- Do I need my own API key?
- No. AI is included in your plan and metered as credits, so every seat has the agent from day one. If you would rather bring your own Anthropic key, you can, and the usage bills to your account instead of your credits.
- How does the AI know about my project?
- Every record you create or upload — drawings, specifications, contracts, RFIs, diaries, applications — is read, chunked into meaningful sections and embedded in your own tenant database. A question is answered by searching those records by meaning, and the answer cites the clause, sheet or record it came from.
- Which agents can I use?
- Three. The Construction AI agent inside the platform, docked in every module with that module’s tools. Claude, on the desktop or the web, connected over MCP. And Microsoft 365 Copilot, the same way. All three see the same records and the same tools, under your permissions. Managed agents also run on a schedule inside the platform: the brief, the project knowledge, the weekly summary and each person’s briefing.
- Can subcontractors use the AI?
- Subcontractors on the platform only see data scoped to their cost codes, and the AI respects the same boundary. A subcontractor asking a question gets answers from their own packages, never from another trade’s or the project’s commercial position.
- Is my data private?
- Every organisation’s data lives in its own database with full tenant isolation; nothing crosses between tenants. Your records go to the model only to answer your request and are never used to train it. Every AI call is logged in your usage ledger, so you can see what was asked and what it cost.
- What if the AI service goes down?
- A circuit breaker is built in. If the model is unavailable or rate-limited, the AI features fail gracefully and the rest of the platform carries on. Nothing you have recorded depends on the model being up.