Everything in Trueleveler, on one page.

Feeds on your scope narrative, specification sections, or trade package and produces a clean RFQ with line items, units, evaluation criteria, instructions to bidders, and a blank bid form. What would take an estimator 40 minutes per package gets compressed to a 2-minute review-and-send.

Normalizes units across quotes (FT/M, LB/KG, EA/PR), flags per-line outliers beyond ±25%, detects labor/material imbalance and mobilization front-loading, scores each vendor on reliability from past projects, and produces a leveled bid tab with a clear recommendation. Upload 2-10 quotes plus the RFQ and get a one-page award recommendation.

Reads your specification or scope document and matches every requirement line-by-line against a vendor's bid response. Catches silent exclusions, missing allowances, and buried qualifications that create disputes in month three. Returns a coverage percentage, a critical-gap list, and a bid/no-bid recommendation with win probability.

Reads AIA A201/A401, ConsensusDocs, NS 8405/8407, VOB/B, JCT, NEC4, and custom subcontracts. Extracts every clause with a risk score, flags pay-if-paid, liquidated damages, consequential damages waivers, termination-for-convenience, indemnification breadth, and flow-down provisions. Produces a jurisdiction-aware redline with specific notes for each risk. Not legal advice, but a senior-PM-grade first pass.

Identifies each distinct trade discipline in the supplied spec text and writes a procurement-ready scope. Grouped by trade with CSI MasterFormat division mapping. Inclusions, exclusions, assumptions, and qualifications are each grounded in the spec — no invented requirements. Persists each scope to the scopes table for downstream RFQ Generator consumption.

Parses specification sections and extracts every submittal requirement — product data sheets, samples, shop drawings, test reports, closeout documents — with responsible party, review type (action/informational), and CSI division. Auto-imports into the Submittal Tracker so your log starts day one, not week four.

Compresses a 40-80 page subcontract into a 2-page brief your field PM can actually use. Payment terms, retainage, LDs, insurance requirements, claims notice periods, dispute resolution, change-order procedure — each surfaced with the clause reference so a PM knows where to look when something goes sideways.

The estimator-to-PM transition is where 60% of cost overruns begin. This engine auto-pulls the latest Contract Review and Bid Scope Compliance results from the active project (no upload step) and assembles a handoff package: key contacts, risk register, scope clarifications, notice requirements, and the decisions the PM needs to make in the first two weeks. Run Contract Review first for the full risk + obligations picture; Bid Scope Compliance is optional but adds the gap analysis section.

The 2am engine. When a CO lands on your desk with a 48-hour deadline, this engine cross-references the original contract to determine if the requested work was already in scope, analyzes the pricing against typical unit rates, flags manipulation patterns (bundled items, inflated overhead, incomplete backup), and produces a verdict with a recommended amount. Pair with Contract Review for full context.

Parses unstructured daily reports (emails, typed notes, photos with captions) and structures them into a consistent format: manpower by trade, work completed by location, deliveries received, safety observations, weather delays, and tomorrow's plan. Searchable across weeks so you can reconstruct what happened when a dispute arises six months later.

Detects when a spec section disagrees with a drawing detail or another spec section, drafts the question, attaches the conflicting evidence, models cost + schedule exposure, and routes to the right architect/engineer with a CC list. Confidence-gated — fields below 70% confidence blank with a "?" prompt instead of guessing.

v1 (text-only): compares pasted submittal text to spec text and produces a numbered list of deviations with severity (critical/major/minor/info), category, side-by-side quote pair, recommended action per finding, and an overall review_result (Approved / Approved as Noted / Revise and Resubmit / Rejected). Persists deviation_findings + review_result onto submittal_revisions. v2 will add drawing-bbox pin overlays when the PDF parsing sidecar ships.

Records or accepts uploaded audio of any meeting, performs ASR via Gemini 2.5 Flash, runs speaker diarization, then extracts a structured minutes document: attendees with talk-time share, topic-grouped summary, action items with assignee + due, decisions with rationale, and risks with severity. Every line carries an audio anchor (▶ MM:SS) that scrubs the recording to the exact moment. Action items auto-promote into the Tasks tracker with source = meeting. Audio retained 18 months by default.

For mid-size GCs sharing cranes, lifts, and hoists across active projects. Reads every project's equipment reservations, detects overlapping demand on the same asset, computes per-asset utilization, and proposes re-sequencing. Pairs with Schedule Risk to surface critical-path impacts.

Pulls today's voice notes, project_media photos with AI captions, procurement deliveries, RFIs created/answered, schedule activities in progress, and submitted change orders, then composes a structured daily report (work performed, manpower, deliveries, safety, issues, weather, follow-ups). Zero new schema — reuses daily_log_entries, voice_recordings, project_media.

Takes a voice recording from a closeout walk + the photos taken on the walk, extracts each distinct deficiency into a structured punch item (title, location, trade, priority, target close date, photo evidence, quote from walk). Trigram-similarity dedupe against existing open items prevents the engine from spamming duplicates.

The clock-protection engine. When an event triggers a notice obligation, Notice Builder reads your contract's notice provisions, identifies which formal notice is required (delay, differing site condition, change directive, claim preservation, cure, termination), and drafts a compliant letter citing the governing clause, the deadline / time-bar window, the required recipients and delivery method, and the supporting documentation to attach. A missed notice deadline is the most common way a legitimate claim gets time-barred — this gets the right notice out, correctly, before the clock runs. Pair with Contract Review for clause context and Change Order Review for pricing-backed claims.

When the vendor's PO arrives, does it actually match what they quoted? This engine catches unit-price creep, quantity mismatches, changed payment terms, and silent scope substitutions that show up between quote and PO. Most procurement teams find 1 in 7 POs has at least one meaningful discrepancy.

Run at the end of the job (or on each progress invoice for longer projects). Reconciles what the vendor billed against what they quoted, flags unit-price walk-ups, unauthorized line items, and double-billed materials. Especially powerful across a portfolio — teams who run this on every invoice recover 2-4% of subcontractor spend in year one.

Catches partial-delivery overbilling, back-charges that weren't approved, freight/handling line items absent from the PO, and quantity mismatches. Pairs especially well with the Procurement Tracker for teams running large material orders across multiple deliveries.

LLM-driven audit that produces a senior-PM narrative read on a pay application. Analyzes progress billing for ≥25% single-period jumps, stored-materials over 15% without documentation, front-loaded unit prices, retainage release errors (the most common dispute), schedule-of-values drift from contract, prompt-payment act compliance, back-charge notice requirements, and lien-waiver conditional/unconditional classification. Companion to the deterministic Pay App Review engine (field phase) — use this for judgment calls, use Pay App Review for line-by-line reconciliation.

Compiles an executive owner-meeting summary — schedule status, budget burn vs baseline, approved/pending COs, open RFIs, safety incidents, and the top 3 risks needing owner attention this period — from the contract/scope and latest progress documents you provide.

v1 (text-based): parses a subcontractor pay application (AIA G702/G703 format) into structured line items, reconciles each line against the project's Schedule of Values, cross-checks the Procurement tracker for billed-but-not-delivered material, flags math errors / over-billing / retainage drift, and recommends an overall review_result (Approve / Approve with conditions / Dispute / Reject). Persists pay_app_reviews + an engine_runs receipt row. v2 will add PDF bbox anchors for click-to-jump in the Document Viewer.

Aggregates the project's warranties (from Warranty Compiler), as-built redlines (from Asbuilt Reconciler), punch closeout state (from Punch List Drafter), and O&M documents into a structured owner-ready deliverable. Computes a completeness score and surfaces gaps before handover. v1 ships the structured package; v1.1 ships the PDF/ZIP export.

Aggregates approved submittal_revisions + procurement_items for the project, then composes one warranty entry per manufactured product with term_months (spec-stated, manufacturer-published, or industry-standard), start date aligned to substantial completion, conditions, and exclusions. Persists each entry to the warranties table for the closeout package. Flags unmatched submittals for PM follow-up.

v1 (text-only): compares construction-set notes against field-marked as-built notes and produces a structured redline log with delta_type (dimension/equipment/route/elevation/quantity/material/other), severity, side-by-side spec ↔ as-built values, location, and rationale. Persists each delta into asbuilt_redlines for the closeout package. v2 will add drawing-image diff via CV pipeline.

Runs CPM forward/backward pass over schedule_items, then samples activity durations from a triangular distribution driven by real risk data (late procurement, stale RFIs, pending COs, overdue submittals). Returns p50/p90 completion bands, named drivers, and ranked pull-forward actions.

After substantial completion, reconciles every bid line item against actual PO + CO data and produces the historical learning report: which trades came in over, which estimating ratios were wrong, which subs delivered to bid. Feeds your future estimate library.