ETTAgentic managed services from ETT Group

We run the work. You pay for what’s proven.

Agentic MSP runs back-office processes with governed AI agents and a trained exception desk. Every outcome is recorded in the Evidence File, and only verified outcomes reach your invoice.

Book a Business Process Workshop See how it works

For independent insurance agencies and mid-market finance teams in the US and Canada.

Evidence FileExample entries

09:14Certificate of insurance issuedACORD 25, general contractor. Record 3f9a…c21eVerifiedBillable
09:16Renewal policy checked against quoteBOP renewal, 4 discrepancies logged. Record b71d…0a94VerifiedBillable
09:21Invoice matched and postedINV-20931 to PO 7718. Record e02c…5f3bVerifiedBillable
09:23Endorsement prepared: add vehicleCommercial auto, licensed review. Record 91be…77d0Awaiting your sign-offNot billed
09:30Loss runs extracted3 carriers, 1 page unreadable. Record 4c6f…a118With exception deskNot billed

3 verified and billable2 in progress, not billed

The gap

AI is everywhere. Finished work isn’t.

Businesses have bought the tools and run the pilots. What most still don’t have is someone who owns the result.

48% of MSPs say AI and automation is their clients’ top need for 2026. Only 13% make meaningful revenue from it.

Kaseya 2026 State of the MSP Report (via VMblog)

About 7 in 10 small businesses use AI regularly. About 1 in 10 pay for dedicated AI tools.

Intuit QuickBooks 2026 AI Impact Report

Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027, for cost, unclear value or weak risk controls.

Gartner forecast (via New Electronics)

The models are not the problem. Nobody owns the outcome.

A service that owns the outcome.

Agents take the routine volume. People take the exceptions. You keep the decisions that need your judgement or your license.

Agents do the routine work
Reading documents, checking policies, matching invoices and updating your systems of record.
Exceptions go to trained people
Anything the agent is unsure of goes to our exception desk, not back to your inbox.
You sign off what matters
Licensed decisions and anything that moves money wait for your team’s approval.
Only verified outcomes are billed
Each outcome is checked against a written definition and recorded before it reaches your invoice.

Read how a piece of work moves through the service

Office workstation with two monitors and paperwork on the desk

The Evidence File

Every outcome, traced back to its source.

The Evidence File is an append-only record of every input, decision, model version and human approval behind each outcome. It is how we prove the work, and what your auditors, regulators and insurers can check.

Governance and compliance is the top barrier to agentic AI, named by 47% of respondents in Omdia’s research. We built the service around answering it. See the research.

Inside the Evidence File

Record b71d…0a94: renewal policy checked

Verified

Input: Carrier renewal policy (PDF, 38 pages) and the bound quote from the agency management system
What the agent did: Compared limits, deductibles, forms and endorsements line by line
Result: 4 discrepancies found and listed with page references
Checks passed: All 38 pages read; every quoted coverage line matched to a policy line or flagged
Human involvement: None required. Discrepancies sent to the account manager for carrier follow-up
Model and version: Recorded per step, with prompt and tool versions
Outcome definition: Policy checked: every coverage line compared and discrepancies delivered to the agency
Integrity: Hash-chained to the previous record, so any later edit is detectable
Billing: Billable once, as one verified policy check

Two kinds of business, one standard of proof.

Hands reviewing a printed document with a pen, beside two open laptops

Small business tier

Independent insurance agencies

Policy checking, certificates, endorsements, renewal preparation and commission reconciliation. Paid per verified transaction. Licensed decisions stay with your team.

Servicing for insurance agencies

A team meeting around a table with laptops while one colleague stands to present

Small enterprise tier

Mid-market finance operations

Order-to-cash, document intake, audit and control testing for companies with 100 to 1,000 staff. A run fee plus a capped share of verified gains.

Finance operations on outcomes

Outcome pricing

A floor fee, then outcomes.

A floor fee covers running the service. On top of it you pay per verified transaction, or a capped share of the value we create against a baseline you have agreed. If it isn’t verified, it isn’t billed.

How outcome pricing works

Monthly invoiceExample

Floor feeFixedPlatform, models, exception desk and governance

Verified outcomesPer item or gain-shareEach line links to its Evidence File record

Agent time, retries, failed itemsNot billed

First steps

Start with one process.

We don’t ask you to transform anything. We pick one process, measure where it is today, and run it.

Business Process Workshop

For finance and operations leaders. A fixed-price engagement that maps your processes, pulls a baseline from your own systems and scopes a paid Proof of Value.

Book a workshop

Insurance agency design partners

We are working with a small group of independent agencies to build the servicing pack. Partners shape the workflows and go live first.

Apply to be a design partner

Questions we get asked first.

Straight answers on risk, control and cost. More detail in our guide to agentic MSPs.

What is Agentic MSP?

Agentic MSP is a managed service provider that runs back-office processes with governed AI agents and a trained human exception desk. We are paid for verified outcomes, such as a policy checked or an invoice posted, rather than for licenses, seats or hours. Agentic MSP is part of ETT Group.

How is this different from buying an AI tool?

With a tool, your team still has to run it, check it and fix it. With Agentic MSP, we run the work, handle the exceptions and prove each outcome in the Evidence File. You pay when the outcome is verified, not for access to software.

What happens when an agent gets something wrong or is unsure?

Low-confidence work goes to our exception desk, where trained operators resolve it. Anything that needs a license, a judgement call or moves money goes to your team for sign-off. Work that does not meet its outcome definition is not billed.

Who do you work with?

Independent insurance agencies in the US and Canada, starting with policy servicing work, and mid-market companies with roughly 100 to 1,000 staff that want order-to-cash, document intake and finance operations run on outcomes.

How do we get started?

Mid-market teams start with a fixed-price Business Process Workshop that maps the process and pulls a baseline from your own systems. Insurance agencies can apply to our design partner program, which shapes the first servicing pack.

Agentic MSP is part of ETT Group, which delivers governed AI and automation from offices in Dallas, London and Vancouver. Agentic MSP runs the work; ETT Group designs and builds it.

Visit ettgroup.ai