Two ways data reaches the portfolio
Automated connectors authenticate with the vendor's admin or analytics API and sync usage, seats, and where the vendor exposes it, spend, on a schedule. Six are implemented today and all six are in beta, enabled per workspace during onboarding rather than self-serve.
Manual entry and CSV import cover everything else: license terms and cost for any provider, plus usage or cost files exported from tools that have no connector. This is a first-class path, not a fallback — a portfolio that only counts automated platforms is not the real portfolio.
Every figure in Midgentic carries its source label, so an automated sync, a CSV import, and a manually entered assumption are never blended into one undifferentiated number.
Platforms Midgentic syncs from directly
Each of these has a working connector in the product today. Capability lists reflect what Midgentic actually ingests, not what the vendor API might theoretically allow.
Microsoft 365 Copilot
Automated connector — betaCopilot seats, active users, per-app activity across Teams, Word, Excel, PowerPoint, Outlook, OneNote and Loop, and usage trends from the Microsoft Graph reports API.
Read more
GitHub Copilot
Automated connector — betaSeat assignment and developer activity for your GitHub organization, with usage trends and identity resolution for engineering adoption.
Read more
Google Workspace with Gemini
Automated connector — betaGemini licenses and Workspace activity from the Google Admin SDK, with identity resolution and department attribution.
Read more
Claude Enterprise
Automated connector — betaEmployee adoption on claude.ai: seats, daily, weekly and monthly active users, per-person activity, and feature usage from the Anthropic Enterprise Analytics API.
Read more
OpenAI API Platform
Automated connector — betaMetered API consumption and spend for your OpenAI organization, broken down by model and project.
Read more
Anthropic API Platform
Automated connector — betaAPI consumption, metered spend, workspace activity, and Claude Code developer analytics from your Anthropic Console organization.
Read more
All automated connectors are currently in beta: implemented and syncing real vendor data, validated with early customers, and switched on per workspace during onboarding rather than self-serve.
Everything else in the portfolio, represented honestly
Most organizations run more AI than any set of connectors can reach: departmental subscriptions, vertical AI features inside business applications, internal tools built on model APIs, and vendors whose analytics APIs are not open. Those belong in the portfolio too.
For any provider in the catalog you can:
- Enter license terms
- Contracted seats, price, currency, and effective dates per product or SKU, so historical periods stay accurate when a contract changes.
- Import usage or cost by CSV
- Bring in an export from a vendor console or finance system and map it once.
- Record value assumptions and outcomes
- Document the assumptions behind estimated value, and record measured outcomes separately.
- See coverage gaps explicitly
- Where a provider has no data yet, Midgentic says so instead of implying completeness.
Examples of providers currently tracked this way in the catalog: ChatGPT Enterprise, Cursor, Atlassian Rovo, AWS Bedrock, Databricks Mosaic AI, and others across assistants, development, collaboration, and infrastructure.
Integrations are a means to an intelligence layer
A long connector list is not the point. The reason to connect platforms at all is to answer questions that no single vendor console can: what AI do we actually run, who genuinely uses it, what does it cost against that usage, and is it worth what we pay. The category those questions belong to is enterprise AI intelligence.
Common questions
Which AI platforms does Midgentic connect to automatically?
Six automated connectors are currently implemented and in beta: Microsoft 365 Copilot, GitHub Copilot, Google Workspace with Gemini, Claude Enterprise, the OpenAI API Platform, and the Anthropic API Platform. Beta connectors are enabled per workspace during onboarding.
What happens to AI tools with no automated connector?
They are still part of the portfolio. License terms, cost, seats, and outcomes can be entered manually, and usage or cost data can be brought in by CSV import. Every metric is labeled with its source, so a manual figure is never presented as an automated sync.
What does 'beta' mean for a Midgentic connector?
The connector is implemented and syncing real vendor data, and it is being validated with early customers. Coverage and vendor API behavior can still change, and connectors are switched on per workspace rather than being self-serve for everyone.
Does Midgentic read the contents of AI conversations?
No. Midgentic works with portfolio-level signals — activity, seats, usage counts, and vendor-reported spend. It does not inspect prompts, responses, or documents, and it is not a DLP or content-monitoring tool.