Send us Fan Mail

Episode Summary

This episode examines five custom Copilot agents that could reduce administrative work across the sales process. The examples cover CRM maintenance, call coaching, pipeline risk, competitive research, and proposal development. The discussion also explains why governed knowledge, accurate data, human review, and a clear data model are necessary for dependable agent outcomes. [captions | Txt], [Subtitle 1.srt | Txt]

Topics Covered

• The cost of non-selling work
• Custom agents within Microsoft 365 workflows
• CRM updates from meeting and email context
• Sales coaching based on an organization’s methodology
• Pipeline risk and engagement signals
• The danger of forecasting from poor CRM data
• Approved competitive intelligence and battle cards
• Proposal and RFP drafting from curated content
• Human review and automated approvals
• Data modeling and agent governance

Key Takeaways

• Agents are most useful when they reduce friction inside tools sellers already use.
• CRM updates can be suggested from customer interactions instead of entered entirely by hand.
• Coaching should be aligned to a defined sales methodology.
• Forecasting output is only as dependable as the underlying data.
• Competitive and proposal agents should use approved, governed content.
• Human review remains important for specialist, legal, and compliance decisions.
• Reliable deployments begin with a fixed data model and clear governance.

Chapters

00:00:00 Welcome to the Copilot Briefing
00:00:32 Five Sales Agent Use Cases
00:01:37 Why Custom Agents Matter
00:02:36 The CRM Guardian Agent
00:04:49 The AI Sales Coach
00:07:09 The Pipeline Forecaster’s Ally
00:08:45 The Data Quality Trap
00:09:42 Competitive Intelligence Agent
00:11:32 Proposal and RFP Agent
00:12:55 What Reliable Agents Require
00:14:24 Final Takeaways
00:14:58 Closing

🔴 Need help with Microsoft 365 Copilot adoption, governance, and employee training? Let's build your strategy ➡️ https://go.stevecorey.com/chatcoworkautopilot

Send us Fan Mail

Send us Fan Mail

Episode Summary

Microsoft’s new Copilot experience brings Chat and Co-Work together under Home while adding Code and Autopilot as distinct modes. This episode explains how each mode is intended to be used, how the managed runtime could simplify application governance, and why organizations need to prepare for consumption-based costs. It also covers Microsoft IQ, plugins, model access, advanced SharePoint capabilities, OneDrive, and Copilot Search.

Topics Covered

• Copilot Home, Code, and Autopilot
• Chat versus Co-Work
• Office editing inside Copilot
• The Today screen
• Mentioning Copilot in Teams
• Personal Autopilot agents
• End-user coding
• Copilot managed runtime
• Microsoft IQ and Dataverse data
• Plugin and MCP governance
• FinOps for AI and Copilot credits
• Model-family access controls
• Hybrid billing for SharePoint
• Copilot Search in SharePoint and OneDrive

Key Takeaways

• Communicate the redesigned Copilot interface before users encounter it.
• Match quick work to Chat and complex multi-step work to Co-Work.
• Use Autopilot for autonomous personal work without abandoning enterprise-agent investments.
• Establish spending policies and approval paths before expanding usage.
• Decide which business processes are suitable for Code, Autopilot, Co-Work, or enterprise agents.
• Treat advanced SharePoint capabilities as a distinct cost and governance category.

Chapters

00:00:00 Welcome to the Copilot Briefing
00:00:32 Microsoft’s Copilot announcements
00:00:58 The new Copilot app
00:02:31 Office inside Copilot
00:04:13 The Today screen
00:04:59 Mentioning Copilot in Teams
00:05:55 Introducing Autopilot
00:06:46 Co-Work versus Autopilot
00:08:03 Autopilot triggers and actions
00:09:47 Home, Autopilot, and Code
00:10:06 End-user application development
00:11:38 Copilot managed runtime
00:13:18 Microsoft IQ and business data
00:14:46 The plugin registry
00:15:47 FinOps for AI
00:19:06 Agent costs and spending policies
00:21:01 Controlling model access
00:22:06 Cost visibility and usage limits
00:24:38 Copilot in SharePoint
00:28:06 OneDrive and Copilot Search
00:29:00 Business priorities and takeaways

Links

Copilot in SharePoint updates - https://techcommunity.microsoft.com/blog/spblog/sharepoint-ai-innovations-hit-ga-powering-new-copilot-app-and-agents/4555724?WT.mc_id=MVP_320501

The new Copilot app - https://blogs.microsoft.com/blog/2026/09/25/introducing-the-new-copilot-with-home-code-and-autopilot/

Send us Fan Mail

Episode Description

Running AI at scale is expensive, and consumption-based services can make budgets harder to predict. Steve Corey explains Microsoft’s MAI family of in-house models, the role of multi-model orchestration, and why routine work should not automatically be sent to the most expensive model. The episode reviews MAI models for reasoning, coding, image generation, voice, and transcription, then considers how efficient routing could affect Copilot credits, pay-as-you-go usage, custom agents, and AI return on investment.

✅ Connect With Steve

☑️ Socials - https://stevecorey.social

🌍 Web - https://stevecorey.com

📝 Show Notes and Resources 📝

Topics Covered

• The cost of operating AI at scale

• Copilot licensing and consumption-based pricing

• Copilot credits and variable AI costs

• Microsoft’s MAI model family

• Multi-model orchestration and automatic routing

• Reasoning, coding, image, voice, and transcription models

• Pay-as-you-go cost management

• Cost-conscious custom agent design

• Matching model capability to task complexity

• AI solution feasibility and return on investment

Key Takeaways

• Not every request requires a frontier model.

• Efficient routing can reserve expensive models for genuinely complex work.

• Model choice directly affects the consumption profile of AI solutions.

• Custom agents can use different models for routine and critical steps.

• Organizations should continually reassess which models their workloads use.

• Lower-cost models may make usage more manageable, but consumption remains variable.

Chapters

00:00 Welcome to the Copilot Briefing

00:32 Why AI Costs Matter

01:37 The Copilot Pricing Challenge

03:14 Introducing Microsoft MAI

03:49 Why Microsoft Built Its Own Models

05:19 Multi-Model Orchestration

06:12 How Model Routing Works

06:56 MAI Model Lineup

07:01 Reasoning Model

07:53 Coding Model

08:23 Image Model

09:25 Voice Model

10:11 Transcription Model

10:49 Implications for Copilot Costs

12:18 Pay-As-You-Go Economics

13:20 Custom AI Cost Control

13:52 The Bottom Line

15:20 Podcast Takeaways

🔗 Links

MAI Models - (Microsoft Techcommunity)