Gartner on AI in EA

Five AI capabilities on a model agents can use

AI on a messy repository is garbage in, garbage out. Gartner names five emerging AI capabilities for enterprise architecture tools. Caplo keeps the model those capabilities read from: a maintained repository, a draft for each proposal, version history, and a canvas.

This page is Caplo's mapping of those capabilities. It is not a Gartner endorsement.

Read the Gartner research: Top 5 Emerging AI Capabilities in Enterprise Architecture Tools

  1. Conversational Interfaces
  2. AI Interoperability
  3. Generation and Discovery
  4. Content Enrichment
  5. Decision Support

Five capabilities, grounded in the model

Each name is how we read the public capability. The foundations under it are what the answer depends on.

  1. 01

    Conversational interfaces

    The in-product assistant sits on the diagram. MCP clients such as Cursor and Claude talk to the same workspace.

    • Repository
    • Canvas

  2. 02

    AI interoperability

    One hosted MCP server and one repository, so outside agents and people work on the same model. Writes from those clients go into a draft.

    • Repository
    • Drafts

  3. 03

    Generation and discovery

    Agents search the repository, follow relations, walk a graph neighborhood, and load BPMN. The assistant can draft and explain diagrams. Generated changes stay in a draft for review.

    • Repository
    • Canvas
    • Drafts

  4. 04

    Content enrichment

    Typed properties, relations, and maintained descriptions are what enrichment lands on. Agents propose missing relations, duplicates, and stale text in a draft.

    • Repository
    • Drafts
    • Versioning

  5. 05

    Decision support

    Impact questions run against the live model or a named scenario draft. The canvas shows the dependency. History and the audit trail keep the recommendation explainable.

    • Canvas
    • Drafts
    • Versioning

How Caplo supports them

A maintained repository, drafts you review before publish, version history, and a canvas.

Foundation

A repository agents can read

Shared entities and relations, maintained by the people who own them. Agents read that model directly: search, relations, neighborhoods, BPMN, and the property shapes this workspace uses. A messy repository produces messy answers, so it has to stay easy to maintain.

  • One repository for architects and for agents.
  • Relations and neighborhoods answer impact questions from the model.
  • Property shapes keep new fields consistent with how the team already models.
  • Conversational interfaces
  • AI interoperability
  • Generation and discovery
  • Content enrichment

Example view

One repository

Entities, relations, and property shapes the team already maintains.

  • Policy adminApplication
  • Claims handlingProcess
  • Policy recordData object

Serves · Policy admin → Claims handling

An agent reads the same entities and relations the architects maintain.

Foundation

Versioning and an audit trail

Per-item history, workspace snapshots, and an audit trail record creates, updates, deletes, and restores. Agent writes are tagged to the draft they belong to, the same way as an edit in Caplo.

  • History shows what changed and who changed it. You can restore a previous version.
  • Snapshots keep the workspace as it was at a point in time.
  • The audit trail tags agent writes to their draft.
  • Content enrichment
  • Decision support

Example view

On the record

The same trail for an edit in Caplo and a write from an agent.

  • History

    What changed, and who changed it

  • Snapshot

    The workspace as it was on a given date

  • Audit

    Create, update, delete, and restore. Tagged to the draft.

History, a snapshot, and the audit trail stay attached to the change.

Foundation

Drafts for collaboration and conflict resolution

A draft is an overlay. People and agents propose there, and publish stays with a person. Live edits the draft has not touched stay visible in the draft. When the draft and live both change the same item, Caplo surfaces a conflict.

  • Named drafts for a retirement, a cleanup, or a target state.
  • Agents create or reuse a draft and return a review link.
  • Overlapping edits surface as conflicts before publish.
  • AI interoperability
  • Generation and discovery
  • Content enrichment
  • Decision support

Example view

Propose, then publish

The draft keeps only what changed. Live stays in view.

Draft · Retire back-office suite

  • Updated Application · phase out
  • Policy adminConflict
  • Live description updated · overlay still wins until publish

The overlay holds the proposal. A conflict appears when live changed the same item.

Foundation

A 2D canvas for diagrams

Inside Caplo, the diagram canvas is where the assistant drafts, explains, and reworks a view of repository entities. Outside Caplo, agents load the same structure as a graph neighborhood and as BPMN, and get a link back to the diagram.

  • The assistant drafts, explains, and reworks diagrams from the side panel.
  • Agents load a graph neighborhood around the entities you ask about.
  • BPMN reads and writes update the process diagram and return a link into Caplo.
  • Conversational interfaces
  • Generation and discovery
  • Decision support

Example view

See the dependency

A diagram in Caplo. A neighborhood for the agent.

Canvas

  • CRM
  • Billing
  • Analytics

Neighborhood

CRM serves onboarding. Billing reads the policy record. Analytics consumes both.

The canvas shows the dependency, and the neighborhood gives an agent that same structure.

Connect an agent

Point a client at the same repository

Cursor, Claude, and your own agents read the workspace and write into a draft. People review and publish.

For AI agents