check before: 2023-10-01
Product:
Copilot, Microsoft 365 Copilot, Teams
Platform:
Developer, Online, World tenant
Status:
Change type:
New feature, Admin impact
Links:
Details:
Summary:
A new declarative agent control allows developers to prioritize curated knowledge sources over internal agent knowledge, improving response alignment and trust. Available now, it affects developers using Agent Builder or Teams Toolkit. No admin action is required; developers should start using the new control.
Details:
What and Why
We're introducing a new declarative agent control that allows creators to discourage responses based on internal agent knowledge. This gives developers more precision in how their agents respond, helping ensure answers are grounded in curated knowledge sources. This update supports better alignment with enterprise content strategies and improves trust in agent responses.
Rollout Schedule
General Availability (Worldwide): Available now.
Impact on Your Organization
Who is affected:
Developers and makers building declarative agents using Agent Builder (Copilot Studio) or Teams Toolkit (TTK).
What will happen:
A new toggle now appears in Agent Builder to control whether agents can rely on internal model knowledge.
The control is off by default.
Developers using Teams Toolkit can enable this behavior via the agent manifest.
This setting helps ensure agent responses are grounded in intended knowledge sources.
Action Required / Recommendations
No admin action is required at this time.
If your organization builds agents using Agent Builder or TTK:
Begin using the new control via the manifest or toggle.
Review the Declarative agent schema 1.4 for Microsoft 365 Copilot - Behavior overrides object for implementation details.
Watch for a dedicated documentation link for this feature, which will be shared soon.
Compliance Considerations
No compliance considerations identified, review as appropriate for your organization.
Change Category:
XXXXXXX ... free basic plan only
Scope:
XXXXXXX ... free basic plan only
Release Phase:
Created:
2025-07-12
updated:
2025-07-12
Task Type
XXXXXXX ... free basic plan only
Docu to Check
XXXXXXX ... free basic plan only
MS Preperations
XXXXXXX ... free basic plan only
MS Urgency
XXXXXXX ... free basic plan only
MS workload name
XXXXXXX ... free basic plan only
summary for non-techies**
XXXXXXX ... free basic plan only
Direct effects for Operations**
Response Accuracy
Without proper preparation, developers may inadvertently allow agents to rely on outdated internal knowledge, leading to inaccurate responses and decreased user trust.
- roles: Developers, End Users
- references: https://learn.microsoft.com/en-us/microsoft-365-copilot/extensibility/declarative-agent-manifest-1.4#behavior-overrides-object
User Experience
If developers do not utilize the new control effectively, users may receive inconsistent or irrelevant information from agents, negatively impacting their experience and productivity.
- roles: Developers, End Users
- references: https://learn.microsoft.com/en-us/microsoft-365-copilot/extensibility/declarative-agent-manifest-1.4#behavior-overrides-object
Configutation Options**
XXXXXXX ... paid membership only
explanation for non-techies**
Imagine you're organizing a library. Traditionally, librarians rely on their extensive knowledge to help visitors find the right books. However, you want to ensure that the recommendations align with a curated list of trusted books that best represent the library's focus. This is similar to the new feature in Microsoft 365 Copilot, where developers can now prioritize specific knowledge sources over the internal knowledge of the AI agent.
This change is like giving librarians a guidebook that highlights preferred books, ensuring that their suggestions are always in line with the library's curated collection. By using this guidebook, librarians can provide more reliable and consistent recommendations to visitors. Similarly, developers can now use a new control to ensure that AI agents prioritize curated knowledge, improving the accuracy and trustworthiness of their responses.
For those building AI agents using tools like Agent Builder or Teams Toolkit, this new control is like a switch that allows them to decide whether the AI should rely on its internal knowledge or stick to the curated sources. It's off by default, meaning the AI will use its internal knowledge unless the developer decides otherwise. This approach helps maintain consistency with the organization's content strategy, much like how a library might focus on promoting certain genres or authors.
No immediate action is required from administrators, but developers are encouraged to start using this new feature to align their AI agents with their organization's knowledge priorities. This change doesn't affect compliance, but organizations should review it to ensure it fits their specific needs.
In summary, this update gives developers more control over how AI agents respond, similar to how a library might guide its staff to recommend certain books, ensuring that the responses are trustworthy and aligned with the organization's goals.
** AI generated content. This information must be reviewed before use.
a free basic plan is required to see more details. Sign up here
A cloudsocut.one plan is required to see all the changed details. If you are already a customer, choose login.
If you are new to cloudscout.one please choose a plan.
Last updated 3 months ago ago
