COREP reporting where every cell reads as a sentence, drills to its source, and changes only when a person approves
Every COREP and FINREP cell traced to its source, and each new release mapped in days.
See one case, screen by screen ↓COREP & FINREP Reporting is COREP reporting and FINREP reporting with agents for UK and EU bank reporting teams. Seven agents read each reporting release, populate the templates from the ledger and risk systems, run the published validation rules, reconcile COREP with FINREP and explain the Basel 3.1 parallel run, with every cell written as a sentence and traced to its exposures, rules, mapping and source records. Analysts accept, managers approve, the Head releases and the CFO attests.
AI use cases in banking: a buyer's guide to agentic AI for bank and credit union operations →
Updated
COREP reporting at a UK bank means 50 COREP and FINREP templates and about 12,600 populated cells, checked against 1,140 published validation rules every quarter, by a team of seven. Each reporting release brings a change log of hundreds of items that a senior analyst reads by hand to find the few that touch the bank's own templates. Mappings from ledger accounts and product codes to template rows are edited in spreadsheets and approved by email, and the templates are reviewed cell by cell in grids.
The rules check arithmetic, not meaning: a mortgage in the wrong exposure class passes every one of them and misstates capital. In 2026 every quarter runs twice, because Basel 3.1 takes effect on 1 January 2027 and the parallel run needs its own templates and its own explanation for the CFO. And the CFO, as the Senior Manager accountable for the return, attests it with evidence assembled from emails and screenshots, where overrides have lost their reasons.
Estimated split for one quarterly return and the reporting release that lands before it, by hand and with the agents. The release row follows the 2–4 weeks the team tracks today and the 4-business-day target; the other rows are our estimate, kept modest.
Seven agents read, populate, validate, reconcile and explain; the analyst maps, the manager approves, the Head releases and the CFO attests.
Corvane Bank UK's Q3 2026 return, reference date 30 September 2026, is due on 11 November 2026, and the team is running Basel 3.1 in parallel. Here is Thursday 22 October, screen by screen, in the working solution.
Tom opens the return with 14 business days left. The left list holds 8 of 50 templates, grouped by why: C 07.00 has 3 rules failing, F 18.00 has an override waiting, C 02.00, C 16.00 and C 47.00 changed since 30 June, C 01.00 and F 01.03 carry a £5.7m cross-template difference, and C 33.00 waits for a feed 15 hours late. In the center, one cell reads as a sentence, with 30 June, today's rules and the Basel 3.1 value beside it.
“C 07.00, row 0010, column 0220: risk-weighted exposure amount £3,412.6m, built from 41,820 exposures.”
The first drill step splits the 41,820 exposures by loan-to-value band and ties the £9,751.0m exposure value to the ledger. Tom searches for buy-to-let and opens MTG-UK-0052306, a mortgage in Bristol with a balance of £241,750, valued at £390,000 at origination: risk weight 35% today, 26.2% under Basel 3.1 once a mapping is approved. The record opens at its source in the credit risk data mart, with the customer's name masked.
Opening a source is recorded in the audit trail.
Mapping rule MR-1104 v12 places 36,206 owner-occupied mortgages in loan-splitting. MR-1105, buy-to-let to individuals with three or fewer properties, covers 5,614 exposures and has no Basel 3.1 mapping, so they fall out of the parallel template. The change waiting for approval, MAP-0417, is shown before and after, with its citations.
“MAP-0417 · RM-BTL-S → regulatory residential real estate, not materially dependent on cash flows, loan-splitting”
The Validation agent ran 1,140 published rules on Draft 3. The 642 rules for today's framework all pass; 3 of the 498 Basel 3.1 rules fail, all traced to C 07.00. Each is written as a sentence with its formula and the amount it fails by: VR-C07-118 and VR-C07-131 by £1,302.7m, VR-C02-044 by £359.5m. The run history shows the quarter's progress: 41 failing on Draft 1, 9 on Draft 2, 3 now.
“Cause and fix: product family RM-BTL-S has no Basel 3.1 mapping. Approving MAP-0417 fixes all three.”
The Release agent read the Bank of England banking reporting framework v4.1.1 of 27 August 2026: 312 changes, 46 touching Corvane Bank UK's templates, 16 mapping proposals, 14 already approved in a median of 4 business days. It proposed MAP-0417 from PRA PS9/24 chapter 3 for 5,614 exposures worth £1,302.7m, at confidence 0.94, and Priya Nair accepted it as the analyst on 21 October. A load that failed its checksum on 28 August is shown too, with the retry.
The Release agent never changes a mapping and never proposes outside the templates the bank reports.
Before he decides, the Preparation agent previews the effect: for the 5,614 buy-to-let exposures, risk-weighted exposure falls from £455.9m to £359.5m, −£96.4m in the Basel 3.1 template. Tom approves MAP-0417 with the reason code “PS9/24 real estate treatment” and a note for the record. Priya Nair accepted the proposal as the analyst; a manager who did not make it approves it.
“Recorded with your name, the time and the evidence you saw. You cannot approve your own work.”
The agents work step by step in view. The Preparation agent maps the 5,614 exposures and repopulates C 07.00 and C 02.00, 84 cells with new lineage. The Validation agent runs all 1,140 rules on Draft 4 and every one passes. The Cross-template agent re-checks C 02.00 against C 07.00: it agrees. The Basel 3.1 cell is now £3,190.0m, −£222.6m against today's rules.
“Draft 4: 1,140 of 1,140 rules pass.”
Total risk exposure amount is £5,948.2m under today's rules and £5,886.7m under Basel 3.1, −£61.5m, so the CET1 ratio moves from 14.17% to 14.31% on CET1 of £842.6m. Five drivers explain every pound: residential real estate −£222.6m, operational risk under the business indicator method +£73.6m, cancellable limits at a 10% conversion factor +£58.3m, the SME support factor removed +£41.9m and investment-grade unrated corporates at 65% −£12.7m. Each driver opens its cells and its source.
Parallel-run results stay inside the bank; the agents never send them to the regulator.
The Cross-template agent checks 86 relationships between COREP and FINREP. Retained earnings are £612.4m in C 01.00 and £618.1m in F 01.03: the £5.7m is Q3 profit the external auditor has not yet verified, which UK CRR Article 26(2) keeps out of CET1. The agent writes the explanation with the article; a manager accepts it or opens both cells.
The Cross-template agent never accepts its own explanation.
Eleanor Marsh, Head of Regulatory Reporting UK, releases the return once every check is green: 44 of 44 quarterly templates approved, 1,140 of 1,140 rules pass, 86 relationships agree or are accepted. David Okafor, the CFO, attests as SMF2 with the Evidence agent's trail on one page: 12,614 cells with lineage, 100% traced to a source record. Because he attested, Eleanor submits, and the portal returns receipt RD-2026-11-10-48213 with 0 errors.
The attester never submits: the solution enforces it.
The same quarter, seen by the five people who carry it: what their quarter looked like, and what it looks like now.
Seven agents read, populate, validate, reconcile and explain; the analyst maps, the manager approves, the Head releases and the CFO attests.
Reads each Bank of England reporting framework release, PRA policy statement and instruction change, keeps only what touches the templates the bank reports, and proposes mapping changes with the cited paragraph and the number of exposures affected.
Populates COREP and FINREP from the general ledger, credit risk data mart, core banking system, treasury system and data warehouse, writes each cell as a sentence with its lineage, and previews what a pending mapping does before anyone approves it.
Runs the 1,140 published validation rules on every draft, for today's framework and the Basel 3.1 parallel run, and explains each failure in plain words with its cause and a proposed fix.
Checks 86 relationships between COREP and FINREP, from own funds against equity to exposures in default, and explains a legitimate difference with the rule behind it.
Attributes every difference between today's rules and Basel 3.1 to a named driver, with its cells and the paragraph behind it, and computes the effect on the CET1 ratio.
Tracks the remittance date, internal milestones, feed arrivals and SLAs, and escalates a late feed, a slipped milestone or an open override to the Head of Regulatory Reporting UK.
Builds the CFO's attestation evidence and the exports for the PRA supervisor, internal audit and the external auditor from the audit trail, the return snapshot and each cell's lineage.
Maps · approves · releases · attests. The agents propose; a named person decides.
Ask about a cell, a template, the parallel run or what still blocks the return. Every answer cites its source, and opens the screen it talks about.
Why did C 07.00 change from last quarter?
Risk-weighted exposure for mortgages rose £56.5m, from £3,356.1m to £3,412.6m, on net new lending: £161.4m of new exposure value at 35%, and the ledger ties to £9,751.0m. In the Basel 3.1 parallel run, 5,614 buy-to-let exposures still need mapping MAP-0417, which is why 3 rules fail.
Which Basel 3.1 drivers move our CET1 ratio?
CET1 goes from 14.17% to 14.31% because total risk exposure falls £61.5m: real estate −£222.6m, operational risk +£73.6m, cancellable limits +£58.3m, the SME support factor +£41.9m and investment-grade corporates −£12.7m.
Can the £5.7m go into CET1?
Only with prior permission once the external auditor has verified the Q3 profit, under UK CRR Article 26(2). The ledger shows £5.7m not yet verified.
What still blocks the return from release?
Four things: MAP-0417 and MAP-0421 wait for a manager, OVR-0031 waits for a checker, XT-07 waits for acceptance, and the treasury feed is 15 hours late.
13 screens from the working solution, on its sample data. Pick one to see it large.
Every change is saved as a new version and goes live when a second administrator approves it.
| Setting | Default | Choose from |
|---|---|---|
| Reporting frameworks | Bank of England banking 3.7.0 live; 4.1.1 as the Basel 3.1 parallel run | Each framework listed with its use, rules and status; the EBA's framework 4.0 and DPM 2.0 are listed, off for a bank that reports to the PRA only |
| Release watch | On, nightly at 02:00 | On · Off |
| Second analyst below confidence | 0.90 | A confidence threshold the administrator sets |
| Quarantine a release that fails its checksum | On | On · Off |
| Escalate a late feed after | 4 hours | Set by the administrator, alongside a slipped milestone (1 business day) and an open override (2 business days) |
| Approve reason codes | PS9/24 real estate treatment, and five more | Each code switched on or off, for approvals, returns and overrides |
| Mask customer names in exposure views | On; reveal needs a reason and is logged | On · Off |
| Retention of returns, lineage and evidence | 7 years | Set by the administrator; customer records at least 5 years |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample return · “target” = the design goal, measured in the live solution · “estimated” = our estimate. Regulatory context: PRA PS9/24, Implementation of the Basel 3.1 standards near-final part 2 (12 September 2024), chapter 3 · PRA PS1/26, final Basel 3.1 rules (20 January 2026), effective 1 January 2027 · Bank of England banking reporting framework v4.1.1 (27 August 2026) · UK CRR Article 26(2) · the Senior Managers Regime (SMF2). Corvane Bank UK, the people named on this page, its exposures and every record number are fictional — characters and sample data in the working solution.
COREP & FINREP Reporting is COREP reporting and FINREP reporting with agents for UK and EU bank reporting teams. Seven agents read each reporting release, populate the templates from the ledger and risk systems, run the published validation rules, reconcile COREP with FINREP and explain the Basel 3.1 parallel run, with every cell written as a sentence and traced to its exposures, rules, mapping and source records. Analysts accept, managers approve, the Head releases and the CFO attests.
COREP is the common reporting of capital and other prudential data, such as own funds, credit risk and capital ratios; FINREP is financial reporting, such as the balance sheet, equity and non-performing exposures. The two must tell the same story, so the Cross-template agent checks the relationships between them, 86 in the sample return, and explains any legitimate difference with the rule behind it.
The sample bank sends 44 quarterly and 6 monthly templates. In the UK, quarterly returns are due on 12 May, 11 August, 11 November and 11 February. The Deadline agent tracks the remittance date and internal milestones, and escalates a late feed after 4 hours or a slipped milestone after 1 business day; it never moves a regulatory date.
The Release agent reads each Bank of England reporting framework release and PRA policy statement and proposes mapping changes only for the templates the bank reports, each with the cited paragraph and the exposures affected. In the sample, release v4.1.1 brought 312 changes, 46 touched the bank's templates and 16 became proposals. An analyst accepts or rejects each one, and a manager approves it after seeing its impact.
Published validation rules check arithmetic, not meaning: a mortgage in the wrong exposure class passes every one. The Validation agent runs all the rules on every draft and explains each failure in plain words, and because every cell drills to its exposures and mapping, a reviewer can check what a number means, not only that it adds up.
Yes. The Parallel-run agent attributes every difference between today's rules and Basel 3.1 to a named driver with its cells and paragraph, and computes the effect on the CET1 ratio. Parallel results stay inside the bank; the agents never send them to the regulator.
Yes. The agents propose and never decide. Every mapping change, cell override, cross-template explanation and template approval needs a second person and a reason code, nobody can approve their own work, and the CFO who attests the return can never be the one who submits it.
The Agentic Solution Engine builds and deploys it from your requirements and documents: your templates, mapping rules, source extracts and a past return. It goes live once every quality gate has passed, and we will walk you through it on your own return first.
Each board figure traced to source, reconciled at every hop, every rating cited.
1-min demoCall Report Instructions SearchA cited answer in under a minute, with the bank’s own precedent beside the instruction.
1-min demoCall Report PreparationHalf the close back, and every figure opens to its ledger and its instruction paragraph.
1-min demoWe'll run COREP & FINREP Reporting on a sample of your own templates, mappings and last quarter's return.