Automation has value when the complete service improves
Zest AI describes client-tailored machine-learning underwriting and automated decisions. Its current product page makes quantitative performance claims, while the cases below describe particular customer experiences. Treat those claims as evidence to examine, not as an industry-wide forecast or the performance of a lender that has not yet deployed the product. [1]
Analysis: the borrower experiences an application, a decision, any document follow-up and delivery of funds. The lender experiences data collection, analysis, notices and exceptions. Automation of one step can create a faster decision while leaving the rest of the journey unchanged.
This distinction matters for staffing and growth. An automatic decline is an automated decision but does not create a funded relationship. An automatic approval may still require documents and customer acceptance. The relevant outcome depends on the problem being solved: capacity, turnaround, sustainable access or contribution after the full cost of service.
What the tool is intended to do
Zest AI markets underwriting technology that uses machine learning to support credit decisions for lenders. Its product materials emphasize predictive performance, automation, and fair-lending analysis. Those are vendor descriptions of capabilities; they are not independent proof that a particular institution will achieve better outcomes.
The relevant unit of evaluation is the complete lending process: applicant population, permitted data, model, policy rules, pricing, explanations, operational exceptions and subsequent performance. A more predictive score can be valuable, but a model result becomes a customer outcome only after these other parts of the process are applied.
What one published case supports
A Zest-published Idaho Central Credit Union case study reports approval increases of more than 30% across specified consumer products and an increase in automated decisions from 50% to 75%. These are claims from a vendor-hosted customer story. They provide a reason to investigate the product, not an independently controlled estimate of its causal impact.
Analysis: before transferring those results to another lender, ask about the baseline policy, approval definition, product mix, observation period and credit performance window. An improvement measured against a restrictive old policy may differ from improvement against a strong existing model. Approval growth also needs to be interpreted alongside pricing, take-up and realized losses.
Bank adoption adds context beyond a vendor feature list
Zest’s August 5, 2026 update says Zions Bank and Hawaii State Federal Credit Union selected its technology during the second quarter and identifies an integration partnership with Launcher Solutions. These are vendor-reported selections and distribution activity; the release does not establish the product-level production share, deployment date or realized credit outcomes for those lenders. Zions Bank is a division of Zions Bancorporation, N.A., rather than a separately incorporated bank subsidiary. [5][8]
The First Hawaiian Bank customer case describes a credit-card underwriting implementation and reports that automated decisions, including approvals and declines, reached 55% within the first year. It provides a concrete product and workflow example. It should not be generalized to every First Hawaiian loan product or treated as an independent estimate of the technology’s causal effect. [4]
Why the public First Hawaiian figures need careful handling
Zest’s customer case gives a 4% starting point for automated decisions. A separate Zest-hosted page introducing a Celent report describes instant decisioning rising from 10% to 55% over one year. The public descriptions use different terminology and starting figures. The full gated report and supporting methodology were not inspected, so comparability cannot be established. This article does not calculate a common improvement multiple or merge the two baselines. [4][6]
For a buyer, this is a useful diligence example: request the numerator, denominator, eligible population and measurement period before using an automation statistic in a business case. Separate automatic approvals from automatic declines and referrals. A larger share of automated decisions can reduce work, but it does not by itself prove more sustainable lending or lower loss rates.
Management commentary and the adoption question
In the August 5 announcement, Zest CEO Mike de Vere said, “AI is becoming the core infrastructure for modern lending.” This is the vendor’s strategic interpretation, not a measured claim that the industry has universally adopted its products. [5]
Our interpretation is narrower: the named selections and implementation case show several routes to deployment, from a customized underwriting model to an origination-platform integration. They also increase the importance of separating underwriting, fraud detection and generative lending intelligence. A relationship covering one module is not proof that the customer adopted all three.
Older case-study language refers to SR 11-7. For current bank governance, the April 17, 2026 interagency revised model-risk guidance supersedes SR 11-7 and SR 21-8; the Federal Reserve letter describes its expected relevance to Fed-regulated organizations above $30 billion and a tailored, risk-based approach. A historical vendor statement about documentation is neither current regulatory approval nor a blanket requirement for every credit union. [7]
Design a pilot that can answer a decision
Start with a specific question: can the proposed system approve additional qualified borrowers at an acceptable risk level, or reduce manual work without worsening outcomes? Preserve the existing model and policy as a benchmark. Use a holdout period that was not used for training, and evaluate performance across economically relevant segments.
Analysis: historical data contain outcomes mainly for applicants who received credit. Results for previously declined populations are therefore uncertain. A controlled expansion can supply new evidence, but should be sized and monitored deliberately. A favorable backtest cannot by itself establish how newly approved customers will perform under a changed policy.
Illustrative economics of automation
Assume a fictional lender receives 100,000 applications annually. Increasing automated decisions from 50% to 75% would reduce manual reviews by 25,000 if every other process remained unchanged. At an assumed $8 of avoidable cost per review, the gross annual saving would be $200,000. These assumptions are illustrative and are not Zest pricing or a forecast for the cited customer.
The net benefit must subtract software, integration, validation, monitoring, exception handling and any additional credit or fraud losses. Some labor cost may remain fixed even when reviews decline. Track actual hours and rework rather than multiplying every automated application by a fully loaded cost that cannot be removed.
A lender’s acceptance scorecard
This proposed scorecard separates model quality from operational readiness.
Scroll horizontally to see all columns.
| Dimension | Evidence to request | Failure mode |
|---|---|---|
| Credit performance | Comparable holdout and seasoned cohort results | Approval lift without comparable risk |
| Fair lending | Outcome testing and documented alternatives | Portfolio averages hide segment harm |
| Reasons | Decision-level validation | Readable reasons that do not explain the decision |
| Operations | Latency, availability and fallback records | A model outage stops applications |
| Economics | Actual costs and retained contribution | Gross benefits presented as net savings |
Explanation and change control
Regulation B’s specific-reasons requirements apply to the credit decision. Analysis: validate how model factors, hard policy rules and human overrides reach the final notice. Keep the deployed model version and input record so a decision can be reconstructed after an update. Vendor explanation tooling can support this work; responsibility for the lender’s process does not disappear.
A model upgrade should have an acceptance standard before deployment. Compare the challenger with the existing model on the same data, inspect segment changes, verify reason mappings and define rollback conditions. Monitor drift in applicant mix and data availability, since those can change results even when the model code remains unchanged.
Follow approvals through to funded relationships
Hypothetical: one process approves 6,000 of 10,000 applications, and 60% of approved applicants complete funding, producing 3,600 loans. A revised process approves 6,600 but has 50% take-up, producing 3,300 loans. Approvals rise 10% while completed funding falls about 8.3%. These are fictional outcomes, not a Zest customer comparison.
Pricing, documentation, competitor offers and customer intent could explain the difference. The model should receive credit only for the change it actually causes. Likewise, employees may use released review time to handle more complicated cases; that is additional capacity, not necessarily an expense reduction.
Evaluate the full process using comparable applicants, completed funding, time and cost to serve, and outcomes after loans have seasoned. The governance work described above is important because decisions affect customers, but the adoption question also includes whether the service becomes more useful and the economics remain sustainable.
Open questions and revision triggers
No universal public price is assumed here. Obtain a quote and clarify data rights, support, audit access, model updates and exit arrangements. The strongest evidence would be lender-specific, reproducible and seasoned results with transparent definitions. Evidence limited to headline approval gains leaves the credit and economic case incomplete.
Revisit this profile when product documentation changes, a customer publishes independently interpretable outcomes or a material legal development changes decision requirements. Preserve the distinction between vendor claims, customer-reported experience and this article’s proposed evaluation method.
Sources
- Zest AI — Underwriting product overviewSourceBack to text: ↑
- Zest AI — Idaho Central Credit Union case studySource
- CFPB — Regulation B §1002.9, specific adverse-action reasonsOfficial text
- Zest AI: First Hawaiian Bank credit-card implementation case; undated, reviewed September 29, 2026; vendor/customer accountSourceBack to text: ↑1↑2
- Zest AI: first-half 2026 business update; August 5, 2026; vendor-reported selectionsSourceBack to text: ↑1↑2
- Zest-hosted landing page for Celent retail-lending report; undated, reviewed September 29, 2026; full gated report not inspectedSourceBack to text: ↑
- Federal Reserve SR 26-2: Revised Guidance on Model Risk Management; April 17, 2026Official sourceBack to text: ↑
- Zions Bank: official careers-site legal footer identifying it as a division of Zions Bancorporation, N.A.; reviewed September 29, 2026SourceBack to text: ↑