InsightRollup · The AI-first acquisition workbench

Buy the agencies worth buying. Every figure from the book.

InsightRollup reads a seller’s book export row by row and computes what you are really buying — retention, concentration, revenue quality, renewals. Ten agents prepare the seller requests, the memo and the 100-day plan; your team decides, on the record.

Request a demo See how it works

6agencies in the demo pipeline, screening to day 100

3,232book rows behind one agency’s figures

10agents, every action capped at ask first

8acquisition controls in Rules Studio

Figures from the illustrative demo book: invented acquirer, agencies, carriers and clients.

The InsightRollup Acquisition Workspace for an illustrative agency, Ironwood Risk Partners, at IC memo: the stage strip; book revenue $3.2M, client retention 90.5%, organic growth 3.4%, top producer 26.1% of revenue, top carrier 29.3% of premium, 22.0% of revenue renewing in 90 days; below, the first decision the book intake agent prepared — ask the seller to correct 70 rows — with Accept, Reject and Ask — **Ironwood Risk Partners, at IC memo.** Every figure computed from the 3,232 rows of the seller’s export, and the first decision the agents prepared. Illustrative data; click any screen to see it full size.

Proof, not promises

Three screens from the demo pipeline, as they are

An invented acquirer is buying six invented agencies. This is what InsightRollup put in front of the deal team without being asked.

A finding on an illustrative agency: three producers hold 61.8% of revenue — Rex Dalton 26.1%, Mae Holt 19.1%, Sol Rivera 16.6% — with chips for the 367, 234 and 194 book rows behind each share — **A finding cites its rows**Three producers hold 61.8% of the revenue. Each share opens the book rows it was computed from.

A decision the carrier mapping agent prepared: request appointments with four carriers, with each carrier's share of premium and appointment status, the ten facts behind it, and Accept, Reject and Ask — **Prepared, never sent**Appointment requests for four carriers and the facts behind them. Nothing reaches a carrier until a person accepts.

The acquisition controls that fired on an illustrative agency: a producer above 25% of revenue and book revenue more than 5% below the seller's, both referred to the practice leader, and contingent income above 15% as a warning — **Policy that fires by itself**Acquisition controls run on every Analyze and say who signs off: the practice leader, or a warning for the deal team.

The problem

An acquisition programme runs on spreadsheets nobody can audit

A broker buying agencies sees a new book export every few weeks: a different management system, different column names, a seller’s revenue figure, a multiple. The analyst rebuilds retention and concentration by hand, the memo quotes numbers nobody can trace, and the integration plan starts the week after close.

Price

A multiple on the wrong revenue

The seller’s revenue includes contingent income and fees the book does not support. Pay the multiple on it and the gap is paid for twice.

People

The book walks with a producer

One producer holding a quarter of the revenue, without a signed retention agreement, is the largest single risk in a small-agency deal.

Carriers

Appointments found too late

Premium with carriers the acquirer is not appointed with has to be appointed or moved, and the renewals do not wait for close.

Record

A memo nobody can trace

The investment committee approves a number. A year later, nobody can say which rows it came from or who agreed to the assumptions.

The product

Pre-deal diligence and post-close integration in one workbench

The book, the agents, the controls and the record — on one page per agency.

Book engines

Retention, growth, concentration, revenue mix, renewals and cross-sell, computed from the seller’s rows.

Learn more

Cited findings

Every figure opens the exact book rows behind it, with their row numbers in the seller’s file.

Learn more

Ten agents

Intake, retention, concentration, revenue, cross-sell, matching, carriers, plan, memo and supervisor.

Learn more

Prepared decisions

Seller requests, retention agreements and appointment requests, ready for Accept, Reject or Ask.

Learn more

Rules Studio

Acquisition policy as rules a practice leader can read, test on a real deal and publish.

Learn more

Decision record

Who agreed to what, on which figures and when — signed, and replayable for an auditor.

Learn more

IC memo

The valuation-support memo from approved findings only, with the arithmetic on each revenue base.

Learn more

Day 100

The plan becomes tasks; the retention tracker follows every client by producer and tier after close.

Learn more

Take the product tour

How an answer is reached

From the seller’s export to a signed decision

Nothing about the price comes from a model. Every step after the import is something you can open and check.

Import

The seller’s CSV or XLSX, row by row onto one schema. Every row keeps its number.

Engines compute

Every figure, with the rows it cites. The model never computes one.

3AI

Agents reason

Ten agents explain what the figures mean for the acquirer.

Grounding check

An answer carrying a number not in the engines’ facts is thrown away.

5AI

Prepared

Seller requests, agreements, appointments, the plan and the memo — never sent.

A person decides

Accept, reject or ask. Controls refer what is outside policy to the practice leader.

Model switched off or failing: the engines’ findings stand on their own, and every prepared decision says it reasoned by rules.

Every answer, approval and sign-off is signed on the decision record. The full architecture is on How it works.

The Acquisition Workspace

One page per agency, from screening to day 100

Findings, Book, Producers, Carriers, Clients, Integration and Memo — seven zones on one page, the stage strip across the top, and the acquisition controls in the rail. Every citation opens the book rows behind it.

The agents

Ten acquisition agents, each with one job and a hard limit

Book review, integration fit, memo and controls. Every action an agent can propose is capped at ask first: what goes to a seller, a carrier or an investment committee is a person’s decision.

Book intake

Data room against the checklist; data quality row by row. Drafts the request to the seller.

Retention

Client and revenue retention by producer and cohort; organic growth.

Concentration

Carrier, producer and client shares; key people. Prepares retention agreements.

Revenue quality

Recurring vs contingent; what renews in 90 days; book against the seller’s revenue.

Cross-sell

Single-line commercial clients and the acquirer’s lines they do not buy.

Client match

Exact, website, InsightPulse company graph, name similarity. A person confirms each.

Carrier mapping

Target carriers against the acquirer’s appointments. Prepares appointment requests.

Integration planner

The 100-day plan from the findings: clients, carriers, producers, systems, finance, people.

Memo writer

The valuation-support memo from approved findings only; quotes the engines’ figures.

Supervisor

Runs the acquisition controls; prepares the practice leader’s sign-off when one fires.

Two agents in the catalogue: the concentration analyst, which proposes retention agreements for key producers and never invents a share, and the retention analyst, which never invents a retention rate — **One job, one limit**Each agent says what it reads, the one action it may propose, the level it starts at, and what it never invents.

The autonomy dial for the concentration analyst: propose retention agreements for key producers, set at L1 draft, with L2 ask first available and L3 act-and-tell and L4 act locked — **The dial, with a ceiling**L1 draft and L2 ask first are open; act-and-tell and act are locked for every acquisition action.

Controls

Acquisition policy as rules a practice leader can read

Carrier concentration over 35%, a producer over 25% of revenue, book revenue 5% below the seller’s — written in Rules Studio, tested on a real acquisition, fired on every Analyze.

Rules Studio: the acquisition controls that fired most in the last 30 days, each with its when and then in plain words — more than 20% of premium with carriers the acquirer isn't appointed with, one producer over 25% of revenue, book revenue more than 5% below the seller's — **Eight controls ship with it**The busiest in the last 30 days, each readable as a sentence.

A rule in the editor: in plain words, when book revenue against the seller's is less than minus 5 percent, refer to a lead; the When condition and the Then action — refer to a lead, with the reason and the practice leader as recipient — **When, then, to whom**Book revenue more than 5% below the seller’s is referred to the practice leader, with the reason they will read.

The broker suite

Part of a family of broker workbenches on one platform

Shared sign-in, roster, decision record, Rules Studio and agent runtime. Company intelligence comes from InsightPulse 360 through a port, not a copy.

InInsightPulse 360Company profile, leadership, news and the parent-company link

InInsightXtractData-room documents read to text

SharedInsightDueDiligenceThe same diligence core: workspace, playbook, cited findings

InsightRollupAgency acquisitions: the book, the agents, the controls and the record

After closeInsightPlacementThe renewal calendar becomes the first renewals worked

After closeInsightProofingPolicies on the moved business checked before the client sees them

After closeInsightAdvocateThe acquired clients’ claims

Governance & deployment

Your cloud, your database, your model account

InsightRollup runs inside your environment on your PostgreSQL, with the model in your own account. Book exports are kept in your object store, never on local disk; every agent step is traced with its model, tokens and cost.

Grounded figures

The model explains; the engines compute. An answer with a number not in the cited facts is discarded and the engine’s finding stands.

A signed decision record

Every answer to a prepared decision, every finding approved for the memo and every control sign-off is signed as the person who made it.

Scoped by deal

A deal lead sees the acquisitions they work; practice leaders and compliance see the programme. Rules and the dial are for leads only.

Security & deployment

From the blog

Worked examples

Acquisitions followed from the seller’s book export to the decision, on the illustrative book.

Use case

The seller’s revenue the book won’t support

An asking multiple of 2.9×, and a book 8.2% short of the seller’s figure — found before the price is agreed.

October 29, 2026 · 14 min read Read article

Use case

Two producers, 55% of the revenue

Key-person risk turned into retention agreements as a condition of signing.

October 29, 2026 · 14 min read Read article

Use case

Day 40 of the integration

97.1% of revenue retained, six tasks done, five overdue — and the plan says which.

October 29, 2026 · 13 min read Read article

All posts

FAQ

What corporate-development leaders ask first

Does it value the agency for us?

No. It computes what the book supports — revenue, recurring revenue, the asking multiple applied to each — and shows the arithmetic. The price is your committee’s decision, recorded as theirs.

What if the seller’s export is messy?

Columns are mapped onto one schema on import and the data-quality engine flags rows with no client code, no producer, impossible commission or reversed dates. The intake agent drafts the request for a corrected export.

Can an agent contact a seller or a carrier?

No. Requests, letters and appointment applications are prepared for a person. Every action is capped at ask first whatever the dial says, and change-of-ownership letters run as a dry run until a person sends them.

Does it work after close?

Yes. The integration planner’s 100-day plan becomes tasks with owners and dates, and the retention tracker follows each client’s transfer by producer and client tier until day 100.