A Path Forward For Enterprise AI: LAB³ embraces the new Microsoft Agent Framework to provide business-level AI Outcomes 

Microsoft has announced its newly available open-source Microsoft Agent Framework and LAB³ is already in prime position to help with deployment as a default.  

“This is not ‘just another framework’ – it’s a major step forward for Agentic AI solutions,” says Jason Leonard, AI Consulting Lead, LAB³.

“In practice, Microsoft’s new framework simplifies production hardening and gives enterprises a single path from prototype to governed, long running agents.” 

Agentic AI frameworks provide the structured foundation for building AI systems that go beyond ‘Question and Answer’ and become goal-driven agents that support multi-step enterprise business processes that integrate with enterprise systems.  

The Microsoft Agent Framework brings together open standards including:

• Model Context Protocol (MCP) for tool use, Agent-to-Agent (A2A) for cross-runtime collaboration
• OpenAPI for systems integration, graph-based workflows for deterministic execution, and
• OpenTelemetry for end-to-end observability.  

“Microsoft has designed the Agent Framework for local experimentation and seamless promotion to Azure AI Foundry with built-in durability, telemetry, and compliance features,” says Jason. 

The Agent Framework provides a unified SDK and runtime that converges the strengths of Microsoft’s pre-existing frameworks – Semantic Kernel (enterprise-ready connectors, observability) and AutoGen (multi-agent orchestration patterns) – into one foundation for building, deploying, and governing agentic applications. 

“LAB³ has a long history of ‘opinionated design’ to help our clients obtain business value faster,” says Jason.

“In this tradition, the Agent Framework will become the default Agentic AI framework deployed by LAB³ as it moves from Preview to General Availability.”  

In addition to the new Agent Framework, Microsoft has also formalized the role of the Microsoft 365 Agents Toolkit to wire agentic apps into familiar tools including Microsoft Teams, Outlook, the M365 Copilot surface, and other messaging channels, with project scaffolding, sideloading, Azure provisioning, and an Agents Playground for local chat testing.

“This makes a messaging UI (like Teams and Slack) first-class front ends for Agentic AI applications,” says Jason.

“Microsoft’s Copilot customers can already create custom AI agents with Copilot Studio. The Agent Framework extends this capability by supporting complex, multi-step or multi-agent workflows.”

LAB³ helps our clients select the right tool (or combination) for the job, taking into account business requirements, costs and privacy/security needs, and then deliver solutions that provide business outcomes. 

LAB³ Deployment of the new Microsoft Agent Framework 

The Agent Framework (AF) is set to become the default Agentic AI framework deployed by LAB³ as it moves from Preview to General Availability. 

• Teams first agent experiences
  LAB³ combines the AF backends with the Microsoft 365 Agents Toolkit to power conversational agents within Microsoft Teams (and Outlook/M365 Copilot) as a familiar user interface for business users. This shortens adoption cycles, leverages single sign-on, and meets users where they already work. 
• Governed, long running workflows
  Using AF Workflows (graph-based orchestration with checkpointing and human-in-the-loop), LAB³ implements end-to-end processes, for example: email received → request triage → data collection → approvals, all monitored by OpenTelemetry tracing so Operations can see every step and cost.
• Practical Implementations in Real-World Complexity
  LAB³ integrates AI applications with business systems via OpenAPI, connect to existing tools via MCP servers, and coordinate multi-agent “handoffs” across runtimes with A2A, thereby allowing our clients to integrate in heterogenous IT environments. 
• Risk managed deployments
  LAB³ enables AF’s built-in observability in Azure AI Foundry, enables Prompt Shields where appropriate, and attaches policy controls so InfoSec can monitor and enforce behaviour from day one. 

“While these latest technology advances extend Microsoft’s lead in agentic frameworks for enterprise use, the need for organizations to target business outcomes remains as critical as ever,” says Jason. 

To help business executives gain a business-level overview of AI terms and capabilities, including Agentic AI, Jason has authored a comprehensive and practical AI Primer.

“In partnering with clients along their AI journey, LAB³ integrates with the line-of-business to thoroughly understand the context of their organization,” says Jason.

This process is facilitated via an AI Business Envisioning Workshop.

LEARN MORE ABOUT OUR AI BUSINESS ENVISIONING WORKSHOPS