Financial workflows often begin with documents, pass through calculations and policy rules, and end with a person accepting responsibility for a decision.
Voyantt builds software around that chain of evidence. We help lending and financial teams extract information, apply rules, route exceptions, connect payment systems, and preserve a decision history that can be reviewed later.
How We Decide
The calculation is rarely isolated. A reviewer needs to know:
When those answers live across spreadsheets, email, shared folders, and human memory, automation can make the process faster without making it safer.
How We Contribute
We build pipelines that read PDFs and other financial documents, extract structured fields, retain confidence, and keep the result connected to its source. The interface should let the reviewer move from a figure to the relevant evidence without searching the file again.
Important calculations belong in testable functions, not in hidden spreadsheet cells or model output. Policy rules should be versioned and configurable where change is expected, with a record of which version evaluated each file.
AI or rules can identify an issue. A product still needs to put that issue in front of the right person, show the evidence, record the action, require a reason where appropriate, and update the workflow state.
Our work includes Payment Intents, subscriptions, off-session charging, credit checks, pay-later decisions, refunds, invoice creation, credit memos, webhook handling, and reconciliation. We design the financial states around the API call, not only the checkout action.
For processes where silent edits are unacceptable, we can use append-only records and hash chaining to make subsequent changes detectable. The exact control should match the regulatory, contractual, and operating requirement of the system.
LoanCite is a feature-complete prototype for non-QM mortgage underwriting. It reads borrower documents, associates extracted values with page-and-line provenance, runs configurable guideline rules, and sends exceptions to an underwriter for Approve, Override, or Reject with a required reason.
The calculation engine has 146 tests. The prototype also produces a binder containing the document index, sourced figures, rule results, exception decisions, and an audit excerpt.
It is a prototype with a live demo—not a production lender deployment. We state that clearly because financial software should be evaluated on evidence rather than implication.
Read the full case study →The Frontier Dental B2B platform includes ERP-synchronised credit limits, pay-later checks, invoice creation, and the application of payments and credit memos across accounts-receivable documents. The industry is different, but the integration and financial-state problems are directly relevant.
Read: Frontier Dental →
First Steps
Start a conversation
No pitch deck. We will ask about the users, the current system, the constraints, and what cannot go wrong. If there is a fit, the next step is a written proposal covering the approach, scope, risk, timeline, and cost.