Agent Catalog · Version 1.0

Five productized reflexes.
Wired to the systems you already run.

Each agent in this catalog is a named, trigger-based autonomous reflex — not a chatbot, not a dashboard, not a "copilot" that waits for prompts. Every agent listens for one specific real-world event in the shop's workfile, executes a defined technical wire against CCC Secure Share, Mitchell Transactional, Audatex OAS3, Enterprise ARMS, or Hertz HIRS, and produces a specific, auditable output. Every entry in this catalog carries the trigger, the wire, the output, and the monthly economic impact. This is the menu. The shop operator picks what they want wired first.

Catalog Version v1.0 · Apr 2026
Agents Shipping 5 of 5 · production
Runtime claude code routines
Custom Builds on request · 2-4 wk
How to Read This Catalog

Four fields. Every agent. No exceptions.

Every autonomous agent is defined by four discrete fields. If an agent can't fill all four fields on one page, it's not productized — it's a prototype. Here is the grammar of the catalog.

01 — THE TRIGGER

The Real-World Event

The specific event that wakes the agent: estimate_saved, ecd_changed, vin_decode == ev. No trigger, no agent.

02 — THE WIRE

The APIs & Systems

The exact integrations: CCC Secure Share, Mitchell Transactional API, Audatex OAS3, Enterprise ARMS, Hertz HIRS, SMS webhooks. Documented endpoints, not screen scraping.

03 — THE OUTPUT

What the Shop Sees

Where the result lands: updated workfile lines, Slack post, SMS text, mobile checklist, insurer portal note. Outputs hit channels the shop already monitors.

04 — THE IMPACT

The Monthly Dollars

Modeled monthly economic value for a mid-size shop (~180 ROs/mo). Full derivation and sensitivity live in ROI Framework.

Ranking Rationale

The catalog is numbered 1–5. The sales order is not.

The agents are presented in numerical order for catalog clarity — matching the order in which they appear in internal documentation, the research report, and the file naming convention. That order has nothing to do with which agent gets pitched first.

The first pitch is always Agent 4 — the E01 Pre-Cognition & Teardown Defender. It is the flagship because it shows the sharpest, most emotional ROI: "we just stopped your technician from wasting eight hours on a total loss." It is also the most defensible pilot: either the E01 calls pay for themselves inside 30 days, or we refund.

The second pitch is Agent 3 (Status Sync) — the natural upsell because it compounds with Agent 4 and solves the most visible daily chaos in the front office (rental extensions, ECD slippage, customer texts).

Agents 1, 2, and 5 stack on top based on the specific pain points the shop surfaces during the discovery call.

Pitch Order vs Catalog Order
Why we open every deck with Agent 4
CATALOG ORDER SALES ORDER 01 P-Page Scrubber 02 Sublet Reconciler 03 Status Sync 04 E01 Predictor 05 EV Compliance 1ST E01 Predictor 2ND Status Sync 3RD P-Page Scrubber 4TH EV Compliance 5TH Sublet Reconciler Flagship route
The Agent Lineup

Five agents. Each one a discrete, productized reflex.

Read any card end-to-end to understand what the agent does, how it's wired, and what it's worth per month. Click through to the full specification for the deep technical wire, edge cases, and day-in-the-life walkthrough.

AGENT 01

P-Page & OEM Not-Included Scrubber

Estimators forfeit thousands per week because they lack the time to cross-reference OEM procedures against the three estimating databases' P-Pages. Every missed manual line — masking, seam sealer, ceramic coating removal, OEM rivets — is gross profit walking out the door.

Trigger event: estimate_saved
Wire
  • CCC Secure Share (BMS / EMS)
  • Mitchell Transactional API
  • Audatex OAS3 /api/v1/Claim
  • Vector index: P-Pages + DEG rulings
Output Pending supplement queue with missing manual lines, each annotated with the exact P-Page reference or DEG inquiry number — DRP-compliant and ready for adjuster defense.
$28,050/mo
Mid-size shop · captured lines
Read full specification →
Flow Diagram
TRIGGER estimate saved AGENT 01 P-Page Scrub DEG + OEM + P-Page OUTPUT lines queued CCC / Mitchell / Audatex reasoning + retrieval supplement draft
AGENT 02

Sublet Invoice & Markup Reconciler

When a mobile sublet vendor emails a PDF invoice, the estimator must read the PDF, calculate the DRP-compliant markup, type the line items into the estimate, and upload the PDF to the insurer portal. Every minute spent reconciling is a minute not spent blueprinting.

Trigger event: sublet_email_received
Wire
  • IMAP / Gmail inbox webhook
  • Claude vision (PDF OCR)
  • SMS REST API (DRP profile lookup)
  • Audatex /api/v1/ImageAttachment
Output Manual sublet lines injected into the workfile with precise DRP-allowable markup math; original PDF attached directly to the digital workfile, ready for supplement submission.
$430/mo
Captured markup · labor reclaimed
Read full specification →
Flow Diagram
PDF EMAIL Precision Dx Elitek · AirPro AGENT 02 Reconciler OCR + markup + attach OUTPUT lines + PDF in workfile inbox webhook vision + DRP lookup SMS + insurer
AGENT 03

Ecosystem Status Synchronizer

Every ECD slip triggers a swivel-chair workflow: log into Enterprise ARMS, extend the rental, switch to Hertz HIRS, repeat, open the insurer portal, write a justification, open the customer CRM, send the update. In a mid-size shop this happens 40+ times a week.

Trigger event: ecd_changed
Wire
  • SMS ECD field webhook
  • Enterprise ARMS (armsweb.ehi.com)
  • Hertz HIRS portal
  • CCC / Mitchell insurer note API
  • Twilio / customer SMS gateway
Output Rental extended with justification code, insurer portal note posted, customer SMS sent — all synchronized in <12 seconds, fully logged to the RO for audit trail.
$883/mo
CSR labor recovered · CSAT lift
Read full specification →
Flow Diagram
ECD SLIP SMS field date changed AGENT 03 Sync 3-way fanout ARMS + HIRS Insurer portal note Customer SMS under 12 seconds end-to-end
Flagship
AGENT 04

E01 Pre-Cognition & Teardown Defender

Shops waste massive technician hours tearing down severely damaged older vehicles that are inevitably totaled — then fight to get paid for the teardown. Total loss frequency hit 22.8% in late 2025. Every teardown on a total is a direct, measurable, avoidable loss.

Trigger event: assignment_dropped
Wire
  • CCC Secure Share assignment feed
  • VIN decode (NHTSA / OEM)
  • ACV lookup (KBB / Black Book)
  • State E01 threshold registry
  • Damage image analysis (vision)
Output High-priority "Probable Total Loss" flag on the RO, preliminary compensation lines (teardown, admin, hazard storage) pre-populated on the workfile to anchor mechanics lien defense — before teardown begins.
$7,480/mo
Teardown hours defended · lien anchored
Read full specification →
Flow Diagram
ASSIGNMENT dropped + FNOL photos uploaded AGENT 04 E01 Predictor ACV × damage × state threshold = TL probability OUTPUT TL FLAG + teardown lines staged + lien anchor before the tech touches the car
AGENT 05

High-Voltage EV Compliance Documenter

EV repair requires rigorous HV isolation, lock-out/tag-out logging, and 50-ft or ecology-block quarantine. Missing documentation exposes the shop to catastrophic liability and guarantees the insurer will not pay for the safety labor. Most shops document intermittently — or not at all.

Trigger event: vin_decode == ev
Wire
  • VIN decoder (NHTSA) — powertrain type
  • OEM service procedure index
  • SMS mobile app (tech checklist push)
  • CCC / Mitchell workfile line injection
  • Photo/log attachment endpoints
Output Mandatory HV safety checklist pushed to the tech's mobile device; quarantine/de-energize/re-energize billing lines injected into the workfile; compliance log archived for OEM/insurer audit.
$280/EV
Per-vehicle · compounds by EV volume
Read full specification →
Flow Diagram
VIN DECODE == EV / PHEV at check-in AGENT 05 EV Compliance OEM + NHTSA + NFPA Tech mobile checklist Workfile HV lines + audit log tech gets checklist · shop bills every line
Agent Interoperability

Agents compose. Two reflexes stacked is more than twice one.

Each agent listens for its own trigger, but all five share the workfile as common context — and they can call one another's outputs as inputs. The stacked combinations are where the real leverage shows up.

Every agent writes its findings, flags, and outputs back to the workfile (CCC ONE, Mitchell RepairCenter, or the shop's SMS of record). The next agent down the chain reads the previous agent's state and can act on it. This is not a manual handoff — it's a single shared substrate that every reflex touches.

In practice this means two things. First: adding an agent never increases shop overhead because nothing new has to be entered manually. Second: the compounding combinations create leverage that no single agent provides on its own.

  • AGENT 01 + AGENT 04 Every estimate is P-Page scrubbed AND pre-checked for total-loss risk. Manual lines added defensively on teardown-likely ROs anchor the lien before the tech starts.
  • AGENT 03 + AGENT 05 Every EV workflow is both compliance-documented and status-synced. The tech gets the checklist, the estimator bills the lines, and the customer knows why the rental is extended — all from one VIN decode.
  • AGENT 02 + AGENT 01 Sublet PDFs are OCR'd, markup-applied, AND P-Page scrubbed for downstream labor operations the sublet visit generated (setup, calibration, test drive).
Shared Context
The workfile is the substrate
THE WORKFILE CCC ONE / Mitchell / SMS via MCP connectors AGENT 01 P-Page Scrub AGENT 02 Sublet AGENT 04 E01 ★ AGENT 03 Status Sync AGENT 05 EV Compliance mcp read mcp read mcp write mcp write
Custom Agents
V2 · Built on Request

Five agents is the catalog. Your pain points set the roadmap.

The five V1 agents address the pain points common to almost every independent or mid-size MSO. Every engagement surfaces shop-specific friction that no generic agent handles — and Intersection Strategies builds a custom agent for it, typically in 2-4 weeks, included free on the Operator Tier. Common V2 candidates that have come up in discovery calls:

  • DRP Scorecard Guardian Monitors your shop's standing inside each DRP program — cycle time, CSI, severity — and flags the specific ROs hurting the scorecard before the monthly carrier review.
  • AI-Estimate Translator Reads the carrier's FNOL photo AI estimate, converts it into an OEM-aligned blueprint, and pre-drafts the rebuttal to every line the AI underestimated.
  • Parts Price Monitor Watches OEM and aftermarket parts pricing in real time; triggers a price-protection supplement the moment a quoted part jumps 5%+ between estimate and order.
  • Insurance Portal Upload Bot Handles the CCC Secure Share / Mitchell Connect uploads: photos, scan reports, calibration verifications — tagged correctly for insurer visibility, every time.
What's Next

Two ways forward from this catalog.

Jump to the first agent specification directly: agent-01-ppage-scrubber.html →