Policy Check: automated credit-policy review of SME loan files

The core per-file check is already taken. Botminds sells exception detection with sign-off by delegated authority to credit leaders. Stack AI offers templates that return pass or fail per policy criterion and draft an exception memo. Underwriting copilots such as Kaaj and CredStack are close to adding it. Incumbents like Alogent AccuAccount, and vendors like Textmine, already cover the exception register. A plain "check the file against the policy" product would be competing on execution alone. The least contested piece is the policy change rerun. My searches found no product that versions a lender's written credit policy and re-tests the live book when the policy changes to list the loans that no longer comply. That makes it a reasonable wedge, because it needs more than a single-file agent: policy compiled into versioned rules, structured data extracted from every live loan and kept up to date, and a diff between versions. A new entrant could win by treating the policy itself as the product. That means turning the written policy into numbered, versioned rules, tying every verdict to a cited passage in the file, and keeping one audit trail that runs from file check to exception register to rerun. The buyer is still open too. Most existing tools target banks, credit unions, CRE lenders or merchant cash advance shops. Fast-moving SME-lending neobanks change policy often and face scrutiny from sponsor banks or regulators, and they are under-served. The risks are that LOS and underwriting copilot vendors can add policy versioning, and some fintech lenders build this in-house with consultancies. These searches were not exhaustive, so a closer look at Botminds, Textmine and the underwriting copilots is worth doing before committing.

• Botminds (Exception Detection and Approval Workflows)

  The closest match. It is sold to chief credit officers and loan operations leaders as a way to catch every policy deviation on the first read of a file, route exceptions by delegated authority, and record each one with a named approver. It overlaps with both Policy Check and the Exception register.

• Textmine

  Its content pitches credit policy exception management across lending portfolios: a register that records the approver, conditions and remediation owner, with trend reporting by product, sector and risk grade. That overlaps directly with the Exception register idea.

• Stack AI (Credit Policy and Exception Memo Agent and Loan File Review templates)

  Ready-made agent templates that check a loan file against credit policy criteria such as LTV, DSCR, borrower experience, approved markets and loan size. They return pass or fail for each criterion and draft an exception memo when one fails. A lender's own team could build the core Policy Check on this platform.

• Kaaj

  An agentic underwriting copilot for SMB lenders, backed by a $3.8M seed round. It has agents for document parsing, diligence, credit analysis and memo generation, with a human kept in the loop. Policy checks fit naturally inside its memo workflow.

• Alogent AccuAccount

  An established commercial-lending exception tracking system used by banks and credit unions. It tracks lending policy exceptions, document exceptions and collateral exceptions against thresholds, and is the incumbent for the exception register.

• CredStack

  AI agents that produce credit memos for credit officers to approve. This is next to the policy-check workflow.

• Draco AI

  Automates underwriting for small business lenders, starting with merchant cash advance. It could add policy-conformance checks.

• GroupBWT (custom build case study)

  A consultancy built AI agents that check policy compliance for a US fintech SMB lender handling about 2,500 applications a week, with analysts making the final decision. This shows fintech lenders also commission custom builds instead of buying a product.

Checked 2026-10-15