Overview
AI bank statement analysis is a feature within Legl's Source of Funds that automatically extracts and summarises key financial information from a client's bank statements. It is designed to reduce the time your firm spends manually reviewing lengthy bank statements, and surfaces the information most relevant to your AML review, without you needing to read every transaction.
What you can use AI bank statement analysis for
Quickly identify a client's regular income sources without reading every transaction
Spot large or unusual one-off payments that may warrant further investigation
Identify higher-risk transaction indicators, such as cash, non-GBP, or cryptocurrency activity
Review account balance trends across the statement period
Reduce manual review time for submissions that include multi-page PDF bank statements
How it works
When a client provides bank statement evidence, either via open banking or by uploading a PDF, Legl automatically runs AI analysis on the data. Before the results are included in the review, they are checked by Legl's mathematical validation algorithm, which confirms that the extracted figures are internally consistent. Only analysis that passes this check is included in the Source of Funds review.
This two-step process (extraction followed by validation) means the analysis your firm sees has been verified for accuracy before it reaches you.
What is extracted and surfaced
The following information is automatically extracted from the client's bank statements and surfaced in the Source of Funds review:
Regular incoming transactions: recurring salary payments, standing orders, and other income streams
Regular outgoing transactions: recurring outgoings across accounts
Largest one-off transactions: significant individual credits or debits
Account balance trends: how balances have changed across accounts over the statement period
Higher-risk transactions: activity associated with elevated risk, such as non-GBP deposits, cash transactions, or cryptocurrency activity
Open banking vs manual PDF upload
AI analysis applies to both open banking and manual PDF bank statement submissions, but the experience differs:
Open banking: data is provided directly from the client's bank in real time. Analysis is highly reliable and also powers SoF Signals, which automatically sends clients follow-up questions for flagged high-risk transactions.
Manual PDF upload: Legl's AI analyses the uploaded document. Where analysis cannot be completed, the review will indicate that manual review is required. SoF Signals does not apply to manual PDF submissions.
ℹ️ Important
Clients should upload a complete, multi-page PDF bank statement showing transactions across the relevant period. Documents that are not bank statements, such as payslips, utility bills, or mortgage offer letters, cannot be analysed by the AI and will result in a failed analysis.
Where to find the analysis
The analysis is surfaced automatically within the Source of Funds review in Legl, once the client has submitted.
It appears as a summary section within the review, alongside the client's questionnaire responses and bank statement evidence.
The full PDF report, which includes the same analysis, is also available for download from the client's Engage request.
Important information
Where AI analysis of a bank statement is not successful or accuracy cannot be verified, the analysis will not be included and the review will prompt manual review.
Only bank statements should be uploaded for AI analysis. Other document types, such as payslips, utility bills, or mortgage offer letters, cannot be analysed and will result in a failed extraction.
SoF Signals (automated follow-ups for flagged transactions) applies to open banking data only, not to manual PDF uploads.
The mathematical validation algorithm is a quality control step, not a compliance check. It confirms the accuracy of extracted data. It does not assess whether the client's source of funds is acceptable.
Legl does not make compliance decisions. All assessment of the evidence remains your firm's responsibility.