Atlas Verum is a specialist firm for regulated finance. We advise on reporting, controls and risk — and automate the work at two levels: the process, where it is done, and the governance, where it is evidenced, reviewed and approved. Senior-led, machine-assisted, to evidential standard.
The FTSE 100 closed the week 0.6% higher at 10,816.56 on 21 August, gold above $4,600 an ounce lifting the miners: Antofagasta up 5.4%, Endeavour Mining 4.1%, Anglo American 2.6%. The FTSE 250 added 0.9% to 24,718.82, helped by August flash PMIs coming in ahead of expectations. Weak July retail sales kept the domestic read mixed.
PS26/17 landed on 13 August: final rules on fund liquidity risk management for authorised managers of UK UCITS and non-UCITS retail schemes, in force 1 February 2027 with transitional provisions to 1 August 2027. Anti-dilution tools must be used effectively, the eligible-market test tightens, and stress testing carries new guidance. T+1 remains the other fixed point at 11 October 2027, with buy-side readiness still the weak link.
Basel 3.1 lands 1 January 2027 with the detail still moving — market-risk IMA adjustments consulted in June, ring-fencing continuity rules in July — while liquidity and reporting expectations tighten for mid-tier firms.
Provision 29 declarations land in 2026 annual reports; IFRS 18 comparatives must be running now for 1 Jan 2027 adoption.
August flash PMIs put composite output at 52.5, a four-month high, with services at 52.8 and manufacturing easing to 51.5. July retail sales volumes fell 0.5% on the month and were up only 1.6% on the year, against 2.2% expected. CPI stands at 2.9% on the 19 August ONS print, up from 2.6% on the Ofgem cap rise, with core steady at 2.6%. Bank Rate stays 3.75% and the next decision is 17 September. Plan funding, ECL and covenant headroom for slow easing, not a pivot.
The Budget is fixed for Wednesday 28 October, with an OBR economic and fiscal forecast published alongside it. That sets the outer edge of the planning year: covenant tests, deferred tax positions and capital plans priced before late October carry re-forecast risk. Underneath it the growth agenda keeps writing the regulatory calendar, from the AI adoption plan for financial services to the consultation on modernising payments regulation.
The Mills Review maps how AI reshapes retail finance to 2030 and floats an agentic supervisory model — no new rules yet, but a visibly higher evidence bar for machine-assisted work.
We believe regulated finance work should stand up to inspection — files that visibly contain their reasoning, controls that certify cleanly, and technology that never outruns the evidence.
Position papers, second opinions, judgement reviews and regulator-facing responses — written by the people whose names go on them. Fixed fee, in writing.
Close, controls, reporting or risk — re-engineered and automated at process and governance level, handed over documented, with your team trained on it. Scoped outcomes, not day rates.
Agents in production under our supervision, senior review on top, service levels in the engagement letter. The function keeps the evidence; you keep the control.
Financial services is where the firm was built and where the standards are hardest. The method travels — from listed groups to owner-managed SMEs, any organisation with a close to run, a control environment to evidence, and routine work worth automating.
IFRS 9 ECL governance, prudential and liquidity reporting, ICAAP and ILAAP, capital and provisioning judgement, PRA-facing evidence.
IFRS 17 measurement and disclosure, reserving governance, Solvency II and ORSA, actuarial-to-ledger reconciliation.
MIFIDPRU and ICARA, client money and CASS assurance, trade-to-ledger completeness, settlement and margin controls.
Fund and AMC reporting, valuation governance, fee and NAV controls, consolidation across fund structures and SPVs.
First statutory audits, safeguarding, scale-up control frameworks, revenue recognition and data completeness at volume.
Portfolio reporting, acquisition accounting and PPA, carve-out and IPO readiness, LP-grade reporting discipline.
Long-cycle revenue, decommissioning and asset judgement, climate and transition disclosure with an evidence trail.
Multi-entity close and consolidation, inventory and margin controls, high-volume transaction analytics.
Revenue recognition at contract complexity, R&D and capitalisation judgement, rapid-growth control build.
The same standard, sized for smaller teams — a first controls framework, audit-ready books, a close that runs itself, and automation priced for SME budgets.
Building and running our own platform taught the firm how AI survives contact with a regulated profession. We work as a solution provider, not only an adviser — scoping, building, integrating and running the systems ourselves, from process mapping through to agentic systems in production, evaluated and supervised. Finance, audit, controls, tax, risk and the operations around them.
Reconciliations, matching, journal proposal, screening, drafting — executed by agentic systems at full population, against the live ledger, on the real deadline.
Review queues, approvals, audit trails, model documentation — every machine action evidenced, attributable and reversible. “The agent did it” is never the answer; the trail is.
Mapping close, reporting, and assurance workflows; deciding what is automatable, what stays human, and where the control points sit.
Ledger and sub-ledger data pipelines, full-population analytics, and the data management discipline that makes AI output defensible.
Scoping, building and evaluating agentic systems for finance and audit work — guardrails, retrieval, tool use, and the product decisions that get them adopted rather than tolerated.
Audit trails for AI-assisted work, model documentation, and the governance a regulator or audit committee will actually accept.
Getting agentic systems into production against real ledgers and real deadlines — integration, monitoring, human-in-the-loop review, and the run-book when something drifts.
Bringing the team with it: training, role redesign, and the operating model that decides who reviews what once the routine work is machine-done.
Every engagement runs on the firm’s own platform — standards codified into executable method, full populations tested rather than sampled, every output generated to inspection standard. Proven on the hardest evidential work; applied across the close, controls, reporting and risk.
The close is where automation pays first and pays most. We rebuild record-to-report end to end — data in from source, reconciliations run continuously, journals proposed and evidenced, consolidation and disclosure assembled — so the cycle compresses without a single control coming off the file.
Continuous reconciliation, automated accruals and allocations, consolidation and multi-GAAP reporting, disclosure assembly.
Touchless invoice matching, exception routing with a recommended resolution, duplicate and fraud screening at full population.
Cash application, credit and collections prioritisation, deductions and dispute handling, revenue completeness.
Variance analysis on close, driver-based reforecasting, board and investor pack generation from one dataset.
Prudential and statistical returns, tax provisioning and CbCR data, filing-grade audit trails on every submission.
Controls executed and evidenced by the same run — no separate testing exercise bolted on at year end.
Almost everyone intends to deploy. Very few have. The blockers are not model quality — they are data, governance, and the fact that “the agent did it” is not an answer an auditor or a regulator accepts. That is precisely the problem this firm is built to solve.
The tools the firm ships are designed in-house — interfaces finance teams actually want to use. A sample of the product and engagement work:
