Construction AI agent · Tender and estimating
Tender & Estimating AI Agent
A lodged tender is a binding offer, and an estimating agent that sees a rival's pricing is a cartel problem. The agent measures and reconciles; your estimator prices, certifies and lodges.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Estimator signs
- Estimate only
- After launch
- Agent Care
What this agent does
Measures the work; it does not price the bid
In
- 01
Extract quantities from the nominated drawing set into a draft takeoff, tied to the revision measured.
- 02
Diff two drawing revisions and list only the changes that move a quantity.
Reason
- 03
Reconcile received subcontractor quotes against the scope schedule and list what nobody has covered.
- 04
Extract each rate with its source and effective date, and the determination the project runs under.
- 05
Diff the working estimate against the estimator's last approved total, itemised, before anything leaves the office.
Decide
- 06
Reconcile the assembled pack against the ITT compliance checklist and report what is missing.
- 07
Route the estimate, its qualifications register and open assumptions to the estimator.
Out
- 08
Retain the takeoff, the quotes, the assumptions and the version each of them was approved at.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent measures and reconciles; the estimator who signs prices, certifies and lodges — after lodgement the route is a formal mistake claim, with the bid guarantee and the next responsibility determination riding on it.
Example workflow
One package, drawings to sign-off
- 1Drawing set receivedITT documents, drawings, specification and the scope schedule
- 2Quantities extractedMeasured quantities, the revision they came from, received quotes and every rate source with its date
- 3Estimate assembledDraft rates, open assumptions, the qualifications register and confidence
- 4Controls appliedTenant-partition checks, rate-source and wage-determination checks, assumption tripwires and confidence threshold
No human action required
Stages 1 to 4 run unaided, and nothing is priced or lodged at any of them — the agent is measuring, and the estimator's lane opens at the confidence gate.
Goes to the estimator to price.
Adds a commercial read first.
Estimator sign-off
The estimate is held with its takeoff, its open assumptions and the confidence.
Price · Amend · Send to commercial review
Priced — released to the signatory
- 6Estimating systems updatedOnly where write access and approval policy allow it
- 7Outcome evaluatedAmendments made, assumption outcomes, quote coverage and post-lodgement corrections
Amendments
Every estimator amendment is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Setting or approving the final tender price.
- Signing, submitting or lodging the tender.
- Signing an independent-pricing or anti-collusion certificate.
- Adding, removing or rewording a qualification, exclusion or provisional sum.
Automation boundary
Agent acts unaided
- ✓Extract quantities from the nominated drawing set into a draft takeoff.
- ✓Reconcile quotes against the scope schedule.
- ✓Diff two revisions and list what moves a quantity.
- ✓Assemble the pack against the ITT checklist and report what is missing.
Any write happens inside the boundaries agreed at implementation, and never onto a lodged bid.
- Approving a bid bond or its penal sum.
- Withdrawing or correcting a lodged bid.
- Making or varying a design or engineering assumption.
- Communicating with a competing bidder.
Example output
One estimate line, annotated
Everything the agent measures is attached to the drawing it was taken from.
Estimating output · single line item
Illustrative example- Package
- Structural steel
- Measured item
- Edge beams to level three, measured off the revision the ITT nominates
- Quantity
- 18.4 tonnes
- Rate source
- Subcontract quote received
- Confidence
- 93%
- Label carried
- Estimate — not a quote or offer
As received
Taken from the nominated drawing set and the quotes on file — nothing here is written by the agent.
What was read
- Nominated drawing revision
- Scope schedule
- Subcontractor quote
Why this rate source
A received quote covers this scope, so nothing here is inferred.
What the score decides
Below the threshold the estimate picks up a commercial read before the estimator.
Action
PriceAmendSend to commercial reviewValue
Where AI adds value
The same four claims, placed at the point in the workflow where each one applies.
Where the value lands
Value 01 – 04
Every package
From your own tender files
Measure inside one bidder's tenant
Drawings, quotes and rate libraries are partitioned to the bidding entity, and each context assembly is logged against that partition.
Price the package, not the risk
Routine takeoff and quote reconciliation are done before the estimator opens the file.
Send review where the exposure sits
Assumption-dependent items and low-confidence rates are marked, so the estimator's read starts where risk concentrates.
Build an evidence trail
The takeoff, the quotes behind it and the sign-off stay attached to the tender.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Estimating and takeoff
- RIB CostX
- Buildsoft Cubit
- Trimble On-Screen Takeoff
- InEight
Drawings and markup
- Bluebeam Revu
- Autodesk Construction Cloud
Bid and tender management
- Autodesk BuildingConnected
- EstimateOne
- SmartBid
- Bidtracer
Agent
Tender & estimating
- Reads the drawings
- Builds the takeoff
- Holds for the estimator
Cost data and portals
- RSMeans
- Rawlinsons
- Wage determinations
- ITT portals
Observability & evaluation
- OpenTelemetry
- Langfuse
- Supported monitoring/evaluation sources
Integration availability depends on the client's existing systems and API access.
Agent controls
Six layers between the model and the tender
Each control wraps the one inside it. What a layer does not catch is named in the failure-mode map below it.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether an estimate may stand
- L3
- Puts the release in the estimator's hands
- L4 – L5
- Keep the record and the version honest
- L6
- Pulls the agent back when signals degrade
How Nestack evaluates it
Evaluate the estimating workflow — not only the final total.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the number the tender carries
- E1Final-output evaluationDid every quantity and rate match the revision and the quote it cites?
- E2Step-level evaluationDid the agent use the nominated drawings, the right quotes and the right rate library?
- E3Tool evaluationDid it read and write the correct estimate file and the correct line?
- E4Confidence calibrationDo low-confidence lines actually attract more estimator amendments?
- E5Slice evaluationHow does performance change across specific tender types?
- E6Business outcomeHow many estimates needed an amendment, or a correction after lodgement?
Floor — the price the business is bound to
Failure modes
Where each failure originates in the agent
Seven failure modes, mapped to the stage that produces each one.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- TE-01
Superseded input priced
A withdrawn revision or an expired wage determination is read as current.
Stage gathers
Drawings, quotes, rate libraries and the ITT documents
02 · Reasoning
2 modes
- TE-02
Assembly double-counted
A model quantity and a manual adjustment both carry the same assembly.
- TE-03
Open item priced as firm
A provisional sum or qualified exclusion is carried as a fixed rate.
Stage proposes
Takeoff, rates, open assumptions and confidence
03 · Tool / write
2 modes
- TE-04
Rate changed after sign-off
A rate is written into the live estimate once the total is already approved.
- TE-05
Portal form auto-filled
A submission form carrying the independent-pricing certificate is completed.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- TE-06
Inferred rate dressed as a quote
An estimated rate is formatted exactly like a received subcontract quote.
Stage returns
The estimate the estimator prices and the tender carries
05 · Change / Version
1 mode
- TE-07
Rounding defaults shifted
Waste factors change mid-cycle, so packages price on different bases.
Stage tracks
Model, prompt, rate library and rounding configuration
- Sev-1 · a held act, or competitor pricing seen
- Sev-2 · a wrong number reaches the tender
- Sev-3 · an input degrades, estimate held
Affected slices
One tender type can carry the whole number
Post-lodgement price-integrity failures are counted against an all-tender baseline, because a lodged price is an offer and the correction route is a formal mistake claim rather than an edit.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Shared-tenant estimating | 4.5% | 3.5× | Review | |
| Publicly funded bonded tenders | 2.7% | 2.1× | Review | |
| Design-and-construct tenders | 1.8% | 1.4× | Watch | |
| Documented BOQ negotiated work | 0.9% | 0.7× | Normal |
Bar: price-integrity-failure lift vs. an all-tender baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
Closing means the suite got longer
Nothing closes on an explanation. The cycle ends when the failure exists as a regression case the next release has to pass.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Price-integrity failures rise in a tender slice.
- 02
Diagnose
The estimator reads the takeoff, the quotes and the version history until the cause is one of them.
- 03
Improve
The fix is written against a version, with the estimates that motivated it attached.
- 04
Verify
Release is blocked until the affected regression cases pass again.
- 05
Learn
The case joins the permanent suite and the estimating standards.
Learn → Detect
The return edge. Every following cycle starts against a longer suite than this one.
Typical build scope
Twelve workstreams across six weeks
The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, sources, estimating workflow, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Tender workflow discovery and boundary definition.Week 1
- 02Drawing and quote source assessment.Week 1, Week 2
- 03Rate-source, wage-determination and partition mapping.Week 1, Week 2
- 04Drawing ingestion and quantity normalisation.Week 2
- 05Takeoff logic and quote reconciliation.Week 2, Week 3
- 06Confidence scoring and assumption routing.Week 3
- 07Estimator sign-off workflow.Week 3, Week 4
- 08Estimating and bid-system integration.Week 5
- 09Evaluation suite and regression cases.Week 4, Week 5
- 10Guardrails and approval controls.Week 4
- 11Observability and trace instrumentation.Week 5, Week 6
- 12Deployment, documentation and Agent Care handover.Week 6
12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallel
Final scope and sequence confirmed in discovery
Engagement tiers
What each tier includes
Rows are the capabilities named in each tier's scope. Higher tiers include everything below them.
| Capability✓ in scope · — not at this tier | PilotOne package, one bidding entity | ProductionProduction estimating systems | AdvancedMultiple entities / regions |
|---|---|---|---|
| Introduced at Pilot | |||
| Takeoff and quote reconciliation | ✓ | ✓ | ✓ |
| Estimator sign-off | ✓ | ✓ | ✓ |
| Baseline evaluation | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Package reporting by trade | — | ✓ | ✓ |
| Review workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Bid-history analytics feed | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-region rate libraries | — | — | ✓ |
| Multi-stage commercial review | — | — | ✓ |
| High tender volume | — | — | ✓ |
| Enterprise tender controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on integrations, workflow complexity, tender volume, approval controls and deployment requirements.
- Separate from build
- Build pricing is separate from recurring Agent Care, which covers managed monitoring, evaluations, incidents and verified improvements after launch.
What we need from you
What you bring, and what we build with it
Each input maps to a piece of build scope and a week in the delivery timeline.
You bring
We build with it
01Your scope schedules and rate libraries
→Rate-source mapping and tenant partition setup
Week 102Representative priced tenders
→Takeoff baseline, quote reconciliation and assumption tripwires
Week 203Your qualification and exclusion standards
→Qualifications-register mapping and automation-boundary definition
Week 104Access to relevant APIs, feeds or exports
→Drawing, quote and rate-source assessment, then integration setup
Week 205Examples of tenders that had to be corrected
→Evaluation suite, regression cases and failure-mode testing
Week 406Your review thresholds and write-back boundaries
→Confidence scoring, assumption routing, guardrails and approval controls
Week 307Named estimators to review the drafts
→Estimator sign-off workflow, then pilot and production validation
Weeks 5–6
Nothing else is required
Deployment, documentation and Agent Care handover are ours.
Delivery timeline
Four phases across six weeks
Each phase covers the weeks it genuinely occupies. Week 5 carries evaluation and pilot together.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Tender workflow discovery, rate mapping and the automation boundary
W2Source integration and the takeoff baseline
W3Quote reconciliation, confidence logic and sign-off controls
W4Evaluation suite, guardrails and failure-mode testing
W5Estimating-system integration, pilot packages and targeted corrections
W6A live tender assembled under the estimator, then Agent Care handover
- Reading the band
- A bar covers only the weeks its work is named in. The week 5 overlap is real, not padding.
- At the end of W6
- Validation closes on a real submission, and Agent Care picks up monitoring.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Construction AI agent
Build a tender agent around the estimator who signs.
Show us your drawing sets, your quote flow and who signs the tender. Partition estimating data per bidding entity first — Australian cartel maximums have doubled — then name what the agent hands the estimator and what it never touches.
- Typical delivery
- 4–6 weeks
- Deployment
- Your estimating stack
- Estimator signs
- Estimate only
- After launch
- Agent Care available
Nestack Agents · Tender and estimating
AGT-CN-01 · Agent Care available after launch