Construction – Design-Build Firm (Mountain West): Preconstruction & Subcontractor Bid AI

Industry: design-build commercial. Size: 190 staff. Complexity: fast-track schedules, union packages. Systems: ACC, Excel legacy macros.

Manual RFP reading, inconsistent sub quotes, hidden scope gaps. Bottleneck at bid leveling and risk surfacing.

RFP Intelligence pipeline, Sub Outreach Agent, Bid Normalization Agent, Risk Scoring Agent, Executive Brief Agent.

1. RFP upload/email.
2. AI extracts divisions and alternates.
3. Canonical scope object.
4. Outreach sequences to qualified subs.
5. Responses normalized to grid.
6. Leadership go/no-go with scenario toggles.
7. Win/loss feeds estimator models.

RFP Parser Agent, Sub Matcher Agent, Leveling Agent, Risk Narrative Agent.

Node orchestration, Postgres, doc pipeline, LLM with table extraction, email/SMS gateways.

Autodesk Construction Cloud, Microsoft 365, PostgreSQL, OpenAI, Twilio, Power BI

Bid preparation hours −47%; subcontractor response rate +29%; estimate error variance −18%; executive review time −63%.

Build: 9 weeks. Deploy: 2 weeks. Optimization: 7 weeks bid-leveling models.