Main site

Construction AI agent · Domestic content

Domestic-Content Certification AI Agent

Key each funded item to the requirement its project funding date fixes, hold the manufacturer and supplier evidence behind each certification, and hand the signer the gaps before they certify.

Typical delivery
4–6 weeks
Deployment
Your stack
Signer approval
Estimate cutoff
After launch
Agent Care

What this agent does

Builds the pack, never signs the certification

In

  1. 01

    An item is funded, so it arrives with its project, its funding programme and the purchase order that bought it.

  2. 02

    A cut-off passes, and each item on that project is re-read against the obligation date the customer supplies.

Reason

  1. 03

    An item is classified as iron or steel, a manufactured product or one of the listed construction materials.

  2. 04

    A certificate lands, and the step chain behind it is laid out link by link with the handler each link names.

  3. 05

    A component cost is quoted, and it is read against the cost definition at 23 CFR 635.410, not invoice price.

Decide

  1. 06

    An input is missing, so the position stays open and the supplier who owes the figure is named in the pack.

  2. 07

    A waiver is relied on, so its version date, its comment window and its Made in America Office status are tracked.

Out

  1. 08

    A pack is complete, so it is held for the signer the recipient contract names, with the open items attached.

  2. 09

    Execute write actions only inside the approval boundaries agreed during implementation.

Product statement

The agent assembles the file; a named person signs it. Procurement asks what arrived and whether it conforms — this page asks what a signature can rest on today.

Example workflow

One funded item, order to signature

  1. 1Funded item identifiedSchedule of values, purchase order, submittal or supplier declaration
  2. 2Dates and rules matchedThe obligation date the customer supplies, the funding programme and the recipient contract
  3. 3Position draftedClassification, evidence chain, cost position and confidence
  4. 4Controls appliedEvidence-chain checks, cost-definition checks, waiver-version checks and confidence threshold

No human action required

Stages 1 to 4 run unaided, and nothing is certified at any of them — the agent is assembling, and the signer's lane opens at the confidence gate.

5DecisionBranches at the confidence threshold
High confidence

Goes to the authorised signer.

Low confidence

Adds a compliance read first.

Signer approval

The pack is held with its evidence chain, its open items and the confidence.

Approve · Amend · Send to compliance review

Approved — file released

  1. 6Project records updatedOnly where write access and approval policy allow it
  2. 7Outcome evaluatedGap-closure time, signer amendments, cut-off misses and post-signature disclosures

Amendments

Every signer amendment, and every later disclosure, is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required

8 items

  • Signing or issuing the certification.
    • Determining that a funded item complies.
    • Requesting a waiver, or predicting its outcome.
    • Inferring a component cost from a comparable part.

  • Automation boundary

    Agent acts unaided

    • ✓Map each funded item to the project date and rule the customer supplies.
    • ✓Assemble the manufacturer and supplier evidence behind each item.
    • ✓Build the cost position to the definition in the rule cited.
    • ✓Name the missing inputs and who owes them, and hold the pack.

    Any write happens inside the boundaries agreed at implementation, never ahead of the signature.

    • Deciding which event fixes the obligation date.
    • Releasing an item for permanent incorporation.
    • Disclosing non-compliant material to the recipient.
  • Changes to classification, threshold or waiver rules.

Example output

One funded item, annotated

Everything the agent assembles is attached to the document it came from.

Certification pack · single funded item

Illustrative example
Item
Signal cabinet
Position drafted
Assembled in the United States; component costs unreturned
Cost position
Open — inputs missing
Source of record
Mill and supplier declarations
Confidence
88%
Requirement applied
Manufactured product standard

As received

Taken from the purchase order and the declarations on file — nothing on this side is inferred by the agent.

Evidence used

  • Supplier declaration
  • Mill test report
  • Purchase-order clause

Why it is still open

A component cost is unreturned, so the position stays open for the signer to weigh.

What the score decides

Below the configured threshold the pack picks up a compliance read before it reaches the signer.

Action

ApproveAmendSend to compliance review

Value

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 funded item

From the purchase order

03Assembly

Build from the documents

Work from the declarations on file and the project dates and rules the customer supplies.

01Approved path

The date fixes the rule

Routine items arrive already matched to the rule their project sits under.

02Human review

Send review to the open gaps

Unsupported inputs and low-confidence packs are marked, so the signer reads where the exposure sits.

04

Build an evidence trail

The item, the funding date behind it and the person who certified stay on the file.

Integrations

Typical integrations

Five system groups connect to the same agent. Which of them are in scope is decided in discovery.

  • Project and cost systems

    • Viewpoint Vista
    • HCSS
    • Sage 300 CRE
    • B2W

  • Funding-programme sources

    • FHWA Buy America notices
    • Recipient bulletins
    • forms

  • Supplier evidence

    • Mill test reports
    • declarations
    • Submittal and PO records

Agent

Domestic-content certification

  • Reads the documents
  • Builds the pack
  • Holds for signature

  • Document control

    • Procore
    • Autodesk Docs
    • SharePoint
    • Egnyte

  • 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 signature

Six filters, tightest last. What none of the six catches is written into the map below.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modePull assembly back to gap-listing only when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt, threshold-rule and recipient-contract configuration changes.Track
L4TraceabilityRecord the documents read, the cost build-up, the open items and the signature time.Record
L3Signer approvalHold packs for the named signer; it governs release, not whether the signed position is right.Gate
L2Policy guardrailsTest the pack against the configured classification, cost-definition and waiver-version rules; a failure returns it.Restrict
L1Confidence thresholdsRoute low-confidence packs to a compliance read before the signer sees them.Require review
Model corePack produced — classification, evidence chain, cost position, open items and confidence
L1 – L2
Test whether a pack may stand
L3
Puts the signature in a person's hands
L4 – L5
Keep the item and the funding date behind it
L6
Narrows to evidence assembly when signals degrade

How Nestack evaluates it

Evaluate the certification workflow — not only the finished pack.

Coverage runs the whole depth of the workflow, and every layer is cut by slice.

Surface — the pack the signer reads

  1. E1Final-output evaluationDid every figure in the pack trace back to a source document?
  2. E2Step-level evaluationDid the agent use the right project date, programme and recipient rules?
  3. E3Tool evaluationDid it read the correct project and the correct supplier record?
  4. E4Confidence calibrationDo low-confidence packs actually attract more signer amendments?
  5. E5Slice evaluationHow does performance change across specific funding programmes?
  6. E6Business outcomeHow many packs needed an amendment, or a disclosure after signature?

Floor — the certification a named person answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, each placed at the stage it actually starts.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • GM-03

      Wrong project date read

      The date is taken from the field the ERP happens to hold.

    Stage gathers

    Purchase orders, declarations, dates and rules

  2. 02 · Reasoning

    2 modes

    • GM-04

      Inferred component cost

      A missing input is closed with a comparable part.

    • GM-06

      Classification skipped

      The predominance test is not run before a rule is chosen.

    Stage proposes

    Classification, cost position and confidence

  3. 03 · Tool / write

    2 modes

    • GM-02

      Cut-off missed

      A late declaration lands after the estimate cutoff.

    • GM-05

      De minimis double-spend

      One contract allowance is drawn against twice.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • GM-01

      Open gap left unnamed

      A component with no supplier evidence is not listed.

    Stage returns

    The pack the signer reads and the auditor tests

  5. 05 · Change / Version

    1 mode

    • GM-07

      Silent rule regression

      A model or rule change widens what the agent will assert.

    Stage tracks

    Model, prompt, threshold rules and recipient config

  • Sev-1 · acts outside the boundary
  • Sev-2 · an unsupported figure reaches the pack
  • Sev-3 · input degrades, pack routes to review
See the complete Construction failure-mode catalogue

Affected slices

One programme can carry most of the gaps

A package-level evidence-provenance figure can look acceptable while a few funding programmes absorb most of the unsupported inputs. Nestack reports the open-gap rate by slice, not only in total.

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
EPA revolving-fund packages8.4%3.7×Review
Waiver-dependent items5.9%2.6×Review
Multi-tier assemblies3.6%1.6×Watch
Federal-aid iron and steel1.1%0.5×Normal

Bar: open-gap-rate lift vs. federal-aid iron-and-steel baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

A cycle ends when a test exists

A cycle is done when the unsupported certification has become a case the next release must pass. That suite is what the next item certified is measured against.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Open-gap rate rises in one funding programme.

  2. 02

    Diagnose

    The item ordered under one rule and delivered under the next is read back to a single cause.

  3. 03

    Improve

    Stamp the change; the items behind it are filed against that number.

  4. 04

    Verify

    Each touched item case is run again, and one red stops the release.

  5. 05

    Learn

    It stays a standing test, and the threshold rules move with it.

  6. Learn → Detect

    The return edge. The next detection runs against a suite one case longer.

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, certification workflow, evaluation, integration, then production validation and handover.

Workstream

Week 1Week 2Week 3Week 4Week 5Week 6

  1. 01Certification workflow and automation-boundary scope.Week 1
  2. 02Project, cost and supplier source review.Week 1, Week 2
  3. 03Funding-date, threshold and waiver-rule mapping.Week 1, Week 2
  4. 04Item ingestion and document normalisation.Week 2
  5. 05Classification logic and evidence binding.Week 2, Week 3
  6. 06Confidence scoring and gap routing.Week 3
  7. 07Signer approval workflow.Week 3, Week 4
  8. 08Project and supplier-record integration.Week 5
  9. 09Threshold and waiver cases.Week 4, Week 5
  10. 10Guardrails and signing controls.Week 4
  11. 11Item-trail instrumentation.Week 5, Week 6
  12. 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 tierPilotOne programme, one jobProductionProduction project systemsAdvancedMultiple programmes / recipients
Introduced at Pilot
Pack assembly to your rules✓✓✓
Signer approval✓✓✓
Evidence-provenance baseline✓✓✓
Introduced at Production
Reporting by funding programme—✓✓
Approval workflow in your systems—✓✓
Approved write-back—✓✓
Supplier-record integration—✓✓
Introduced at Advanced
Multi-recipient contract rules——✓
Multi-stage signing approvals——✓
High item volume——✓
Multi-programme certification controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build price
Confirmed after discovery based on integrations, workflow complexity, transaction 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.

  1. 01Your funded jobs and their project dates

    Item ingestion and funding-date mapping

    Week 1

  2. 02Representative certification packs

    Assembly baseline, evidence binding and cost build-up

    Week 2

  3. 03Your recipient contract and its certification forms

    Recipient-contract, threshold and waiver-rule mapping

    Week 1

  4. 04Access to relevant APIs, feeds or exports

    Project, cost and supplier assessment, then integration setup

    Week 2

  5. 05Certifications you would not want audited

    Cut-off cases and the evaluation suite

    Week 4

  6. 06What no certification may take on trust

    Confidence scoring, gap routing, guardrails and approval controls

    Week 3

  7. 07Named signers to approve the packs

    Signer approval 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

Every band is drawn to the weeks that phase truly needs, and so week five carries a pair of them.

Phase

W1

W2

W3

W4

W5

W6

  1. DiscoveryW1
  2. BuildW2 – W3
  3. EvaluateW4 – W5
  4. Pilot & LaunchW5 – W6

Week focus

W1Certification workflow discovery, rule mapping and the automation boundary

W2Source integration and the assembly baseline

W3Assembly workflow, confidence logic and signing controls

W4Evaluation suite, cut-off cases and failure-mode testing

W5Supplier-record integration, pilot packages and targeted corrections

W6One funded package run under the authorised signer, then Agent Care handover

Reading the band
Each bar spans only the weeks its own work is named in. The week 5 overlap is real work, not padding.
At the end of W6
When the certification file validates, Agent Care takes the agent on.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Construction AI agent

Build a domestic-content agent around the person who signs.

The date on the funding paperwork decides which rule an item is certified under, and one project carries several. Show us your funded jobs and the person who signs. The False Claims Act counts per claim, so each pay estimate on a false certification is another one.

Typical delivery
4–6 weeks
Deployment
Your funded jobs
Signer approval
Estimate cutoff
After launch
Agent Care available

Nestack Agents · Domestic-content certification

AGT-CN-14 · Agent Care available after launch

Build a domestic-content agent

Four questions so the call starts from your actual setup — then pick a time that suits you.

About you
Your projects

No obligation. We’ll map the workflow, name the automation boundary and tell you if an agent is the wrong answer.