testing gets more attention in this release. Flow now has a dedicated test mode where teams can create and run test scenarios, use assertions, and review test coverage. Mocks for actions and subflows also allow teams to test responses without relying on external systems.
For developers, synchronous Apex heap limits increase to 10 MB and asynchronous limits to 25 MB. Apex integration tests can also now call any endpoint rather than being limited to Agentforce and Data 360 endpoints.
These changes give teams more ways to test and manage automation before it reaches production.
Reporting and analytics become more flexible
There are several Winter ’27 updates aimed at making reporting and analytics more useful in day-to-day work.
Lightning reports can now provide record previews directly within the report, while Lightning reports and dashboards can be embedded into Experience Cloud sites built with Lightning Web Runtime.
Data 360 adds joined reports, allowing teams to analyse information from different data model objects together. Dimensional hierarchies are also moving to general availability, making it easier to drill down and roll up data within reports.
CRM Analytics adds Gantt charts to visualise schedules and dependencies, plus a recycle bin for deleted lenses and dashboards.
Data 360 reporting also gains an AI sentiment function in beta, allowing text such as support tickets, reviews and call transcripts to be classified by sentiment.
The useful takeaway is less about the number of reporting features and more about how these changes could reduce manual work and give teams better visibility into their data.
More changes across Salesforce
The webinar also covers a much wider set of Winter ’27 updates across Salesforce.
In Marketing Cloud Next, the session looks at buyer engagement capabilities, agentic content creation, completion actions and bulk content deletion. There are also improvements around email testing, multilingual content, image recommendations and interactive email conversations.
Revenue Management sees changes to promotions, product catalogues, pricing, transaction management, approvals and billing. Promotions move from beta towards general availability, while new capabilities include compound uplift, weekly proration, backdated asset changes and automatic refreshes within transaction management.
For Sales, Agentforce Prospecting can now be scheduled to run automatically, while teams can prioritise prospects based on previous engagement. The release also introduces call coaching and additional controls over what Einstein Conversation Insights captures.
In Slack, Agentforce Sales agents can work directly within Slack, with activity synchronised back to Salesforce.
For Service, the session covers updates to Agentforce Service Assistant, including dynamic service plans, real-time issue resolution, Service Assistant chat, case catch-up insights and suggested replies. Winter ’27 also introduces knowledge blocks, knowledge similarity, and further improvements to case management.
There is a lot here, but the aim of the session is not to work through every feature. It is to highlight the changes most likely to affect how teams build, manage and use Salesforce.
What should you actually do with Winter ’27?
A Salesforce release does not automatically mean you need to change your platform.
The useful starting point is to look at what has changed and ask how it relates to your existing environment. Does it affect an integration you rely on?
Could it change how you manage security or permissions? Does it give you a better way to handle something you are already building? Or does it solve a problem that simply isn’t relevant to your organisation?
That distinction matters because not every new feature deserves a place on your roadmap.
It is also worth keeping an eye on what Salesforce is releasing in beta. A beta feature may change before it reaches general availability, or it may not make it that far at all. But it can still give you a useful view of the problems Salesforce is investing in and the direction the platform is taking.
The earlier you understand those developments, the more time you have to assess them properly. You can test where appropriate, consider the implications for your architecture and decide whether a capability is worth adopting when it becomes generally available.
Winter ’27 also reinforces the importance of reviewing older parts of your Salesforce environment. API retirements and authentication changes may not come with an obvious day-to-day benefit, but they can become significant if an integration or piece of Apex code has been running unchanged for years.
If Winter ’27 has raised questions about which changes deserve attention, how the new AI capabilities fit into your platform, or what should make it onto your roadmap, we’re happy to talk it through.