Quillon Claims AI

Advanced

0 words

Company overview

Quillon Claims AI sells software that reads, sorts and settles insurance claims using large language models. Property and casualty insurers pay an annual platform fee plus a fee per claim processed, with contracted annual minimums. Quillon was founded in 2019, is backed by growth-equity investors (it raised a $60M Series C in 2024 at a $780M post-money valuation) and has 142 insurer customers. It is growing quickly but has never generated positive free cash flow.

Quillon's costs are shaped by AI. Every claim it processes runs through models: about 65% of compute spend is on a major cloud provider's servers, where Quillon runs its own fine-tuned models under a three-year committed-spend contract signed in January 2026, and about 35% is paid per use to a third-party foundation-model provider. Compute cost per claim fell for two years as model prices dropped, then rose in the last twelve months as Quillon launched multi-step "agentic" claim handling that uses far more computation per claim.

Quillon is asking for a $100M senior secured term loan, sized at about 1.0x annual recurring revenue (ARR), to repay $35M of venture debt and add about $61M of cash. The loan is interest-only for three years and has two financial covenants: minimum liquidity and maximum debt to ARR. Management's plan reaches positive free cash flow before interest in FY2028.

The IC question: is this a lender's credit or an equity bet? Is growth durable as retention slows and the largest customer approaches renewal; does each claim get cheaper or more expensive to serve; how long does the cash last if the plan slips; and what would the lender recover if Quillon had to be sold?

Your memo should answer

01Would you make this loan on these terms? State your recommendation and the conditions or changes you would require.
02How durable is growth? Use the ARR bridge, retention trends, cohorts and customer concentration, including the largest customer's renewal.
03Is Quillon getting cheaper or more expensive to serve? Assess gross margin after compute, compute cost per claim, unit economics and the cloud commitment.
04Build a downside: the largest customer leaves at renewal, new customer ARR halves, and compute rises 5 points of revenue. How long does the cash last, and when does each covenant trip?
05What would the lender recover if Quillon had to be sold, and how does the risk of its product being displaced by foundation-model providers affect that?
06Is the pricing enough for the risk, and what would you change in the terms?

Financials ($M)

	FY2023	FY2024	FY2025	LTM Jun-26
ARR, end of period	38	61	88	102
Revenue	31	52	77	92
Compute cost	7.4	11.4	16.2	21.2
Gross margin, after compute	58%	61%	63%	62%
Sales and marketing	21	27	34	38
EBITDA	(34)	(29)	(21)	(18)
Free cash flow	(41)	(36)	(27)	(24)
Cash, end of period	52	76	49	37
Claims processed (millions)	16.5	30.0	47.6	52.0
Net revenue retention	118%	121%	114%	109%
Gross revenue retention	93%	92%	91%	91%

Term sheet

Facility: $100M senior secured term loan, fully drawn at closing; first lien on all assets including intellectual property and customer contracts; guarantees from all subsidiaries
Tenor and amortization: 5 years; interest-only for 36 months, then 5% a year paid quarterly; balance at maturity
Pricing: SOFR + 650 bp, cash pay (assume SOFR of 4.0% for analysis); 2.0% upfront fee; warrants for 0.5% of fully diluted equity
Pricing context: Teaching assumption: recurring-revenue term loans to venture- and growth-backed software companies recently priced at SOFR + 600 to 800 bp with 1–2% fees and warrants; sponsor-backed recurring-revenue loans at SOFR + 525 to 650 bp
Financial covenants: Minimum liquidity (unrestricted cash) of $20M at all times; maximum total debt to ARR of 1.25x, tested quarterly; from the first quarter of FY2029 the ARR test is replaced by maximum total debt to EBITDA of 6.0x
Other terms: Prepayment premium of 3%, 2% and 1% in years 1–3; no other debt except capital leases up to $5M; change-of-control put at par plus premium
Use of proceeds: Repay $35M of existing venture debt, pay $4M of fees and expenses, and add $61M of cash to the balance sheet

ARR bridge, Jun-25 to Jun-26 ($M)

Opening ARR (Jun-25): 76.0
New customers: +19.0
Expansion from existing customers: +14.0
Contraction: −2.5
Churned customers: −4.5
Closing ARR (Jun-26): 102.0
Customers: 142, of which 42 were won in the last twelve months
Largest customer: 14% of ARR (a national property and casualty insurer); contract renews in March 2027
Top five customers: 41% of ARR

Cohort net revenue retention (ARR of each annual customer cohort as a % of its first-year ARR)

2021 cohort: Year 1: 100% · Year 2: 128% · Year 3: 141% · Year 4: 146% · Year 5: 148%
2022 cohort: Year 1: 100% · Year 2: 124% · Year 3: 133% · Year 4: 137%
2023 cohort: Year 1: 100% · Year 2: 119% · Year 3: 125%
2024 cohort: Year 1: 100% · Year 2: 112%
2025 cohort: Year 1: 100%

Unit economics, LTM Jun-26

Sales and marketing spent on new customers: $26M of the $38M total; 42 new customers
Average first-year ARR of a new customer: $0.45M
Net new ARR: $26.0M (opening $76.0M to closing $102.0M)
Net burn: $24M (free cash flow)

Compute cost and contracts

Compute cost per claim processed: FY2023: $0.45 · FY2024: $0.38 · FY2025: $0.34 · LTM Jun-26: $0.41
Why it rose: Agentic claim handling, launched in Q3 2025 and now about 30% of claims, uses about three times the computation of a standard claim; it is priced about 40% higher
Cloud provider (about 65% of compute): Three-year committed-spend contract from January 2026: $12M in 2026, $18M in 2027 and $24M in 2028 ($54M in total); unused commitment is paid as a shortfall fee; about $48M remains at closing
Foundation-model provider (about 35% of compute): Paid per use with no commitment; the provider can change prices on 90 days' notice; its per-use prices fell about 60% over the last two years
Competitive context: Two foundation-model providers have announced insurance-claims products; Quillon's defences are seven years of labelled claims data (licensed from customers), integrations with the main claims-administration systems, and audit trails insurers need for regulators

Management plan ($M; free cash flow is before interest on the new loan)

FY2026E: ARR 118 · revenue 104 · gross margin 63% · EBITDA −12 · free cash flow −14
FY2027E: ARR 140 · revenue 128 · gross margin 64% · EBITDA 2 · free cash flow −3
FY2028E: ARR 165 · revenue 152 · gross margin 65% · EBITDA 18 · free cash flow 12
First half of FY2026 (actual): Free cash flow −12

Sources and uses at closing ($M)

Sources: new term loan: 100.0
Uses: repay venture debt: 35.0
Uses: fees and expenses: 4.0
Uses: cash to the balance sheet: 61.0
Pro forma cash at closing: 98.0 (37.0 existing plus 61.0)

Valuation context (teaching assumptions)

Last equity round: Series C, April 2024: $60M at $780M post-money
Listed vertical software peers: About 4–7x forward revenue
Sales of distressed growth-stage software companies: Have cleared at about 1–3x ARR, lower where the product was being displaced

01 Recommendation0 / 50–200

Your call (approve, approve with conditions, or decline), the size and type of loan, and the two or three reasons that decide it.

02 Business and key risks0 / 150–400

What the company does and why its cash flow holds up, then the two or three risks that drive the outcome, ranked, each with evidence from the packet.

03 Financial analysis0 / 150–400

Pro forma leverage, coverage and free cash flow with the inputs shown, the trend behind them, and what they mean for the lender.

04 Downside case0 / 120–350

One plausible stress anchored in the company's history or a named shock, carried through to leverage, coverage, liquidity and covenant headroom.

05 Structure, terms and covenants0 / 100–300

Whether the terms protect the lender against your downside, pricing against the risk, and the changes you would ask for.

06 Conditions and conclusion (optional)0 / 30–150

The conditions your approval depends on and what would change your mind.

Saved in this browser as you type.

← Case page