AI for Private Credit Underwriting: Direct Lending Analysis, DSCR, and Loan Structuring with Claude (2026)

Private Credit Underwriting and AI

Direct lending underwriting requires a comprehensive credit analysis: EBITDA quality assessment (add-back scrutiny), leverage test against market standards, DSCR analysis across scenarios, loan structure design (covenants, amortization, PIK features), and ongoing monitoring templates. Claude with FinSkilz accelerates the underwriting memo — from deal screening to credit committee presentation.

Can private credit funds use Claude for underwriting? Yes, with a clear boundary: Claude doesn't replace the underwriting decision or the credit committee's judgment, but it accelerates every analytical and documentation step feeding that decision — EBITDA add-back review, DSCR/leverage calculations across scenarios, covenant drafting, and PIK economics modeling. The underwriter still makes the credit call and signs the memo; Claude removes the hours of spreadsheet and drafting work in between.

How Fast Is Private Credit Actually Adopting AI?

Adoption-rate claims in this space vary enormously depending on who's asking and how "using AI" is defined — most coverage cites a single number without saying so. Here's what's actually out there, reconciled rather than cherry-picked:

• The clearest survey data point: an Armentum survey, reported via Debtwire/Creditflux, found AI use among private credit originators rising from roughly 20% to about 50% between survey waves, with underwriting-specific AI use roughly doubling within that — the strongest, most specific figure found for this niche (source: ionanalytics/Debtwire via Creditflux).
• The funding signal is real and recent: EnFi raised a $15M Series A in February 2026 specifically for AI agents in bank credit decisions; PactFi raised a $25M Series A in March 2026 to scale private-credit operations — both 2026 rounds, not legacy fintech.
• Named point solutions are multiplying: F2, Arc, and Qualitas are all actively building AI underwriting-support products for private credit as of 2026, alongside EnFi and PactFi above — five distinct funded entrants in roughly a year is a real signal of category formation, not a single vendor's marketing claim.
• Where the data gets soft: individual funding-round sizes sometimes conflict across outlets (one 2026 seed round for a private-credit AI vendor was reported as both $14M and $24M by different publications, with neither traceable to a primary filing), and broader financial-services adoption surveys (not private-credit-specific) report figures as high as 90%+ depending on how loosely "any AI use" is defined. Treat any single adoption percentage you see cited elsewhere as directional, not precise — the Armentum 20%→50% figure above is the most specific and sourceable number for this niche specifically, which is why we lead with it.

The practical takeaway for a credit team evaluating whether to adopt AI in underwriting: you're not early anymore, but you're also not late. Roughly half of originators report some AI use as of the most recent survey wave, concentrated in exactly the workflow this page covers — EBITDA add-back review, DSCR/leverage calculation, and covenant drafting — rather than in the credit decision itself, which stays with the underwriter.

EBITDA Quality and Add-Back Analysis

• "Analyze EBITDA add-backs for middle-market software company seeking direct lending: management EBITDA $18.5M. Proposed add-backs: (1) Restructuring charges $1.8M — review documentation: one-time office closure, non-recurring, ACCEPT; (2) SBC $2.2M — non-cash, ACCEPT; (3) Cost savings from offshore transition $3.5M — 80% completed, run-rate established, ACCEPT $2.8M (80%); (4) Revenue synergies from acquisition $4.5M — speculative, not completed, REJECT; (5) Management fees $0.6M — standard PE sponsor fee, ACCEPT. Total accepted add-backs: $1.8M + $2.2M + $2.8M + $0.6M = $7.4M. Add-back as % of base: $7.4M / $18.5M = 40% — EXCEEDS 25% market standard cap. Negotiate cap at $4.6M (25% of base)."
• "Test EBITDA sustainability: compare LTM EBITDA to 3-year average. LTM: $18.5M. Year -1: $15.2M. Year -2: $13.8M. Year -3: $11.2M. CAGR: +18.1%/year — consistent organic growth. Q4 run rate: $19.5M annualized — momentum positive. Assess customer concentration risk: top 5 customers = 54% of revenue ($12.5M). Largest single customer: 18% ($4.2M). EBITDA sensitivity to loss of largest customer: -$4.2M × contribution margin 75% = -$3.2M EBITDA impact = 17% reduction. Adjusted leverage for customer concentration risk."

Leverage and Coverage Analysis

• "Compute credit metrics for this direct lending opportunity: LBO purchase price $135M (7.3x adj. EBITDA $18.5M). Capital structure: First Lien Term Loan $90M (4.9x leverage), Revolver $15M (undrawn), Equity $45M (33% equity cushion). Cash interest: SOFR ($5.25%) + 575bps = 11.0% on $90M = $9.9M. Mandatory amortization: 1% per year = $0.9M. DSCR: $18.5M EBITDA / ($9.9M interest + $0.9M amortization) = $18.5M / $10.8M = 1.71x. Compare to market standard (private credit minimum DSCR: 1.30x). Current DSCR 1.71x provides 31% cushion. At what EBITDA does DSCR hit 1.30x? Break-even EBITDA = 1.30x × $10.8M = $14.0M (24% decline from LTM — meaningful buffer)."

Loan Structure and Covenants

• "Design the covenant package for this direct lending transaction: given 4.9x first lien leverage and 1.71x DSCR, propose: (1) Financial maintenance covenants: maximum first lien leverage 5.50x (tested quarterly, 60bps of headroom), minimum DSCR 1.20x (tested quarterly), minimum liquidity $5.0M (restricted cash + revolver availability); (2) Reporting: quarterly financial statements within 45 days, annual audit within 90 days, monthly management accounts within 30 days; (3) Excess cash flow sweep: 75% sweep of excess cash flow after permitted debt service; (4) EBITDA add-back cap: 25% of LTM EBITDA. Include equity cure: 2 cures of 4 quarters in any 12-month period."
• "Analyze the PIK toggle feature: borrower requests PIK election for cash interest: in any quarter where liquidity falls below $8M, borrower may elect to PIK all or a portion of cash interest (adding to principal instead of paying cash). PIK rate: 13.0% (vs cash rate 11.0% — 200bps PIK premium). Model PIK for 2 quarters at current balance $90M: Quarter 1 PIK: $90M × 13.0% × 0.25 = $2.925M added to principal → $92.925M. Quarter 2 PIK: $92.925M × 13.0% × 0.25 = $3.020M → $95.945M. Lender's perspective: PIK is compounding and increases loss severity in default — requires larger discount to par on purchase price."

Loan Monitoring Template

• "Design a quarterly loan monitoring template for a 15-loan private credit portfolio: for each credit: (1) Borrower name and loan summary (size, rate, maturity); (2) Q3 2025 EBITDA (LTM) vs underwriting case; (3) Current leverage vs covenant; (4) Current DSCR vs covenant; (5) Liquidity (cash + revolver availability); (6) Covenant headroom % (leverage headroom / leverage); (7) Performance vs underwriting assumptions (revenue growth vs plan, EBITDA margin vs plan); (8) Risk rating (1-5 scale: 1=watch list, 5=performing in-line); (9) Notable developments (management change, customer loss, etc.). Export as structured table for quarterly credit committee."

Where to Start

For a new deal, start with the EBITDA quality analysis. Paste the borrower's management EBITDA and all proposed add-backs, and ask Claude to evaluate each add-back against private credit market standards — which ones are acceptable, which need reduction, and whether the total add-back exceeds the typical 20-25% cap. Once you have the adjusted EBITDA, the leverage and DSCR analysis follows directly from the proposed capital structure.