Architect · Microsoft MVP · Speaker

Holger
Imbery.

I build enterprise AI as an architect. Through instruo365, I turn that work into practical talks, tools, and labs, sharing what works, what doesn't, and how to take ideas into production.

instruo, Latin for “to teach, inform, instruct.”

Agentic AICopilot StudioMicrosoft FoundryEnterprise architectureAzure
Holger Imbery Berlin, Germany
About

Architecture that works beyond the demo.

Principal Architect at NTT DATA, Microsoft MVP and MCT. I help teams turn emerging Microsoft technologies into secure, scalable solutions.

I work at the intersection of enterprise AI, intelligent automation, and the Microsoft ecosystem. My focus is turning promising ideas into secure, scalable, well-governed solutions that work beyond the demo.

Through talks, tools, labs, and field notes, I share practical approaches and lessons learned from architecture and implementation.

Sessionize · Current sessions

Talks built from real architecture work.

Sessions on customer-facing AI, agent economics, and the observability and operational practices needed to build, govern, and run agentic solutions in production.

Agent operations · Copilot Studio

When Copilot Studio Agents Go Wrong: Designing for Observability, Incident Response, and Recovery

Design agents to be diagnosable before incidents happen. Trace conversations, orchestration, tools, APIs, identities, and business systems to investigate failures, contain impact, find root causes, and turn incidents into evaluation and regression tests.

Session details
Customer service · Copilot Studio

Building Agentic Customer Service Solutions with Copilot Studio and Real-Time Voice

Build a customer service agent with Copilot Studio's Customer Conversation Harness, then connect enterprise knowledge, business actions, Dynamics 365, Teams Phone Agents, voice, and human escalation.

Session details
Architecture · Voice self-service

One Goal, Two Paths - Choosing the Right Architecture for AI-Powered Voice Self-Service

Compare Dynamics 365 Contact Center and Teams Phone Agents with Copilot Studio through one customer scenario, and leave with a practical framework for choosing the right architecture.

Session details
ALM · GitHub Actions

From Maker Success to Enterprise Success: ALM and GitHub Actions for Microsoft Copilot Studio Agents

Move Copilot Studio agents from proof of concept to production with solutions, source control, lifecycle management, automated deployments, governance, and operational ownership.

Session details
Agent evaluation · Copilot Studio

Rethinking Test Automation: Ship Your Copilot Studio Agents with Confidence

Build realistic test sets, choose evaluation methods for variable agent responses, and gate deployments on evaluation results. Explore multi-turn scenarios, custom graders, and the Copilot Agent Kit's value, limits, and costs.

Presenting with Jennifer Krause · LinkedIn

Presented at ColorCloud 2026.

Session details
FinOps · Agentic enterprise

FinOps for Agents: Managing the Economics of the Agentic Enterprise

Connect agent runtimes and consumption telemetry to cost allocation, governance, and outcomes, measuring cost per successful task and business value, not just tokens.

Session details
Multi-agent systems · Copilot Studio

From Harness to Handoff: Multi-Agent Architecture in Microsoft Copilot Studio

Choose the right harness and composition pattern for each workload, then design clear context handoffs, access controls, evaluation, monitoring, and human checkpoints.

Presenting with Jennifer Krause · LinkedIn

Presented at European Power Platform Conference, Copenhagen (2026) & Nordic Summit, Billund (2026).

Session details

For the complete session list and latest updates, visit Holger's Sessionize profile.

Previous talks

More topics from the stage.

Previous sessions, with slides and demos in the talk repository.

Copilot Studio

Building Agents with Copilot Studio: From Orchestration to Customer Service

Choose the right harness for your workload, connect enterprise knowledge and tools, and design event-driven workflows with clear human checkpoints. Explore customer-facing chat and voice scenarios, with the Customer Conversation harness explicitly covered as preview.

Dynamics 365

From Inbox to Resolution: Copilot Studio, Microsoft Foundry & Dynamics 365

Build an email-first service solution that automatically classifies incoming messages with Azure AI Foundry models and Copilot Studio, then routes them into Dynamics 365 case management. Provide self-service replies and guided resolution for common requests, trigger business actions such as updating orders or creating cases, and hand off to a human agent with full context when the issue needs one.

Workshop

Agent in a Day: Build, Test & Publish

A hands-on introduction aligned with Microsoft's current Agent in a Day learning path: declarative agents in Microsoft 365 Copilot Chat and custom engine agents in Copilot Studio. Explore knowledge, tools, agent flows, generative orchestration, and autonomous functionality, with practical testing and publishing steps.

Presented with Jennifer Krause · LinkedIn

Monthly, at various locations.

Microsoft Foundry

Agentic AI with Microsoft Foundry: From Prototype to Production

Design an agent solution with Microsoft Foundry Agent Service and choose an appropriate runtime and tool integration strategy. Cover Microsoft Entra identity, access controls, tracing, evaluations, and operational monitoring, separating production architecture from preview capabilities.

Browse previous talk content

Topic Areas

Where I work and speak

My work sits at the intersection of enterprise AI, intelligent automation, low-code development, and agentic systems.

I focus on how organizations can move from experimentation and individual copilots toward scalable AI and automation capabilities that work across applications, data, processes, and teams. From low-code solutions to autonomous agents and enterprise-wide orchestration, I connect strategic direction with the practical realities of architecture, implementation, governance, security, operations, and adoption.

AI Foundry & Agentic AI

Designing and building autonomous and multi-agent systems with Microsoft AI Foundry, Azure AI, and the broader Microsoft agent ecosystem.

Topics include agent architecture, reasoning and orchestration patterns, grounding and enterprise data integration, model selection, tools and actions, multi-agent collaboration, observability, evaluation, security, and responsible AI. I’m particularly interested in what happens after the prototype: how organizations operate, govern, monitor, scale, and financially manage agentic AI in production.

Copilot Studio

Building enterprise agents and conversational AI experiences with Microsoft Copilot Studio.

This includes custom agents, voice-enabled experiences, autonomous agents, agent flows, test automation, connectors and integrations, multi-agent orchestration, and extending Microsoft 365 Copilot with organization-specific knowledge and actions.

Beyond building the agent itself, I focus on the enterprise lifecycle around it: environments, ALM, governance, security, testing, monitoring, adoption, and establishing repeatable patterns for moving agents from experimentation into production.

Power Platform

Using Power Apps, Power Automate, Power Pages, Dataverse, and the wider Power Platform to turn business processes into scalable digital solutions.

My focus extends beyond individual apps and flows to enterprise automation architecture: combining low-code, APIs, connectors, cloud services, RPA, AI, agents, and human-in-the-loop processes. This also includes governance, ALM, environment strategy, reusable components, Center of Excellence approaches, and designing automation that remains maintainable as adoption grows.

Microsoft 365 Copilot

Helping organizations turn Microsoft 365 Copilot from an individual productivity tool into part of a broader enterprise AI transformation.

Topics include adoption strategies, use-case discovery, Copilot extensibility, agents, enterprise search and knowledge, governance, information readiness, security, and measuring business value. I’m especially interested in how Microsoft 365 Copilot connects employees, organizational knowledge, business processes, and specialized agents to create new ways of working across the enterprise.

Azure AI Services

Integrating Azure AI capabilities into real-world business processes and enterprise applications.

This covers language, speech, vision, document processing, search, content understanding, and other cognitive capabilities, as well as how they can complement generative AI and agentic architectures.

The emphasis is on composable solutions: combining AI services, enterprise data, APIs, automation, agents, and existing applications rather than treating AI as an isolated technology. Architecture, security, scalability, responsible AI, observability, and operational cost are important parts of that conversation.

Dynamics 365

Transforming CRM, customer service, and contact centre operations with AI, automation, Copilot, and agents.

My focus includes Dynamics 365 Customer Service, AI-supported service processes, conversational and voice experiences, case and knowledge automation, intelligent routing, agent assistance, and automation across the customer journey.

A key theme is connecting Dynamics 365 with Copilot Studio, Power Platform, Azure AI, Microsoft 365, and enterprise systems to create end-to-end processes where humans, copilots, automation, and autonomous agents can work together.

The common thread

Across all six areas, the question I’m interested in is increasingly the same:

How do we move from isolated AI demos and automations to a governed, scalable enterprise capability where people, copilots, agents, data, and business processes work together?

That means looking beyond the technology itself and addressing the things that ultimately determine whether enterprise AI succeeds: architecture, governance, security, observability, FinOps, lifecycle management, adoption, and measurable business value.

Tools I build

From the stage to the tenant.

MATE365 — Tenant Manager

Automate workshop, lab, and development tenants for Power Platform, Copilot Studio & AI. Pre-provisioned environments per participant, governed self-service for dev teams, deployed to your own Azure subscription.

mate365.app · See it in the labs

Bridge365 — Cross-device sharing

Move clipboard text, screenshots, and files directly between Windows 11 and iOS. Transfer over your local network or Tailscale, with no transfer cloud or extra account.

bridge365.app

costs365 — Agent cost visibility

Being developed to bring Microsoft agent licenses, credits, tokens and capacity into role-aware cost views for individuals, teams and tenants.

In development. The synthetic demonstrator has no real integrations and does not accept real tenant data.

Planned delivery options are hosted SaaS at app.costs365.app on Azure or deployment in your own Azure tenant through Azure Marketplace once ready. Neither offer is available yet; no launch date or final pricing is confirmed.

costs365.app · Explore the synthetic demo

Speaking & architecture

Let's make it practical.

For conference sessions, workshops, or conversations about enterprise AI architecture, send me a note.

Bring real-world experience to your event.

Sessions can be adapted to your audience, event format, and technical depth.

  • Conferences, meetups, and corporate events
  • Sessions in English and German
  • Practical workshops and architecture discussions

Your request is sent through Formspree. Read the privacy information, or email contact@holgerimbery.blog directly.