HOW IT WORKS

From ERP data to a closed case

Four steps: Understand, Plan, Execute, Explain. AI performs the analysis; your experts keep every judgement.

01

Understand

foretale.ai reads your enterprise data, relationships, policies, manuals, and business context, and builds the data model itself. This is the step that normally consumes most of a risk analytics project, and it is why engagements usually start in weeks rather than months.

02

Plan

The platform plans which analytics to run for the risk objective you set, based on the processes, entities, and data actually present in your environment—rather than applying a fixed test list regardless of context.

03

Execute

Analytics run across the full transaction population, not a sample. Hundreds of tests can execute across millions of records, including fuzzy matching, relationship analysis, and behavioural comparison against peers and history.

04

Explain

Findings are returned with the underlying transactions, the logic applied, and the data lineage, ranked by risk and materiality. From there they move into investigation, where your team validates, escalates, tracks, and closes them.

WHAT YOU PROVIDE

Data and access

Transaction extracts from the processes in scope—invoices, orders, receipts, payments, claims, journals, or payroll registers
Master data for vendors, customers, or employees, including bank and tax attributes where duplicate and conflict testing is required
Policy context—approval thresholds, discount bands, and procedure documents, so analytics test your rules rather than generic ones
Reviewer time to validate the first findings, which is what tunes precision for subsequent runs

WHAT STAYS WITH YOU

Judgement and decisions

Every risk conclusion—AI explains, people decide
Escalation and reporting decisions, including what reaches the audit committee
Authorisation for any high-impact action; these require explicit human approval
Challenge and dismissal of findings, with the reasoning retained as feedback

Frequently asked questions

How long does it take to get the first findings?

Typically weeks rather than months, because the data modelling step that normally dominates a risk analytics project is automated. The main variable is how quickly data extracts can be provided and access approved.

Does the AI make the final risk judgement?

No. AI performs the analysis and explains it; people decide what it means. Findings are presented with evidence and reasoning so a reviewer can confirm, challenge, or dismiss them, and high-impact actions require human authorisation.

What happens to false positives?

Findings are ranked by risk and materiality rather than reported flat, and data integrity issues are profiled first because they generate most false positives. Reviewer feedback then tunes subsequent runs.

Can analytics run on a schedule?

Yes. Analytics can run continuously or on a defined cadence—per pay run for payroll, before payment runs for accounts payable, around period end for journal entries—so exceptions surface while they are still correctable.

NEXT STEP

See it on your data.

Request a focused walkthrough—usually within one business day.