Seat-licensed AI raises seat-licensed questions
- How many Claude Enterprise seats are active weekly rather than merely assigned?
- Is adoption deepening, or concentrated in a small enthusiastic group?
- Which teams rely on it, and which never onboarded?
- What is it costing per active user under our agreement?
- How does it compare with our other assistant licenses — and with what we already spend on the Anthropic API?
The last question is the one most organizations cannot answer, because seat licensing and metered API usage are billed and reported on entirely separate tracks.
What Anthropic provides — and where the boundary sits
Anthropic's Enterprise Analytics API exposes genuine adoption telemetry: seats, daily, weekly and monthly active users, per-user activity, and feature usage. Midgentic reads those endpoints directly and treats them as the authoritative source for Claude Enterprise usage.
What it does not include is your contracted price, your other vendors, or the relationship between employee adoption on claude.ai and the metered API consumption your engineering teams may run under the same vendor relationship. Reconciling that is portfolio work.
What Midgentic actually ingests from Anthropic
- Seat data
- Claude Enterprise seat counts, which form the utilization denominator.
- Active user counts
- Daily, weekly, and monthly active users retained as a historical series.
- Per-user activity
- Individual activity records used for utilization and identity resolution, never conversation content.
- Feature usage signals
- Which capabilities adoption is actually concentrated in.
- Claude Code signals
- Where exposed, developer-side Claude Code activity attributable to the organization.
- License terms (entered by you)
- Seats, price, currency, and effective dates — analytics endpoints do not expose negotiated pricing.
The questions this data answers
- Employee adoption
- Real active users against purchased seats, tracked over time.
- Depth of use
- Whether adoption is habitual or occasional, and which features carry it.
- Vendor-level spend
- Claude Enterprise seat cost and Anthropic API consumption rolled up per vendor without double counting.
- Assistant comparison
- Claude Enterprise set beside other assistant licenses covering similar users.
- Renewal evidence
- Utilization and value assumptions documented before the seat count is renegotiated.
Two Anthropic relationships, one portfolio number
Most organizations that adopt Claude Enterprise also consume the Anthropic API Platform. Midgentic keeps them distinct — one seat-based, one metered — while rolling both into a single vendor view, which is what spend intelligence and ROI reporting require.
Common questions
What Claude Enterprise data does Midgentic collect?
Seat counts, daily, weekly and monthly active user counts, per-user activity records, and feature usage signals from the Anthropic Enterprise Analytics API. It is adoption telemetry, not conversation data.
Does Midgentic read Claude conversations?
No. The connector uses the enterprise analytics endpoints only. Prompts, responses, projects, and uploaded files are never accessed.
How does Midgentic avoid double-counting Claude Enterprise and Anthropic API spend?
They are modeled as separate providers with separate cost bases: Claude Enterprise is seat-licensed, the API platform is metered consumption. Portfolio rollups combine them at the Anthropic vendor level without adding the same money twice, and both remain individually attributable.
Does Claude Enterprise seat cost sync automatically?
No. Seat price comes from license terms you enter — seats, price, currency, and effective dates — which Midgentic reads against active-user data to produce utilization and cost per active user.
Is the Claude Enterprise connector generally available?
It is implemented and in beta, enabled per workspace during onboarding while it is validated with early customers.