Underwriting intelligence for MCAs and ISOs
The true-revenue layer for MCA underwriting.
Vyaso re-underwrites every bank statement in 90 seconds. It surfaces loan stacking, account kiting, and revenue inflation before funds wire — and gives committees a number they can defend.
Runs on top of Ocrolus · Heron Data · Validis · raw PDF or CSV uploads
Your parser tells you what each transaction is.
Vyaso tells you what they add up to.
The same statement, two ways of reading it.
What your parser sees
Total deposits: $158,000.00. Looks like revenue.
What Vyaso sees
Adjusted revenue: $48,000. One of four credits is real.
Why now
MCA underwriting got harder in 2024.
MCA default rates climbed through 2024 and into 2025 as pandemic-era expansion cohorts matured into repayment. Funders that lent against gross deposits are absorbing the delta now, in collections. Industry estimate¹
Loan stacking has gotten harder to spot. Lender deposits are routinely described to look like platform payouts or customer ACH. The keyword-only detection that worked five years ago no longer does. Industry estimate¹
AI-generated forged bank statements became trivial in 2024. Font drift, balance breaks, and metadata anomalies are signals a parser does not surface. They need a forensic layer. Industry estimate¹
The intelligence that worked five years ago does not work for the files arriving today.
Designed for how MCA portfolios actually default.
Speed
Per file, end to end.
Replace 60–90 minutes of manual review per applicant. Industry estimate¹
Accuracy
Typical adjusted-revenue delta.
The amount Vyaso strips out of gross deposits to land on a number you can defend in committee.
Vyaso benchmark across the pilot corpus to date. Per-file delta varies; some files show no delta, others exceed 50%.
Coverage
Detection layers.
Symmetric flows, graph cycles, window dressing, statistical anomalies, counterparty risk, loan stacking, account kiting, daisy-chain inflation, and PDF integrity. Every credit passes through all of them.
Built for both sides of the deal.
Funders re-underwrite every file. ISOs route every applicant to the funder whose box the real revenue actually fits.
For MCA funders
Re-underwrite every file in 90 seconds.
Stop funding stacks. Re-underwrite every file at the same speed: clean files clear in seconds, risky ones surface before the funds wire.
- Adjusted revenue you can defend in committee
- Loan-stacking and kiting flags ranked by severity
- Executive summary, paste-ready to credit memo
For ISOs and brokers
Submit smarter. Decline less.
Pre-screen every applicant before submission. Match the file to the funder whose box the real revenue fits. Protect your reputation by never sending a stacked merchant.
- True-revenue figure before you submit
- Stacking and kiting visibility on first read
- Funder-routing intelligence built on the same analysis underwriters trust
How a file moves through Vyaso.
Upload
Drop the bank statement PDFs or CSVs. Optional: company domain.
Auto-classify
Vyaso detects the merchant's industry and tunes the analysis automatically.
Re-underwrite
Every credit passes through nine detection layers and an agentic synthesis step.
Receive
Risk score, adjusted revenue, ranked flags, network graph, revenue waterfall, and an executive summary.
Median time end-to-end: ~90 seconds. ✓ verified
The five patterns we keep seeing.
Every MCA portfolio has them. Most underwriting tools don't.
| Pattern | What it looks like |
|---|---|
| Loan stacking | Multiple lender deposits arriving in a short window. |
| Account kiting | Self-transfers between the merchant's own accounts inflating deposit volume. |
| Circular B2B counterparty laundering | Money moving through three or more counterparties in a tight loop and returning to the merchant. |
| Daisy-chain inflation | The same flow stretched out over weeks, evading short-window detectors. |
| Forged statements | PDFs that look real but were generated outside bank software, with font drift, balance breaks, or telltale metadata. |
Each pattern maps to a dedicated detection layer. See the platform page for what catches what. →
Vyaso runs on the data your stack already produces.
If you already use Ocrolus, Heron Data, or Validis, Vyaso reads their output and re-underwrites it. Raw PDF and CSV upload also work today. API integrations with these parsers are available in the pilot program. Roadmap
Run Vyaso on your portfolio.
Free 30-day pilot. Bring 50–100 files. We'll show you what the model would have flagged, what would have been approved, and how the adjusted revenue compares.
No commitment. No setup fee.
Design partners
Working with funders and ISOs through the pilot program.
Logos shown after pilot completion and partner approval. Currently anonymized by mutual agreement.