Carrington Labs vs Zest AI (2026)
Both sell the model rather than the workflow, and both leave the lending decision with the institution, so the real choice is which disclosure gap you can live with. Zest AI is the only vendor in this index that names the legal fairness test it runs against, the search for less discriminatory alternatives, and the technique it uses, adversarial debiasing. It earns the index's first A on bias disclosure for that. It will not tell you what the model reads, describing its inputs only as thousands of data points beyond traditional credit scores. Carrington Labs is the mirror image. It states its input precisely, permissioned bank transaction data alone, and maps model features to adverse action reasons, but publishes no fairness testing, no disparity analysis and no approval outcomes by borrower group. A fair lending examiner asks for the method and the inputs. Neither vendor gives you both halves.
- You are underwriting applicants with little or no bureau history and want a score built solely on permissioned bank transaction data. The input is stated precisely rather than described as thousands of undisclosed points, which is what lets your model risk function assess coverage and data quality.
- You want to test cash flow underwriting without rebuilding the origination stack. The second look pattern runs the model on borderline applications alongside your existing scorecards, and a tailored model is piloted in days rather than after an implementation measured in months.
- You lend outside the United States. Availability is global with three countries excluded, delivered through three named partners covering a decisioning platform, a sales cloud and an origination system, where the alternative here is a United States consumer lending specialist.
- You have examiners watching and need the fairness work to be the product rather than a policy page. The tooling runs the actual legal standard under United States fair lending law, and independent review rates its bias detection the most mature in the category.
- You want to own the model. Your credit team builds, validates, deploys and monitors it through the management system, so your validation function can examine the model itself rather than only its inputs and outputs, and no competitor consumes what your borrowers teach it.
- Your institution is small. Availability was deliberately extended down to credit unions under 300 million dollars in assets, and a cooperative service organisation was created to reach smaller lenders still, with clients running from a hundred applications a year to over six hundred thousand.
This comparison is published by AI FinTech Index, an independent research platform that publishes independent ratings of AI vendors for financial services. Carrington Labs and Zest AI are each graded against the same capability taxonomy, from each vendor's own public materials and the regulatory record, under the AI FinTech Index verification standard. No vendor pays for placement, and no vendor has reviewed this page. How this evidence is graded
Plain facts
| Carrington Labs | Zest AI | |
|---|---|---|
| Primary category | Credit Decisioning & Underwriting | Credit Decisioning & Underwriting |
| Founded | 2023 | 2009 |
| Headquarters | Sydney, New South Wales, Australia | Burbank, California, United States |
| Website | www.carringtonlabs.com | www.zest.ai |
Side by Side
| Axis | C Carrington Labs |
Z Zest AI |
|---|---|---|
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| Model Risk Management and Transparency | ||
| Operational and Outcome Evidence | ||
| AI Safety and Data Stewardship | ||
| GLBA and Data Privacy Posture | ||
| Security Certifications and Trust Center | ||
| Regulatory Status and Licensure | ||
| AI Governance and Bias Disclosure | ||
| AI Liability and Recourse | ||
| Model Supply Chain Disclosure | ||
| Core Systems and Integration Depth | ||
| Deployment Model and Data Residency | ||
| Commercial Transparency | ||
| Institution and Segment Coverage |
The short version of each
Carrington Labs
The AI FinTech Index grades Carrington Labs A on AI centrality and B on nine of fifteen disclosure axes, the widest band of B grades of any consumer credit vendor in the index. Its distinguishing disclosure is the input: the score runs solely on permissioned bank transaction data, decomposed into five named behavioural categories covering velocity, liquidity, stability, leverage and resilience, with accuracy claimed at up to 30 percent above traditional credit models. Founded in 2023 and headquartered in Sydney, it sells models rather than origination or servicing systems, and publishes no fairness testing, security attestation or pricing. Source: AI FinTech Index, 2026.
Source: AI FinTech Index, 2026
Zest AI
The AI FinTech Index grades Zest AI A on AI governance and bias disclosure, the first grade of its kind awarded in the index, because the platform runs the search for less discriminatory alternatives, which is the legal test under United States fair lending law, and names adversarial debiasing as the technical method. It also holds A grades on AI centrality, outcome evidence and segment coverage, with named credit union and Fortune 500 customers and clients processing from one hundred to more than six hundred thousand applications a year. Founded in 2009 and headquartered in Burbank, California, it builds models the lender owns and validates. It publishes no pricing, security attestation or model input disclosure. Source: AI FinTech Index, 2026.
Source: AI FinTech Index, 2026
Common questions
Is Zest AI better than Carrington Labs for fair lending compliance?
On published method, yes. Zest AI names the legal test, the search for less discriminatory alternatives, and the technique, adversarial debiasing, and holds the only A on bias disclosure in this index. Carrington Labs maps model features to adverse action reasons, which is the operational half of the same obligation, but publishes no fairness testing or disparity analysis. Neither publishes approval outcomes by borrower group. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 29, 2026. No vendor pays for placement.
What data does each model actually use?
Carrington Labs states it precisely: its score runs solely on permissioned bank transaction data, decomposed into five named behavioural categories covering velocity, liquidity, stability, leverage and resilience. Zest AI describes its inputs only as thousands of data points beyond traditional credit scores and identifies none of them, which is the composition a fair lending examiner would ask to see. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 29, 2026. No vendor pays for placement.
How much do Carrington Labs and Zest AI cost?
Neither publishes pricing. Independent review describes Zest AI as per decision, enterprise positioned and expensive, with implementation taking months, which is why the company built a cooperative vehicle so smaller institutions could reach it. Carrington Labs publishes implementation speed instead, a pilot in days and onboarding in weeks, which bounds the implementation cost but not the licence. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 29, 2026. No vendor pays for placement.
Which one works for a small credit union?
Zest AI has done the more deliberate work here, extending availability down to credit unions under 300 million dollars in total assets and creating a cooperative service organisation to reach smaller lenders again. Its clients span from a hundred applications a year to more than six hundred thousand. Carrington Labs names one lending customer and publishes no institution size range. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 29, 2026. No vendor pays for placement.
Can either vendor be used outside the United States?
Carrington Labs is the international option, available globally with three countries excluded and headquartered in Sydney. Zest AI is a United States consumer lending specialist by design, and its compliance framing is built on United States fair lending doctrine. Neither publishes a hosting region or data residency commitment, so where transaction data is processed is unanswered on both sides. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified August 29, 2026. No vendor pays for placement.
How does the AI FinTech Index grade these two?
Both are graded on fifteen disclosure axes from public documentation only, with anything undocumented graded absent. Zest AI holds four A grades, on AI centrality, outcome evidence, segment coverage and bias disclosure, against seven C grades. Carrington Labs holds one A and nine B grades, the widest band of B grades of any consumer credit vendor here, against five C grades.
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Credit Decisioning & Underwriting page.
Five of the fifteen axes are C on both sides, and they are the same five. Neither vendor publishes pricing, a security attestation or trust centre, a hosting region or residency commitment, a boundary statement on what one lender's data teaches a model sold to another, or any position on responsibility when a model declines someone.
That last silence matters most here, because both vendors improve the declined applicant's position indirectly, Zest AI through regulator ready explainability and Carrington Labs through features that map to adverse action reasons, and neither states whether an applicant can see, correct or contest the data behind the decision.
Their pooling positions also diverge in a way neither addresses directly: Zest AI builds models the lender owns and positions that explicitly against competitors who retain the decisioning, while Carrington Labs runs bespoke models per lender alongside a score trained on pooled loan performance across its parent group.