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.
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.
The specific event that wakes the agent: estimate_saved,
ecd_changed,
vin_decode == ev. No trigger, no agent.
The exact integrations: CCC Secure Share, Mitchell Transactional API, Audatex OAS3, Enterprise ARMS, Hertz HIRS, SMS webhooks. Documented endpoints, not screen scraping.
Where the result lands: updated workfile lines, Slack post, SMS text, mobile checklist, insurer portal note. Outputs hit channels the shop already monitors.
Modeled monthly economic value for a mid-size shop (~180 ROs/mo). Full derivation and sensitivity live in ROI Framework.
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.
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.
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.
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.
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.
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.
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.
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.
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: