From Clicks to Code: Reimagining Dataverse with Coding Agents

Helping organisations become intelligent enterprises requires more than adopting new AI tools. It requires modern applications, connected business systems, trusted data, and a strong architectural foundation that allows technology to evolve with changing business needs.

That is why From Clicks to Code: Reimagining Dataverse with Coding Agents was one of the most interesting sessions at Microsoft Integrate 2026. Through a series of demonstrations involving coding agents, natural language driven development and Dataverse Model Context Protocol (MCP) capabilities, Kent Weare provided a glimpse into how Microsoft is preparing enterprise platforms for the agent era.

As someone working across Azure Integration Services, Power Platform and enterprise architecture, I was particularly interested because the session brought together several themes that are becoming increasingly important: AI agents, Dataverse, MCP and enterprise integration. While the session focused on Dataverse, the broader story is about the future of enterprise architecture, integration, and AI-enabled business systems.

What the Session Demonstrated

One of the strengths of the session was that it moved beyond theory and demonstrated how coding agents can actively participate in the Dataverse development lifecycle.

Coding Agents and Natural Language Development

The demonstrations showed how Dataverse development is evolving from a portal-driven experience towards an agent-assisted experience powered by natural language, coding agents, and MCP.

Instead of manually creating tables, columns, and relationships through the Maker Portal, users could describe a business scenario and allow the agent to generate the underlying Dataverse assets.

What was particularly striking was the speed at which a business idea could be transformed into a working Dataverse structure. Tasks that would traditionally require navigating multiple screens, creating metadata, and configuring relationships could be initiated through a natural language conversation.

The demonstrations suggested a future where developers spend less time performing platform configuration and more time focusing on business outcomes, governance, and solution design. AI can understand business intent, translate it into Dataverse artefacts and while still allowing developers to review and refine the results.

This represents a significant departure from traditional low-code development, where most configurations were performed manually through the user interface.

Dataverse Model Context Protocol (MCP) Plugin

Another important demonstration focused on the Dataverse MCP plugin. The MCP plugin enables coding agents to interact directly with Dataverse capabilities through a standard protocol rather than relying on custom integrations.

From a developer perspective, this allows agents to:

• Discover Dataverse metadata

• Create and modify tables

• Understand relationships

• Generate Dataverse assets

• Interact with the platform through natural language instructions

The significance of the MCP plugin extends beyond Dataverse development itself. By exposing Dataverse capabilities through a standard protocol, Microsoft is enabling a growing ecosystem of AI tools, copilots and coding agents to interact with business platforms in a consistent manner.

What I found most interesting

From Platform Expertise to Agent Expertise

One aspect of the session that stood out was not the AI-generated code itself, but the way Microsoft is beginning to package platform expertise into reusable agent capabilities.

Rather than requiring developers to understand every Dataverse capability, the agent determines:

• Which tools to use?

• Which APIs to call?

• Which scripts to generate?

• Which sequence of actions should be performed?

This feels less like AI-assisted coding and more like AI-assisted platform expertise.

The End of ClickOps

Something else that resonated with me was Kent’s reference to “ClickOps”.

Many architects and developers have spent years moving away from portal-driven administration towards Infrastructure as Code, source control and automated deployments.

What the session demonstrated was an evolution beyond both traditional coding and traditional low-code development. Rather than manually creating Dataverse assets through user interfaces, developers increasingly describe business intent and allow AI agents to generate the implementation details.

This represents a shift from ClickOps towards intent driven development.

Key Takeaways

The Bigger Microsoft Story

While the session focused on Dataverse, one of my key observations was that the surrounding Microsoft announcements reveal a much larger strategic direction.

Historically Microsoft built ecosystems around:

Applications >> Connectors >> APIs

Today Microsoft appears to be building the next layer:

Applications >> Connectors >> APIs >> MCP >> Agents

Recent announcements around MCP certification, tool registries and governance capabilities suggest Microsoft is thinking beyond individual agents and towards an enterprise-scale ecosystem.

While still early, the direction increasingly resembles a governed platform for discovering, managing and operating AI capabilities.

A Thought That Stayed with Me: Data Is the Moat, MCP Is the Access Layer

Much of the current AI conversation focuses on models, reasoning capabilities, coding agents, and agent frameworks.

However, a growing number of industry discussions are reaching a similar conclusion: as AI capabilities become more widely available, differentiation increasingly comes from the quality of an organisation's data, processes and business context.

Viewed through this lens, the Dataverse MCP Plugin becomes much more interesting. At first glance it appears to be primarily a developer productivity feature, but I believe it is much more significant than that.

Microsoft is not simply helping agents access data. Microsoft is enabling agents to securely access:

• Business context

• Relationships

• Workflows

• Security models

• Organisational knowledge

In other words, Microsoft is enabling agents to interact with information that differentiates one organisation from another.

The long-term value of AI is unlikely to come from the model alone. It will increasingly depend on the quality of the business data, processes, and organisational knowledge that AI systems can access and act upon.

What This Means for Organisation

The session highlighted that successful adoption of coding agents and AI powered development is not simply about deploying new tools.

Organisations need strong foundations in data modelling, metadata, governance, security and integration architecture before agents can operate effectively.

Some practical actions organisations can take today include:

Improve Metadata Quality

• Agents depend heavily on metadata.

• Poorly named tables and columns will produce poor outcomes.

Strengthen Governance

• Security models, business units, field security and role design become increasingly important when agents gain access to business capabilities.

Invest in Documentation

• Kent repeatedly reinforced the importance of: Plan >> Review >> Execute >> Document

  rather than relying on uncontrolled AI generated development.

Design AI Ready Architectures

• Future projects should consider:

  • MCP

  • Agent access patterns

  • Governance

  • Observability from the beginning.

How Valorem Reply can help 

One reason this topic is particularly relevant to Valorem Reply is that many of the challenges emerging around MCP, agent governance and enterprise AI are fundamentally integration challenges.

Organisations have spent years exposing business capabilities through APIs, integration platforms, and service contracts. MCP appears to be the next layer on top of that foundation.

While many discussions around AI focus on prompts, models and copilots, the real challenge is often enabling AI to interact safely with enterprise systems, business processes, and organisational data. This is precisely where integration architecture becomes critical.

As a Microsoft Solutions Partner with expertise across Azure Integration Services, Data & AI, Power Platform, Business Applications and enterprise architecture, Valorem Reply is uniquely positioned to help organisations prepare for this transition.

Potential areas where we can help include:

• AI Ready Enterprise Architecture
  Establishing governance, security, observability, and architectural foundations for agent-driven solutions.

• API-First and MCP Ready Architectures
  Designing integration strategies that enable business capabilities to be safely consumed by applications today and AI agents.

• Enterprise Data and System Enablement
  Connecting ERP platforms, student systems, CRM solutions, business applications and data platforms to support AI powered experiences.

• MCP Enablement Services
  Helping organisations identify candidate systems, expose business capabilities through MCP, establish governance controls, and prepare internal systems for agent consumption.

• AI Powered Business Solutions

  Designing and implementing solutions that combine Azure, Dataverse, Power Platform, Azure AI Foundry, Copilot Studio, and Integration Services.

Conclusion 

The most significant announcement from this session may not have been the Dataverse MCP plugin itself. The bigger story is Microsoft’s investment in making enterprise platforms consumable by AI agents through standards such as MCP and natural language-driven development experiences.

Dataverse is no longer just a low-code platform. Microsoft is positioning Dataverse as a strategic enterprise data platform for the agent era.

Perhaps the most important takeaway from the session is that the long-term value may not lie in the agent itself, but in the business knowledge, workflows and proprietary data the agent can access.

Microsoft’s investment in Dataverse MCP suggests the company recognizes this shift and is positioning Dataverse as a strategic platform for the next generation of AI powered business solutions.

For organisations that have spent years building API first architectures, the next evolution may be:

API-First Architectures + MCP-Ready Capabilities

As Microsoft continues building its enterprise agent platform, organisations that establish strong governance, metadata standards and well-designed business capabilities today will be best positioned to take advantage of the next generation of AI-driven solutions.

The question may soon shift from:

Do we have APIs?

to

Are our systems agent-ready?