Main site

Communications AI agent · Disclosure drafting

Statement and Release Drafting AI Agent

The results release is the exhibit to the filing, so the agent drafts it against the rule that actually reaches it, records the build, and holds it for the disclosure officer who signs.

Typical delivery
4–6 weeks
Deployment
Your stack
Named officer
Filing first
After launch
Agent Care

What this agent does

Drafts the release, never releases it

In

  1. 01

    A results release is drafted, and that same text goes to the Commission as the Item 2.02 exhibit.

  2. 02

    A non-GAAP measure appears, and Regulation G asks for a reconciliation, not for a place on the page.

Reason

  1. 03

    The release reports a closed quarter, so Instruction 2 carries Item 10(e)(1)(i) across, and no more.

  2. 04

    A deal release quotes an adjusted margin, and the prominence rule does not reach that release.

  3. 05

    A cautionary paragraph is carried forward, and it is marked as a claim about present fact.

Decide

  1. 06

    A forward-looking line is drawn, and the facts qualifying it under 78u-5 sit outside the draft.

  2. 07

    A figure moves, and the bridge beneath it is re-added rather than reformatted around the change.

Out

  1. 08

    A release cannot be tied back to the closed ledger, and that is recorded as the finding.

  2. 09

    Execute write actions only within the approval boundaries fixed at implementation.

Product statement

Drafting, reconciliation and the draft trail belong to the agent. Release belongs to a named disclosure officer, who signs the report on behalf of the company and owns it there.

Example workflow

One release, close to wire

  1. 1Inputs receivedClose package, adjustment schedules, the guidance range, prior releases and counsel notes
  2. 2Release class fixedWhether this is a results release under Item 2.02 or a release governed by Regulation G alone
  3. 3Draft raised against that classThe headline, the reconciliation table, the cautionary block and the officer quote
  4. 4Checks appliedProminence checks, bridge arithmetic, cautionary-language diffs and drafting confidence

No human action required

Stages 1 to 4 run unaided, and nothing reaches a wire at any of them — the agent is drafting, and the officer lane opens at the release gate.

5DecisionSplits at the release gate
Clean against the fixed class

Goes to the named disclosure officer to release.

Anything contested

Adds a securities-counsel read first.

Officer review

The draft is held with its release class, its reconciliation and the source each figure came from.

Release · Append source · Send to counsel review

Released — by the named disclosure officer

  1. 6Filing and wire records updatedOnly where write access and the records policy permit it
  2. 7Release evaluatedClass accuracy, bridge ties, officer corrections and what counsel review found

Officer corrections

Each correction the officer makes is scored in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required

8 items

  • Putting a statement on the wire.
    • Deciding what this quarter makes material.
    • Judging which rule reaches this release.
    • Signing the current report for the company.

  • Automation boundary

    Agent acts unaided

    • ✓Fix the release class before a line is drafted.
    • ✓Reconcile the non-GAAP measure to its nearest GAAP twin.
    • ✓Diff the cautionary block against the one shipped last quarter.
    • ✓Hold the draft, the tables and the sources in one numbered record.

    Nothing crosses a wire except by a named disclosure officer, inside the agreed boundaries.

    • Deciding a named risk has not yet materialised.
    • Telling the market a figure is final.
    • Choosing the guidance range a release carries.
  • Changes to the release, the tables or the filing.

Example output

One results release, annotated

This serves a communications team whose draft becomes the exhibit to a federal filing; below is one release exactly as the agent leaves it.

Results release · completed quarter

Illustrative example
Release
Results release, one closed quarter
Recorded as
Reconciled to the close package
Class
Item 2.02, furnished
Evidence of record
Close package, 6 August 2026
Confidence
Held unreleased
Held for
The disclosure officer, by name

As received

Built from the close package and the adjustment schedules under it, and it claims nothing past them.

What the draft record holds

  • Close package
  • Adjustment schedule
  • Prior-quarter release

Why no release here

Putting a release on the wire is a call a named officer alone makes.

What the score decides

Below the configured threshold a draft gets a counsel read before the officer sees it.

Action

ReleaseAppend sourceSend to counsel 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

Each release drafted

From the figure that carries it

03Evidence

Where the evidence lands

The agent does not vouch for a figure, only for where it came from, which rule the release sits under, and how the draft was built from it.

01Approved path

The release is the filing

Regulation G asks for a reconciliation and says nothing about placement; the prominence rule sits in Item 10(e)(1)(i)(A), and one instruction carries it over.

02Human review

Checked for, and not found

No published evaluation was found of a model drafting a securities-compliant results release, and no enforcement action was found whose central allegation is the press release itself, so the evidence here is analogical and is reported that way.

04

Build an evidence trail

The sentence, the figure under it and the officer who released it stay together.

Integrations

Typical integrations

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

  • Close and consolidation systems

    • ERP
    • group consolidation ledger
    • Closed-period GAAP statements

  • Adjustment and non-GAAP work

    • FP&A workbooks
    • schedules
    • Adjustment conventions on file

  • Prior releases and exhibits

    • Wire archive
    • furnished exhibits
    • Cautionary language shipped before

Agent

Statement and release drafting

  • Reads the close
  • Drafts against the class
  • Holds for the officer

  • Wire and exchange desks

    • Wire vendor
    • MarketWatch notices
    • Notice timings and dispatch times

  • Observability & evaluation

    • OpenTelemetry
    • Langfuse
    • Supported monitoring/evaluation sources

Integration availability depends on the client's existing systems and API access.

Agent controls

Six gates between the model and the wire

Six gates along one corridor, the last the narrowest. What clears is set out in the map below.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modeNarrow the agent to source assembly when evaluation or production signals degrade.Roll back
L5Version trackingTrack model, prompt and rule-class logic, and note the version each draft was written under.Track
L4Draft trailRecord each draft, the figures beneath it, the class it was written under and each read of the file.Record
L3Officer releaseHold the draft for a named officer; the hold governs the wire, not whether the text is right.Gate
L2Prominence guardrailsTest each draft against its fixed class, and refuse a headline that leads on an adjusted measure alone.Restrict
L1Confidence routingRoute a low-confidence draft to a counsel read before the release reaches the officer.Require review
Model coreDraft written — the class, the headline, the reconciliation and its sources
L1 – L2
Test whether a draft may stand
L3
Leaves the release to a named officer
L4 – L5
Keep the release and the reconciliation behind it
L6
Drafts no figure at all when signals degrade

How Nestack evaluates it

Evaluate the whole drafting run — not only the release that comes out.

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

Surface — the release a wire carries

  1. E1Final-output evaluationDid the draft record the rule class it was actually written under?
  2. E2Step-level evaluationDid the agent read the closed period, the right schedule and the live conventions?
  3. E3Tool evaluationDid it read and write the correct period and the correct measure?
  4. E4Confidence calibrationDo low-confidence drafts actually attract more officer corrections?
  5. E5Slice evaluationHow does performance change across specific measure classes?
  6. E6Business outcomeHow many drafts needed a correction before the officer released?

Floor — the close a release rests on

Failure modes

Where each failure originates in the agent

Seven failure modes, each set at the stage it first becomes visible.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • SL-07

      Quote taken in unlabelled

      A supplied quote arrives with its connection undisclosed.

    Stage gathers

    The close, the tables, the prior text and the quotes

  2. 02 · Reasoning

    2 modes

    • SL-01

      Prominence inverted

      The headline leads on the adjusted figure alone.

    • SL-03

      Wrong rule class applied

      A deal release is drafted as a results release.

    Stage proposes

    The class, the headline, the bridge and confidence

  3. 03 · Tool / write

    2 modes

    • SL-04

      Bridge that does not tie

      A component figure is plausible and wrong.

    • SL-06

      Confident half-truth

      A cause is supplied that the evidence does not carry.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • SL-05

      Draft out before the wire

      A version reaches a reader ahead of the release.

    Stage returns

    The release a wire carries and the market reads

  5. 05 · Change / Version

    1 mode

    • SL-02

      Cautionary block carried forward

      The warning stands after the risk it names occurred.

    Stage tracks

    Model, prompt, rule class and quarter dates

  • Sev-1 · a release out on no record
  • Sev-2 · a stale figure reaches the wire
  • Sev-3 · signals degrade, draft held back
See the complete Communications failure-mode catalogue

Affected slices

Newly adjusted measures absorb the corrections

A release-level correction figure can read clean while newly adjusted measures carry most of the rework. Nestack reports the correction rate by measure class, not only across the release.

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
Newly adjusted measures13.2%3.7×Review
Dual-listed releases9.4%2.6×Review
Forward-looking guidance5.8%1.6×Watch
Settled GAAP line items2.6%0.7×Normal

Bar: correction-rate lift vs. settled-line baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

What an inverted headline costs

The loop shuts when the measure given more prominence than its GAAP twin is a standing case. That suite is what the next release is measured against.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Correction rate rises on newly adjusted measures.

  2. 02

    Diagnose

    The headline that led on an adjusted figure because it simply read better is worked backwards until one cause is left standing.

  3. 03

    Improve

    Every release goes out numbered, with the drafts that preceded it attached.

  4. 04

    Verify

    Nothing goes on the wire while one touched measure case is red.

  5. 05

    Learn

    The case stays on, and the drafting rules are redrawn beside it.

  6. Learn → Detect

    The return edge. The next release is measured 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, drafting logic, evaluation, integration, then production validation and handover.

Workstream

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

  1. 01Release-classification and automation-boundary work.Week 1
  2. 02Close, schedule and prior-release sources.Week 1, Week 2
  3. 03Draft-source-to-release and prominence-rule mapping.Week 1, Week 2
  4. 04Close-package ingestion.Week 2
  5. 05Class, figure and period binding.Week 2, Week 3
  6. 06Prominence scoring and review routing.Week 3
  7. 07Officer release workflow.Week 3, Week 4
  8. 08Filing and wire integration.Week 5
  9. 09Measure and prominence cases.Week 4, Week 5
  10. 10Guardrails and drafting controls.Week 4
  11. 11Draft-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 release, one cycleProductionProduction disclosure workflowAdvancedMultiple releases / jurisdictions
Introduced at Pilot
Drafting to your disclosure rules✓✓✓
Named officer release✓✓✓
Disclosure-calendar baseline✓✓✓
Introduced at Production
Reporting by measure class—✓✓
Counsel review workflow in your systems—✓✓
Approved filing write-back—✓✓
Wire-and-filing integration—✓✓
Introduced at Advanced
Multi-measure reconciliation——✓
Cross-release language packs——✓
Dense release calendars——✓
Multi-jurisdiction release controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build price
Confirmed after discovery based on integrations, workflow complexity, release 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 live releases and the class each one falls under

    Class capture and prominence versioning

    Week 1

  2. 02Representative close, schedule and prior-release material

    Source binding, drafting logic and the release baseline

    Week 2

  3. 03Your results calendar and the officers it names

    Rule-class mapping, source binding and the automation boundary

    Week 1

  4. 04Access to the relevant systems, feeds or exports

    Close, schedule and wire-system assessment, then integration setup

    Week 2

  5. 05Sentences you would not want deposed on

    Prominence cases and the evaluation round

    Week 4

  6. 06What no release may imply

    Prominence scoring, review routing, guardrails and release controls

    Week 3

  7. 07A named disclosure officer who releases

    Disclosure-officer sign-off, 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

Widths follow the work rather than the grid, which is why one band sits beneath another in week five.

Phase

W1

W2

W3

W4

W5

W6

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

Week focus

W1Rule-class discovery, prominence versioning and the automation boundary

W2Close and schedule integration, and the disclosure-calendar baseline

W3Drafting logic, reconciliation checks and release controls

W4Evaluation suite, prominence cases and failure-mode testing

W5Wire and filing integration, pilot drafts and targeted corrections

W6One reporting cycle run under the disclosure officer, then Agent Care handover

Reading the band
Each bar spans only the weeks its own work is named for, and week five is shared by design.
At the end of W6
Once the release record validates, Agent Care adopts the agent.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Communications AI agent

Build a release drafting agent around the rule your last results release was actually written under.

Show us one results release you publish each quarter and the reconciliation sitting behind its headline. If the adjusted figure leads and the GAAP figure waits in a later paragraph, the layout is the exposure. A release nobody can class comes back as a finding.

Typical delivery
4–6 weeks
Deployment
Your release calendar
Named officer
Filing first
After launch
Agent Care available

Nestack Agents · Disclosure drafting

AGT-COM-02 · Agent Care available after launch

Build a release drafting agent

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

About you
Your releases

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