AI and Data Advisory
Your data, your models and your use of AI, put in order.
For lenders who know their data could do more and that their use of AI will be asked about. People who build lending software, working with you on your data, your credit models and your governance, each engagement scoped to your book and quoted before it starts.
AI and data readiness review
Know what your data can do, and where your decisions are exposed.
We read what you hold — application answers, bank statements, bureau files, the ledger, collections notes — and how your decisions are actually made: the policy rules, any scorecards or vendor models, the overrides your assessors make and why. Then we set it against what you will be asked about: the APP 1 disclosure of automated decisions from 10 December 2026, ASIC’s expectations of AI governance in REP 798, and your responsible lending file.
What you get
- A map of your data: what is there, how good it is, and what it could answer that nothing reads today
- Every automated or model-assisted decision you make, found and listed
- The gaps, ranked by the harm and the regulatory exposure each carries
- A plan in order: what to fix first, what to build, and what to leave alone
AI governance and automated decisions
Use AI and automation where it helps, and be able to show you govern it.
A governance framework sized for a small or medium lender rather than a bank: who may approve a model or an automated rule, how it is tested before it goes live and monitored after, when a person must review its output, and how a borrower hears that a program helped decide. We write it with you, in your words, around the systems you actually run.
What you get
- A register of every model, scorecard, vendor algorithm and automated rule, with its owner and its risk
- An AI and automated decision policy your board can approve
- The automated-decision wording APP 1 asks of your privacy policy, and where it must appear
- A board briefing on the risks, the obligations and the controls
Credit models, built or validated
Models fitted to your own book, every factor explained.
A scorecard, a set of policy rules or an affordability model fitted to the outcomes on your own loans, built the way the software is: every factor named with the evidence behind it, nothing that cannot be explained to a borrower or a regulator, and no protected attribute read. Each model is tested for stability over time and segment, for how it treats hardship apart from wilful default, and for whether it errs against the borrower. Already running a model, your own or a vendor’s? We validate it independently, as a credit risk function should.
What you get
- A model you own and can run in Creditcrest Decisioning or anywhere else
- Its documentation: data, factors, performance, limits and the tests it passed
- A monitoring pack that says when it has drifted
- Or, for a model you have: an independent validation report with findings in order of importance
Data and reporting
Your data where your people, your board and your funder can use it.
Applications, decisions, the ledger, collections and complaints joined in one place you control, so a question about your book is answered in minutes rather than a week of spreadsheets. We build the reports a lender is asked for most: the loan tape and borrowing base a warehouse funder wants, arrears and roll rates by vintage, approval and loss by channel, and the board pack.
What you get
- A warehouse of your lending data, documented, in your own account
- A funder-ready loan tape and borrowing base, reconciled to the ledger
- Portfolio, collections and conduct reporting for management and the board
- Your people shown how to change and extend it themselves
Fractional head of data and AI
A senior data and AI lead, a few days a month.
For a lender that needs the judgement of a head of data and models but not a full-time hire: someone who sits in your credit and risk meetings, owns model and data risk, reviews the vendors and models you are offered, decides with you where AI is worth using and where it is not, and answers your board’s and your funder’s questions about it.
What you get
- Model and data risk owned and reported each month
- Vendor and model reviews before you sign
- A standing view of where AI could cut cost or loss in your business, and what it would take
- Someone accountable for the answers when a regulator, auditor or funder asks
