Saris AI
Saris AI builds agents that run back office work for banks and credit unions across the systems a lender already has: the origination system, the core, the document repository and the inbox. Its Document AI Agent extracts and verifies data from loan packages, handwritten forms and poor scans, the System AI Agent works in core banking, origination and back office platforms with or without an API, and the Inbox AI Agent triages messages and drafts replies for staff to send. Other agents calculate income across complex cases, pull data from outside sources such as Zillow, county tax portals and UCC filings, and log every action for examiners.
The institution sets its policies and approval points, and agents route exceptions and consequential decisions to staff with the evidence attached. Saris covers mortgage, consumer and commercial lending, post close quality control and risk and compliance reviews, with connections to Fiserv, Jack Henry and FIS cores, ICE Encompass and MeridianLink, OnBase and Outlook.
Vibrant Credit Union cut HELOC file review from 15 to 22 minutes to between 30 seconds and two minutes and cleared a 600 loan backlog in four days. Catalyst Corporate Federal Credit Union reports a regulatory filing workflow about 75% faster, with 99.8% field accuracy. Saris announced a $28.8 million Series A in June 2026.
Capability Axes
Capability grades
15 of 15 axes rated · 8 graded A or B
Agents are the operating product. Saris configures them around a bank's applications and procedures so they read documents, interpret policy and prepare data or actions across systems: a document agent extracts a value or prefills a form, a system agent reads and writes in existing banking software, and an inbox agent handles incoming context and drafts messages. Each depends on models interpreting unstructured information from whatever files arrive in a lender's queue.
The bank sets the workflow and keeps approval at meaningful decision points. An agent cross checks figures against policy and presents the case for sign off, exceptions arrive with the discrepancy, source and policy context, and inbox drafts wait in a queue for staff review before sending. The Audit and Monitoring Agent snapshots systems and timestamps every action, whether the agent works through a native connection or the screens staff use.
Each action is recorded and each exception links to the file evidence and the bank's policy. Catalyst's field and workflow accuracy are reported separately, at 99.8% and 94.2%, and the model risk documentation is stated to align with SR 11-7, NCUA guidance, OCC Bulletin 2023-17 and CFPB expectations. That documentation and the accuracy method go to institutions under a nondisclosure agreement during evaluation, and no test before deployment is described.
Vibrant Credit Union's HELOC team cut review time from 15 to 22 minutes per file to between 30 seconds and two minutes, cleared a 600 loan backlog in four days and gave each employee back about 2.5 hours a day. Catalyst Corporate Federal Credit Union reports about 75% less processing time on a regulatory filing workflow, one representative run falling from nine hours to 45 minutes, with 99.8% field accuracy and 94.2% workflow accuracy. Unnamed clients add a 90% cut in cost per lending workflow at a regional credit union and 70% of operational tasks automated at a $5 billion regional bank.
Agents are configured on each institution's own systems, policies and procedures, and the lender decides where sensitive information may move. The service terms let Saris collect diagnostic and usage metadata to improve the service, and the bank's agreement governs institutional data. How model training, supplier retention and deletion apply to each lender's loan files is not stated.
Bank data moves between document repositories, the origination system and the core through permissions the institution grants, encrypted in transit and at rest, with role based access and logged actions. Saris states GLBA adherence, and its service terms put the treatment of institutional data under the master agreement. Processing roles, retention by record type and the subprocessor chain for bank information sit outside the public materials.
Saris is SOC 2 Type II audited and names encryption, role based access and a full audit trail of agent actions. Full security and compliance documentation, the report included, goes to institutions in active evaluation under a nondisclosure agreement.
Bank staff stay accountable for lending and compliance decisions, and the platform prepares the evidence and checks behind them, with an audit trail built for examiners. Catalyst's regulatory filing workflow runs under human oversight, but the filing is unnamed and direct submission by Saris is not described, nor any supervised authorization for the service. Saris lists associate membership in a state bankers association.
Institutions define the policies agents apply, and staff check exceptions with the source information and rule in front of them. Logs of agent actions and governance documentation are kept for bank review. Agents check supplied documents and calculations while staff keep the credit call, and no demographic evaluation or fairness testing across borrower groups is described.
Staff can inspect the evidence and controls behind an agent's action and check an exception or proposed decision before approving it. The user terms supplement the institution's master agreement and disclaim product warranties and liability for specified categories of loss. An incorrect extraction or loan file check carries no accuracy warranty or compensation, and the bank keeps responsibility for the lending decision.
Agents combine access to the institution's software with models that read files and interpret policies, configured by Saris for each lender. Fiserv, Jack Henry, FIS, ICE and MeridianLink systems supply the records, and Zillow and public sources supply outside data. The developers and hosted model services behind document reasoning, message drafting and exception routing are unnamed.
Connections cover Fiserv, Jack Henry and FIS cores, ICE Encompass and MeridianLink for origination, and OnBase and Outlook for documents and email. Where a system has no direct interface, the System AI Agent works through the same screens staff use, with institution credentials and permissions. The External AI Agent checks data against Zillow, the Social Security Administration, UCC filings, county tax portals and insurers' systems.
Saris runs as a hosted platform that connects to an institution's systems, with encryption in transit and at rest and access limited by role. The lender keeps its core and origination applications while agents handle tasks across them. The hosting region and the location of agent evidence and input files are unstated.
Saris starts with a workflow analysis before a bank commits, identifying slow steps across origination, servicing, quality control or compliance and configuring agents for the institution's systems. Implementation runs 90 to 120 days without replacing the core or origination system, under a master service agreement. Subscription tiers, implementation fees and a billed unit per workflow or agent are quoted.
Community banks and credit unions are the core buyers, with a $5 billion regional bank among the clients. Consumer lending teams automate file review and data checks for home equity, auto and personal loans, mortgage teams review documents from origination through post close quality control, and commercial lenders compare entity and guarantor information and structure personal financial statements. Compliance staff use the same platform for regulatory reviews. Vibrant, AmFirst and Catalyst each run a different lending or compliance workflow.
Alternatives to Saris AI
Other products that do the same kind of work as Saris AI, and what each one does.
Infrrd is a San Jose intelligent document processing company founded in 2015 by chief executive Amit Jnagal, built on deep learning, computer vision and natural language processing rather than templates.
Proximitty runs autonomous agents across the commercial loan servicing lifecycle for banks, credit unions and fintechs, covering commercial and industrial, commercial real estate and small business administration lending.
UPTIQ sells AI agents that take on lending and banking operations work for financial institutions. Banks, credit unions, wealth managers, equipment finance firms and non bank lenders use them to open accounts, underwrite loans, monitor portfolios and run compliance tasks.
Informed.IQ verifies income and loan documents for consumer and auto lenders. Its software reads pay stubs, bank statements and lender forms, extracts relevant fields and compares the results with each institution’s lending policy.
Alogent sells deposit automation and check fraud prevention to community banks, credit unions and regional institutions.
Zafin AIOS is an agent orchestration platform for banks and other regulated institutions, sold separately from Zafin's Banking Platform and Zafin IO.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | Data Protection Terms | Implementation | Source |
|---|---|---|---|---|
|
Quoted after a workflow analysis.
|
Saris starts with a workflow analysis, then configures agents around a bank or credit union's existing lending, compliance and banking systems. The institution buys under a master service agreement. | The institution's master agreement governs customer data. Security and compliance documentation goes to institutions in active evaluation under a nondisclosure agreement. | Implementation typically takes 90 to 120 days without replacing the core or origination system. Implementation and integration charges are quoted. | Vendor Published |
Saris states no price per workflow, number of included agents or minimum contract term.