Demonstration — recorded run of a sample CIM · identities redacted · no live data processed
DealScreen
Agentic deal screening

Screen a CIM in minutes,
not an afternoon.

Drop a confidential information memorandum. A deterministic pipeline does the parsing, extraction, and scoring you can check — and agents do the two things that actually need judgment: researching the market and reviewing their own work. Out comes an IC-ready screen memo, scored to your box.

Drop a CIM here
PDF up to ~100 pages · or click to browse
Sample: Project Peach — a restoration-services franchise, 45-page broker CIM (redacted)
2Agents, used deliberately
MinutesVs ~4 hrs manual
3Issues the critic caught
HumanOwns the verdict
The architecture

A pipeline for the facts. Agents for the judgment.

Most of screening is deterministic work you want to be able to verify — parse, extract, score against the box. We use agents in exactly two places where iteration and judgment earn their keep. No agents for the sake of agents.

Deterministic pipeline · verifiable, no model judgment
Ingest & parse Extract & recompute Score to the box
Agents · used only where they pay off
🔎 Research agent + Draft Review Refine
Orchestrated run · recorded

Screening in progress

Project Peach — CIM (redacted).pdf · 45 pages
Deal Screening Orchestrator
Planning run…
agent activity log
Investment screen memo · Confidential · Generated by DealScreen

Project Peach — Restoration Services Franchise

Full-service disaster restoration & reconstruction contractor · Major Southeast metro · Franchisee of a national restoration brand · Founded 2001 · Screened from broker CIM (identifying details redacted).

3Pipeline steps
2Agents
4Research sources
3Issues auto-corrected
Review gate passed
!

Verdict: CONDITIONAL — inside the deal box on the numbers; franchise control & carrier concentration cap the platform thesis

A genuinely on-thesis lower-middle-market target — fragmented, buy-&-build sector, founder-owned, cash-flow positive. Two structural issues keep it from being a clean control platform. Advance to IC only if the two gating conditions below can be met in diligence.

Confidence: Medium-High·Downgraded from "Advance" by the review loop·2 HIGH flags open

Fit against the deal box

Pipeline · scoring

Scored against the deal box — < $50M revenue · $1–5M EBITDA (cash-flow positive) · healthcare / consumer / business services · fragmented & add-on-rich · control ownership with management co-investment. The box is a config file (deal_box.yaml); point it at another fund's criteria and the same pipeline re-screens.

CriterionResultEvidence from CIM
Revenue < $50MPASS$11.0M FY22 (recast), $13.5M FY23E — comfortably inside.
EBITDA $1–5M, cash-flow positivePASS$1.47M FY22 / $1.81M FY23E · positive free cash flow. Lower third of the range, margin 13.4% and expanding.
Target sector (healthcare / consumer / business services)PASSBusiness services — insurance-funded property restoration. Squarely inside the mandate.
Stable-to-positive demandPASSNon-discretionary, claims-driven demand — recession-resistant. Matches "stable to positive underlying demand trends."
Fragmented & add-on-rich (Buy & Build)CAUTIONCategory is highly fragmented and roll-up friendly — but the franchise agreement constrains brand control and M&A flexibility, blunting the buy-&-build engine.
Control ownership attainableFLAGFounder will sell 100%, but franchisor consent-to-transfer means control is encumbered — a gating item for a control fund.
Management partnership & co-investmentCAUTIONFounder (65) is retiring; senior PM layer stays. The box requires management co-investment — a rollover/partner isn't yet identified.
Conservative structure / low-risk profileCAUTIONStrong NWC ($2.8M), clean S-corp, no litigation — but carrier concentration ~68% and a related-party lease cut against the low-risk profile. Reconciled to Appendix B by the review loop.

Company snapshot

Pipeline · extraction
Business
Water / fire / storm restoration, reconstruction, contents & mold — residential (80%), commercial (16%)
Scale
30 FTEs · 23 active carrier/client accounts · 20,000 sq ft leased facility
Financials
FY22: $11.0M rev · $1.47M EBITDA (13.4%) · FY23E: $13.5M / $1.81M (recomputed from source tables)
People
Senior PM layer in place, expected to remain post-close — partially mitigates founder key-man risk

Risk flags

Pipeline · scoring
HIGH
Carrier concentration: loss of the top carrier program (~27%) would erase most of EBITDA. Diligence carrier scorecards & program tenure.
HIGH
Franchise agreement: transfer consent, territory limits, royalty load & renewal terms determine whether a control platform thesis is even executable.
MED
Revenue quality: percentage-of-completion accounting on reconstruction jobs; verify WIP schedule and normalize for CAT-event years.
MED
Related-party lease + owner-operator dependency during the transition window.

Market context — gathered live, external to the CIM

Agent · research

The research agent ran targeted queries to sanity-check the CIM's story against the outside world — the kind of context a deal team would otherwise pull by hand. Figures below are illustrative for this demo; the pattern is the point.

MARKETLarge, fragmented, insurance-funded. US property restoration is a multi-billion, non-discretionary category — demand is driven by claims, not the economy. Highly fragmented locally, which is exactly what makes it buy-&-build territory.
COMPSRestoration platforms trade ~6–9× EBITDA; single-unit franchisees typically clear at a discount to independents of similar scale, reflecting the brand/royalty overhang. Frames a disciplined entry multiple.
FRANCHISETransfer-consent and territory rights are recurring blockers when PE buys franchised units — corroborates elevating the franchise agreement to a HIGH, gating flag rather than a footnote.
CARRIERManaged-repair / carrier-program dependence is the sector's signature risk. The concentration in the CIM is structural, not idiosyncratic — which raises the bar on diversification credibility.

Review & refinement — what the agents caught

Agent loop · review → refine

Before the memo was finalized, a review agent audited the draft for unsupported claims and math errors; a refinement agent resolved each finding against the source. This loop is why the verdict you're reading is not the verdict the pipeline first drafted.

C-01Overstated growthHeadline CAGR includes a CAT-event year
Review agent flagged

"23.9% revenue CAGR (FY20–FY22) spans FY21, an above-trend storm year. Presenting it as organic durability overstates the run-rate."

Refinement agent fixed

Added the normalization caveat, noted organic CAGR ≈15%, and downgraded the growth read in the snapshot.

C-02Figure mismatchConcentration % didn't tie to the appendix
Review agent flagged

"Draft cites Top-5 carriers ≈ 71%, but the Appendix B carrier table sums to 68% for named carriers. Reconcile before IC."

Refinement agent fixed

Corrected to 68% named / ≈71% incl. affiliated programs, and tagged the 3% delta as a diligence item.

C-03Unsupported verdictDraft recommended "Advance to IC"
Review agent flagged

"Two HIGH flags (franchise consent, carrier concentration) are unresolved gating items. A clean 'Advance' isn't supported by the evidence."

Refinement agent fixed

Downgraded the verdict to CONDITIONAL with two explicit, testable gating conditions.

Recommendation

Refinement agent

On-thesis on nearly every dimension — size, sector, demand stability, fragmentation — with real momentum and margin expansion. But two structural issues cap it as a control platform. Advance to IC only if (1) the franchisor grants multi-territory expansion rights and consents cleanly to PE control, and (2) a management co-investment / rollover partner is credible and second-tier carrier diversification is achievable within 24 months. Otherwise, revisit as an add-on to an independent (non-franchise) restoration platform, where the local reputation and carrier relationships are worth more than the franchise flag.

Auto-generated diligence questions

Memo composer
  • Franchise agreement: royalty %, term/renewal, territory definition, transfer-consent standard, franchisor ROFR?
  • Carrier programs: tenure, scorecard performance, TPA / managed-repair dependencies, pricing power by program?
  • Management co-investment: will the founder or PM layer roll equity, and on what terms?
  • Revenue normalization: how much of FY21–22 growth is storm/CAT-event driven vs. genuine share gains?
  • WIP schedule and revenue-recognition policy on long reconstruction jobs — any margin fade at completion?
  • Related-party lease: market-rate comp and willingness to sign a long-term lease at close?

If acquired: phase-aware AI roadmap — sequenced to the hold period, not the technology

Memo composer

Stabilize

First 100 days
  • AI intake agent on emergency lines — capture every FNOL job while staff is in transition
  • AR follow-up agent on carrier receivables (protects NWC)
  • Estimate-prep assist: photos + notes → draft scopes for adjuster submission

Standardize

Months 4–18
  • Job-documentation agent: field photos → carrier-compliant reports
  • WIP + margin dashboards from job-costing data
  • Carrier SLA monitoring — defend the concentration before diversifying it

Scale

Year 2+
  • Cross-territory crew scheduling & CAT-surge capacity planning
  • Add-on integration playbook: a repeatable agent stack per acquisition
  • Commercial BD agent: target second-tier carriers & property managers