Insurance AI agent · Rate filing
Rate-Filing & SERFF Workflow AI Agent
Assemble the filing and its exhibits against each state's checklist, keep the filing calendar and effective dates straight, draft objection responses with their support attached, and hold every filing for the actuary who signs.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Actuary review
- Pre-submission
- After launch
- Agent Care
What this agent does
Assembles the filing, not the decision to submit
In
- 01
An indication is prepared, and the agent pulls the checklist for the state and line it is headed to.
- 02
A state is selected, and the agent maps the exhibits and rate pages that state's checklist requires.
Reason
- 03
Each exhibit is drafted from the support already on file, not invented from a template.
- 04
The carrier's own filing calendar and configured effective-date rules are applied to the filing.
- 05
Every drafted line in an exhibit is bound to the workpaper or memorandum it was drawn from.
Decide
- 06
An objection letter is matched to the support already on file, and what's missing is flagged.
- 07
The assembled package is routed to the actuary and the filings manager for sign-off.
Out
- 08
The exhibit, its source support, the draft language and the sign-off stay on the filing.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent assembles the filing package; a qualified actuary signs the support and the filings manager submits, and neither step happens without them.
Example workflow
One filing, indication to disposition
- 1Indication receivedRate indication, target state and line, and the actuarial support behind it
- 2Checklist assembledExhibits, rate pages, loss experience and the state's own filing checklist, each with a source
- 3Filing draftedExhibit language, disclosure answers, flagged gaps and confidence
- 4Controls appliedChecklist completeness, disclosure-question checks, regime fit and confidence threshold
No human action required
Stages 1 to 4 run unaided, and nothing is submitted at any of them — the agent is drafting, and the actuary's lane opens at the confidence gate.
Goes to the actuary and filings manager to approve.
Adds a compliance read before sign-off.
Actuary review
The filing is held with its exhibits, its flagged gaps and the confidence.
Approve · Edit · Request more support
Approved — released to submit
- 6SERFF record updatedOnly where write access and approval policy allow it
- 7Outcome evaluatedExhibit accuracy, objection outcomes, disposition and post-submission corrections
Edits
Every edit made before submission is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Submitting the filing to SERFF or a state department.
- Signing or certifying the actuarial opinion or memorandum.
- Deciding whether a filed rate is approved for use.
- Implementing a rate in a policy administration or rating system.
Automation boundary
Agent acts unaided
- ✓Assemble the filing package and its exhibits from the support on file.
- ✓Bind each exhibit line to the workpaper or memorandum.
- ✓Apply the carrier's own filing calendar and configured checklist rules.
- ✓Flag gaps against the checklist and objection support.
Any write happens inside the boundaries agreed at implementation, never ahead of sign-off.
- Treating a submitted filing as an effective or usable rate.
- Answering a state's AI-use disclosure question without a named human confirming it.
- Deciding which filing regime governs a state-and-line combination, unreviewed.
- Changes to the filing-checklist library or regime classifications.
Example output
One exhibit in the filing, annotated
Everything the agent drafts is attached to the support it was drawn from.
Filing-exhibit output · single filing
Illustrative example- Filing
- Personal auto rate revision, TX
- Drafted line
- The indicated change is supported by the loss-development pattern in the exhibit, not asserted on its own
- Stated impact
- Exhibit 4, p. 2
- Support type
- Loss-development exhibit
- Confidence
- 91%
- Attribution
- Actuary of record and licence ID
As received
Taken from the exhibit and the citation given — nothing on this side is written by the agent.
Support used
- Loss triangle
- Rate-impact exhibit
- Prior filing history
Why this exhibit
It states what the exhibit supports, not a judgment about the rate.
What the score decides
Below the configured threshold, the filing picks up a compliance read before the actuary sees it.
Action
ApproveEditRequest more supportValue
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 filing
From the filing checklist
Draft from the checklist
Draw on the support on file and the carrier's configured filing calendar and checklist rules.
Filed is not the same as usable
Routine exhibit language and disclosure answers arrive already drafted.
Point the actuary at what needs support
Gaps against the checklist and unsupported lines are flagged, so review starts where support is thin.
Build an evidence trail
The exhibit, the support behind it and the actuary who signed stay on the filing.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Rating & filing systems
- SERFF
- state DOI portals
- WKFC Sircon
- Insurity
Actuarial & pricing tools
- Rate and loss exhibits
- Milliman ARC/Radar
- Earnix
Policy admin & rate data
- Rate manuals
- loss data
- Guidewire
- Duck Creek
Agent
Rate-filing & SERFF workflow
- Reads the checklist
- Drafts the exhibits
- Holds for sign-off
Objection & case tracking
- SERFF correspondence
- Objection-letter tracking
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 filing
Controls nest inward. What the stack misses is set out in the map beneath it.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether the filing may proceed
- L3
- Puts the sign-off in the actuary's hands
- L4 – L5
- Keep the exhibit and the support behind it
- L6
- Holds the filing at draft when signals degrade
How Nestack evaluates it
Evaluate the filing workflow — not only the finished exhibit.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the filing the regulator sees
- E1Final-output evaluationDid every exhibit support the not-excessive, not-inadequate, not-unfairly-discriminatory standard?
- E2Step-level evaluationDid the agent use the right checklist, calendar and regime configuration?
- E3Tool evaluationDid it read and write the correct filing and the correct field?
- E4Confidence calibrationDo low-confidence exhibits actually attract more compliance reads?
- E5Slice evaluationHow does performance change across specific filing types?
- E6Business outcomeHow many filings needed a compliance edit or drew an objection?
Floor — the outcome the carrier answers for
Failure modes
Where each failure originates in the agent
Seven failure modes, placed at the stage each one originates.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- FI-03
Superseded checklist
A state's filing checklist is read from a version the department has since revised.
Stage gathers
Indication, target state and line, checklist and history
02 · Reasoning
2 modes
- FI-04
Regime conflated
A file-and-use state is treated as if it were prior approval, or the reverse.
- FI-06
Support overstated
An indicated rate is presented as the actuary's final justified figure.
Stage proposes
Exhibit lines, disclosure answers, confidence
03 · Tool / write
2 modes
- FI-02
Disclosure misstated
A generic tool name is given instead of the specific AI system used.
- FI-05
Objection history dropped
A refiled filing loses the objection context from its earlier filing number.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- FI-01
Disclosure question skipped
The filing template predates the state's AI-disclosure question and was never updated.
Stage returns
The filing the actuary signs and the regulator receives
05 · Change / Version
1 mode
- FI-07
Silent checklist drift
A checklist or regime-configuration change widens what the agent will draft unflagged.
Stage tracks
Model, prompt, checklist and regime configuration
- Sev-1 · a filing submits outside the boundary
- Sev-2 · unsupported line reaches the filing
- Sev-3 · support degrades, routes to review
Affected slices
A clean rate can hide where objections land
A state-and-line approval rate can look clean in aggregate while a handful of combinations carry most of the objection load. Nestack reports the objection rate by slice, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Filings in prior-approval states | 7.6% | 3.9× | Review | |
| Filings supporting a model-based rate | 5.7% | 3.0× | Review | |
| Multi-state programme filings | 3.4% | 1.8× | Watch | |
| File-and-use rate revisions | 1.9% | 0.7× | Normal |
Bar: objection-rate lift vs. the file-and-use baseline · scale 0–4.0× · tick at 2.0×
2 of 4 slices over threshold
Evidence-linked improvement
Every cycle earns the filing its next case
A cycle closes when the objection is a case the next release has to answer. That suite is what the next filing assembled is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Objection rate rises in a state-and-line slice.
- 02
Diagnose
The objection letter that asked for one exhibit is traced back through the filing until the cause narrows to one line.
- 03
Improve
The correction is versioned, with the filings that motivated it attached.
- 04
Verify
A failing exhibit case holds the release until it passes.
- 05
Learn
The case is added permanently, and the state rules move with it.
Learn → Detect
The return edge. Detection next time runs against a suite one filing 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, filing sources, drafting workflow, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Filing workflow discovery and scope and boundary definition.Week 1
- 02SERFF and filing-system assessment.Week 1, Week 2
- 03State checklist and regime mapping and rule mapping.Week 1, Week 2
- 04Indication ingestion and normalisation.Week 2
- 05Exhibit-drafting logic and support binding.Week 2, Week 3
- 06Confidence scoring and gap routing.Week 3
- 07Actuary review workflow.Week 3, Week 4
- 08SERFF and rating-system integration.Week 5
- 09Exhibit and objection cases.Week 4, Week 5
- 10Guardrails and submission controls.Week 4
- 11Filing-trail 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 state, one line | ProductionProduction filing systems | AdvancedMultiple states / lines |
|---|---|---|---|
| Introduced at Pilot | |||
| Drafting to your checklist and rules | ✓ | ✓ | ✓ |
| Actuary review | ✓ | ✓ | ✓ |
| Exhibit-completeness baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by state and line | — | ✓ | ✓ |
| Actuary workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Filing-system integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-state regime complexity | — | — | ✓ |
| Multi-stage filing approvals | — | — | ✓ |
| High filing volume | — | — | ✓ |
| Multi-state filing controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on integrations, workflow complexity, filing volume, review 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 state list and filing-checklist library
→Checklist ingestion and state-and-line mapping
Week 102Representative past filings, including objections
→Exhibit-drafting baseline and support-binding baseline
Week 203Your filing calendar and effective-date rules
→State checklist and regime mapping
Week 104Access to relevant APIs, feeds or exports
→SERFF and rating-system assessment, then integration setup
Week 205Filings you would not want submitted
→Objection cases and the evaluation suite
Week 406What no filing may go in without
→Confidence scoring, gap routing, guardrails and submission controls
Week 307Named actuaries and filings managers to review
→Actuary review 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
Weeks are drawn where the work sits, so the fifth carries both evaluation and the first live filings.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Filing workflow discovery, checklist mapping and the automation boundary
W2SERFF and rating-system integration and the drafting baseline
W3Drafting workflow, confidence logic and actuary-review controls
W4Evaluation suite, disclosure-question checks and failure-mode testing
W5SERFF integration, pilot filings and targeted corrections
W6One filing cycle assembled under the pricing actuary, then handover
- Reading the band
- A bar covers the weeks its work is named in, and nothing else. The week 5 overlap is real, not padding.
- At the end of W6
- Validation closes on live filings, and Agent Care takes over monitoring.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Insurance AI agent
Build a rate-filing agent around your state and SERFF checklist.
Show us your filing checklist and your state list. Prior approval or file-and-use? The calendar and the automation boundary follow from the answer, and we'll map both around the actuary who signs.
- Typical delivery
- 4–6 weeks
- Deployment
- Your SERFF filings
- Actuary review
- Pre-submission
- After launch
- Agent Care available
Nestack Agents · Rate-filing & SERFF workflow
AGT-INS-17 · Agent Care available after launch