Financial services are moving from dashboards to decisions. From applications to agents. From passive data access to delegated financial action. Agentic Open Finance™ is the consent, authority, governance, and evidence layer for financial agents across banking, payments, lending, insurance, pensions, wealth, utilities, payroll, and trade finance — helping organisations design, govern, deploy, and commercialise agents that read, reason, recommend, and act only within approved consent, policy, jurisdiction, risk, and evidence boundaries.
The problem
Open banking and open finance focused on access; the next phase is delegation. Agents will monitor cashflow, prepare lending evidence, spot better products, time payments, manage recurring mandates, detect risk, explain decisions — and act. But in finance, action is not just automation. Action requires authority. Financial agents are coming, and most organisations are not ready: the moment an agent enters the workflow, the core question changes from "can the system do this?" to "who authorised it — and what exactly?"
Consent is granted once, broadly, in advance. The action happens later, specifically, at machine speed. Between those two moments lives every incident report this industry will ever write.
An agent permitted to "manage payments" will, eventually, manage one you did not mean. Not malice — inference. A standing permission plus a reasoning engine is an unbounded mandate.
Observability reconstructs behaviour. The auditor, the ombudsman and the regulator ask about authority: in scope, live grant, named approver. Most stacks simply cannot answer.
Without a clear authority layer, agentic finance becomes a liability. Agentic Open Finance™ starts with authority.™
Anatomy
Authority before autonomy.™ Not a workflow diagram — an authority chain. It plays below; tap any step to jump.
The category
The use of authorised financial agents to perform governed tasks across open banking, payments, lending, insurance, wealth, pensions, payroll, utilities, trade finance and other financial data ecosystems. It is not just data access. Not just automation. Not just AI chat. Not just a dashboard. Consent and authority are different instruments — put them side by side once, and the gap stops being abstract.
Doctrine
Authority before autonomy.™ If the industry adopts nothing else, it should adopt these. Each closes a failure class consent tokens were never designed to close.
An agent should never act simply because it can. No standing permission: a grant is scoped to the act — this payee, this ceiling, this window — and checked at execution, not at onboarding. Boring by design: boring is what auditable looks like.
Consent names a purpose, a scope, an agent and a window — never a blanket. A token that opens everything authorises nothing in particular, and the agent allowed to pay this morning must be allowed again this afternoon.
Rules that live in PDFs govern nothing. Policy is executable — it allows, denies, escalates or demands approval before the action, not in the post-mortem. A denied act never executes; an escalated act waits.
Every verdict is sealed when it happens: the mandate that applied, the grant that was live, the person who bound it. Not logs assembled after the incident — evidence that already exists before the question is asked. March's dispute is answered from January's seal.
The same agent, the same action, a different regime — a different answer. Jurisdiction is resolved before execution, natively, never retrofitted. An action lawful in one market stops quietly at the border of another.
Notification, approval and revocation stay in the customer's hands — and finality has a name. Above the line — new payee, new ceiling, irreversible transfer — the act stays provisional until a person with the standing to bind says so. When the regulator asks "who authorised this?", the answer is a name, never a model version.
Every governed action is attributable, meterable and billable from day one — a commercial event with evidence attached, not an afterthought. If you cannot govern how an agent earns, you cannot govern the agent.
Positioning
We help you define what agents may see, recommend, initiate — and never do; when human approval is required; when a mandate is valid; when consent expires; when a policy blocks action; when a customer must be notified; when evidence must be preserved; when a workflow escalates; when an action is commercially billable; and when an agent is safe to deploy. Not a chatbot with compliance added later — an execution spine agents cannot bypass. The agent proposes; the rails decide; the record outlives both.
The platform
A horizontal control plane, sold as focused modules — each one useful alone, all of them sharing the same registries, policies and evidence spine.
Agent catalogue
Twelve governed agents — each one registered, scoped, policy-bound and evidence-native. Every agent runs inside a live mandate and can be refused, escalated or revoked per action. None of them can exceed what it was allowed to do.
Reads accounts, forecasts cashflow and proposes payment timing — and cannot exceed its mandate.
SMEs · accountantsWatches recurring charges, flags creep and duplicates, and prepares governed cancellations.
households · subscription businessesMonitors recurring mandates for limits, expiry and drift — escalating before an unwanted charge, not after.
payments teams · billingAssembles consent-scoped affordability evidence into a decision-ready pack.
lenders · brokersMatches invoices, payments and accounts — and files exceptions with a reviewer instead of guessing.
finance operationsWatches payroll runway and reserves, warning before payday is at risk.
payroll providers · SMEsSets aside tax reserves under governed sweep rules — so the quarter never arrives as a surprise.
SMEs · accountantsPrepares a business to borrow: evidence, ratios and gaps — before the application, not during it.
SME platforms · lendersPrepares renewals with context — cover, claims and affordability — under explicit consent.
insurers · brokersReviews outcomes against duty-of-care boundaries and flags the ones that need a human.
compliance · customer careReconstructs any disputed action from its sealed evidence — the mandate, the verdict, the approver.
support · legal · auditContinuously checks live consents against actual use — expiring, revoked or overreached.
compliance · payment firmsSolutions
The gap looks different from every seat at the table. The fix is the same discipline — consent, authority, policy, evidence — applied to each sector's own workflows.
You are the account. You wear the loss. When a third-party agent misfires against an account you hold, the customer calls you — the token was valid, the connection worked as documented, and the money is still gone. Your defence is not "consent was granted". It is "the action was checked against a live mandate, and here is the sealed record."
Representative work: agent access policies, per-action authority checks, dispute-ready evidence, agent-safe payment products. The bank that can prove per-action authority will set the terms of agentic access. The bank that can't will switch it off — and lose the channel.
Nobody lets a brilliant agent near serious money on brilliance alone. The unlock is bounded autonomy you can demonstrate: show that your agent cannot exceed its mandate even when its reasoning says it should. "It asks before it binds" is a feature you can sell. "Trust the model" is not.
Representative work: governed agent features, consent-scoped data products, partner assurance evidence, faster risk sign-off on new agent launches.
Client permissions become explicit, scoped and renewable — and the practice becomes the governed operator of its clients' financial agents rather than a spreadsheet janitor.
Representative work: client cashflow monitoring, accountant–client permission management, tax reserve sweeps, lending readiness preparation.
Affordability stops being a PDF bundle and becomes a consented, verifiable evidence pack — assembled by a governed agent, defensible months later.
Representative work: affordability evidence packs, consent-scoped verification, decision evidence for underwriting, dispute replay on contested outcomes.
Renewals, claims and cover checks gain live financial context — under explicit consent, with vulnerable-customer safeguards enforced by policy rather than training slides.
Representative work: renewal preparation with context, contextual cover checks, claims evidence, fair-value and vulnerability safeguards.
Discovery, consolidation prompts and drawdown guardrails run under suitability boundaries — with escalation to a named adviser designed in, not improvised.
Representative work: discovery under consent, suitability boundaries on recommendations, adviser escalation, evidence of the advice context.
Switching support, payment plans and collections gain affordability context — governed, notified and evidenced, so help arrives before arrears and dignity survives the process.
Representative work: switching support, affordability-aware payment plans, early arrears detection, governed collections journeys.
Mandates with limits, expiry and transparent evidence turn billing disputes from churn events into replayable records — and failed-payment recovery into a governed retry, not a surprise.
Representative work: mandate design, failed-payment recovery, churn-safe retries, transparent billing evidence.
Tenant affordability, rent mandates and deposit evidence run under scoped consent — the letting decision and the arrears conversation both get a defensible record.
Representative work: tenant affordability evidence, rent mandate design, deposit evidence, arrears workflows.
Counterparty checks, invoice verification and payment scheduling become governed agent tasks — with the evidence trail financing partners actually ask for.
Representative work: counterparty evidence, invoice verification, payment scheduling under mandate, financing readiness packs.
Failure files
None of these require a malicious agent. Each one only requires a standing permission, a reasoning engine, and time. Each has a structural fix — none of the fixes is "better prompts".
An agent told to "reduce monthly costs" cancels a subscription the user was mid-dispute on, then switches an insurance product with a worse exclusion clause. Every individual call was inside the consent. The outcome was never meant.
Structural fixProduct-switching and cancellation are separate action classes with their own grants — "reduce costs" is a goal, not an authority.
A payment agent resolves an invoice to a counterparty the account has never paid, from an email thread it was asked to "handle". The transfer clears. The invoice was fraudulent. The consent token was valid the entire time.
Structural fixFirst-time payees are above the finality line by default — provisional until a named person binds the act.
The employee who configured the treasury agent leaves the company. The agent keeps sweeping balances for months under a delegation nobody currently holds. Every sweep is logged beautifully. None of them had a living mandate behind it.
Structural fixGrants are checked live against current authority — an authority that lapsed an hour ago fails at the next action, not at the next audit.
A customer disputes a transfer their agent made in January. It is now June. The provider has request logs, the bank has settlement records, and neither can prove what the user's mandate actually permitted at that moment. The regulator reads both files and believes neither.
Structural fixThe verdict, the live mandate and the approver are sealed at the moment of the act — the dispute replays the seal, not the recollection.
The artefact
Not a legal PDF and not a scope string — a small, checkable object that exists per agent, per action class, and dies on schedule.
Everything on this card is checkable by a machine and readable by a human. That double property is the whole trick: the agent can be refused in microseconds, and the ombudsman can understand the refusal in seconds.
Notice what is absent — no model name, no prompt, no vibe. Authority attaches to the act, not to the intelligence proposing it. Swap the model, keep the mandate; upgrade the agent, keep the ceiling.
And notice the expiry. A mandate that cannot die was never a mandate — it was a transfer of ownership nobody priced.
Why now
From open banking to open finance to agentic finance.™ Access came first; then programmable payments, customer control, automation, accountability, AI adoption — and now agentic systems inside real workflows. Delegation without authority is dangerous. Automation without evidence is fragile. Personalisation without consent is risky. Financial action without governance is unacceptable.
Products & programmes
Six commercial products — each one a fixed-scope engagement with named deliverables, all of them feeding the same registries, policies and evidence spine.
Opportunity map, risk map, agent use-case design, consent and authority model, product roadmap, governance checklist, commercial SKU map, board-ready briefing.
Book a Readiness SprintPayment use-case mapping, recurring mandate design, journey review, consent scope mapping, failure and dispute scenarios, evidence requirements, revenue model design, partner readiness.
Design Agentic Payments™One high-value workflow, governed end to end — pilots drawn from the agent catalogue above, from cashflow to consent compliance.
Launch a PilotAgent risk register, consent checklist, authority model, policy controls, evidence requirements, incident playbook, dispute replay model, board briefing, customer explanation templates.
Get the Evidence Pack™Agent catalogue, ownership model, approved capabilities, prohibited actions, approval thresholds, customer-facing descriptions, evidence requirements, governance map.
Create Your Agent Registry™Training, certificates, templates, playbooks, co-branded workshops, referrals, sector packs, partner directory, delivery frameworks.
Become a Partner™Engagements run from a readiness assessment through pilots, platform adoption, certification programmes and partner delivery — education and assessment come first, so governance never waits for procurement. Detailed pricing is shared with registered prospects.
Academy
Six certificates — one per seat at the table. Every certificate is casework on the same control model this page describes: consent, authority, policy, evidence.
Templates & partners
The control model travels: providers plug in beneath it, professional firms deliver on top of it, and every partner ships the same templates their clients' regulators will eventually ask for.
Straight answers
Asked by banks, builders and regulators — answered without hedging.
The use of authorised financial agents to perform governed tasks across open banking, payments, lending, insurance, wealth, pensions, payroll, utilities and trade finance — with consent, authority, policy and evidence resolved per action. It is not just data access, not just automation, not just AI chat, and not just a dashboard: it is the control layer that makes delegated financial action defensible.
No. Banks matter because they hold the account — but the same control model serves fintechs, payment firms, lenders, insurers, wealth platforms, accountants, payroll providers, utilities, property platforms, subscription businesses and the public sector. Anyone whose software can move, commit or evidence money needs an answer to "who authorised this?"
No. It is the layer organisations use to design, govern, deploy and commercialise financial agents. Customers feel it indirectly — as clearer consent, safer mandates, honest notifications and outcomes that can be explained back to them.
Automation repeats a defined task; an agent reasons and chooses. In finance a choice is an action with consequences — money moves, products switch, credit extends. That is why an agent needs authority, policy and evidence per action, where a scheduled job only ever needed a checklist.
Only when the consent, mandate, policy, approval and evidence requirements are all satisfied — and above the finality line, only after a named person binds the act. A governed agent that cannot prove its authority does not act. That is the point.
No. SCA proves a human is present and is who they claim at a login or payment moment. Agentic finance is defined by the human being absent — the entire question is what the software may do while nobody is watching. Authentication answers "who?"; a mandate answers "what, how much, until when, and who carries it?"
A mandate check is a lookup against a small object, not a committee meeting — microseconds, not minutes. The slow path is reserved for the acts you'd want slow anyway: new payees, raised ceilings, irreversible transfers. Below the line, governed agents run exactly as fast as ungoverned ones. The difference only appears when something should stop.
A model's judgement is a capability, not a boundary. It improves on average and fails on the tail — and finance is priced on the tail. The mandate exists precisely for the day the reasoning is confidently wrong. "The model would never do that" is not a sentence anyone should have to say to a regulator.
The mandate belongs to the principal — the customer or firm whose money moves — and must be checkable by whoever executes. In practice that means the executing institution enforces it, the interface renders it, and the builder honours it. Anyone in the chain who can't check it shouldn't be executing.
No — consent still opens the door, and nothing here removes it. Authority is a second layer under the same door: consent admits the agent to the account; the mandate decides, act by act, what may bind. Regimes evolve by addition here, not demolition.
Inventory. List every agent that can touch money, what each is currently able to do, and under whose standing permission. Most teams discover the honest answer to the third column is "nobody is sure" — that inventory is the business case, and everything on this page follows from it.
The Agentic Open Finance™ Readiness Sprint. It packages that inventory with an opportunity map, a risk map, a consent and authority model and a board-ready briefing — so the first pilot starts from a governed baseline instead of a blank page.
Get started
Agentic Open Finance™ helps you move first — without losing trust. Financial agents are inevitable. Ungoverned financial agents are optional.™ The winners will know who authorised the agent, what it was allowed to do, which policy applied, what evidence exists, how to explain the outcome — and how to scale safely.
Agentic open finance will be judged not by what agents could do — but by what they were allowed to do.
— Authority before autonomy.™
Tell us who you are and what you need — a person replies, usually within one business day.