VantedgeAI

The work being sold is reading. Prospectuses, credit agreements, data rooms and financial statements go in, structured data, benchmarked risk factors, built models and valuation summaries come out, and the retrieval sits on ontology based frameworks and vector search rather than keyword lookup. There is no residual product if the models are removed, only a document store, and the company's own framing is that manual data handling is the thing being eliminated.

Review is built into the workflow as a product feature rather than assumed. Teams invite colleagues to review, comment and refine outputs before a summary or memo goes to an investment committee or a client, so the artefact that leaves the building has passed a named human step. The gap is upstream: nothing describes what happens when an agent extracts a figure incorrectly and the reviewer is checking prose rather than re reading the source document.

The stated controls are the right kind and are unquantified. The platform is described as parsing, validating and organising submitted documents into templates, and the company claims auditability, transparency and custom training as differentiators, but no extraction accuracy rate, error taxonomy, validation methodology or evaluation summary is published.

That matters because the failure mode here is quiet: a covenant misread from a credit agreement or a figure lifted from the wrong column flows straight into a built financial model and looks authoritative on the other side.

Descriptions are detailed and proof is thin. No fund is named, no extraction accuracy, time saved or deal throughput figure is published, and the strongest available line is an invitation to join an unnamed elite group of credit teams. The founders ran a credit hedge fund for fifteen years, which is credible domain grounding and is not evidence about the product.

The cross client boundary is addressed in architecture rather than in a policy sentence, which is the stronger form. Multi tenant deployment with isolated workspaces, secure data isolation, role based access, custom training on a client's own material, and the option to run the workload inside the client's own cloud environment together answer the question a private credit fund asks first, which is whether its deal documents can reach a competitor through the model. What is not published is a plain statement about whether client documents are ever used to improve shared models, so the answer is implied by the architecture rather than stated.

Worth reading the attestation scope closely, because it is precise about what it does not cover. The published trust criteria are security, availability and confidentiality, and privacy is not among them. Confidentiality protects the client's deal material, which is what a fund cares about, and it is a different commitment from privacy handling for the individuals inside the documents, which matters once merchant cash advance underwriting brings small business owners and their personal data into scope.

It publishes a trust centre and enumerates rather than asserts, naming a service organisation control type one attestation across security, availability and confidentiality. Read the type carefully, because the distinction is the whole point of the axis: a type one report is an auditor's opinion on whether controls are suitably designed at a point in time, while a type two tests whether they operated effectively over a period. This is the weaker of the two and it is published openly, which is still ahead of the vendors that assert certifications without naming any.

No licence is required to sell diligence tooling and none is claimed. The merchant cash advance lane is where the regulatory position deserved a sentence and does not get one, since commercial financing disclosure regimes now apply in several states and the product's economics have drawn sustained supervisory and enforcement attention. A tool underwriting those advances sits close to that perimeter without acknowledging it.

Two very different exposures sit under one roof and neither is addressed. Institutional credit analysis carries little protected class risk, so silence there is defensible. Merchant cash advance underwriting is a different matter: the applicants are small businesses, frequently owner operated, and commercial credit is inside the equal credit opportunity framework, with small business lending data collection rules bringing further scrutiny. No fairness testing, adverse action handling or outcome monitoring is published for that side of the product.

Nothing published describes what happens when an extracted term is wrong and a fund prices a deal on it. No warranty, accuracy commitment, service level or remedy appears anywhere, and the collaborative review step, useful as it is, quietly relocates responsibility to the analyst who approved the output. On the merchant cash advance side the affected party is a small business that never sees the platform and has no route to challenge what it concluded.

The marketplace makes this axis more consequential here than for most vendors and the disclosure has not kept up. Third party agents are described as hand picked and integrated into the workflow, which means components the vendor did not build are processing client credit documents, and none of them are named, nor is there any published statement of what they are permitted to do with the data, how they are assessed, or what a client is told when one is added. No underlying model or provider is named either.

Integration here means agents rather than systems, and it is a deliberate architecture: one dashboard hosting the vendor's own expert agents plus hand picked third party agents that complement the credit workflow, with new agents deployed continuously. Documents arrive in the formats credit teams actually use, spreadsheets, portable documents and presentations. What is absent is the institutional plumbing, since no portfolio management, loan administration, data room or customer relationship system is named.

Among the more specific deployment statements in the index. Single tenant and private cloud environments are offered with full data isolation and encryption, multi tenant workspaces are isolated, and the company describes enabling clients to run workloads inside their own cloud, which is the arrangement a fund with confidentiality obligations to its borrowers will want. No geographic region or residency commitment is named, which is what keeps it from an A.

A free trial with no card required is published, which tells a prospective buyer the product can be evaluated without procurement, and no rate, tier or seat basis appears anywhere. The agent marketplace raises a second unanswered commercial question, since nothing states whether third party agents are bundled, priced separately or revenue shared.

Coverage is specific rather than generic: private credit funds, hedge funds, merchant cash advance lenders, other financial institutions and private investors, with two named specialisms in private credit and merchant cash advance. Merchant cash advance underwriting is an unusual second lane for a credit research tool and is close to empty in this index, which makes the coverage claim more interesting than its size would suggest.