Bank core automation

Fiserv core automation when there is no API to call

"Fiserv core" is not one system. It is Premier, Signature, DNA, Cleartouch, Precision, and the newer CoreAdvance, each exposing a different screen. The teller and back-office workflows on most of them have no public API, so automating them means driving the UI. This is how that is done without recording brittle scripts that break on every release.

Matthew Diakonov, Written with AI

Published June 17, 20268 min read

Direct answer (verified June 17, 2026)

Yes, you can automate Fiserv core workflows but the legacy cores expose no public API for the teller and back-office screens, so automation has to drive the user interface. An accessibility-API agent reads the same Name, Role, and ControlType properties a screen reader sees. That works where browser bots and pixel-matching RPA fail, and it self-heals when Fiserv ships a quarterly release.

Source for the core lineup: Fiserv account processing platforms.

The map nobody draws: which Fiserv core exposes which interface

Most advice about Fiserv core automation talks about "the core" as if it were a single green screen. It is not. Fiserv sells a lineup, and the right automation approach depends entirely on what a given core puts on the screen. Here is the honest breakdown.

Core	Where the work happens	Public API for teller / back office?
Premier	Terminal / green-screen sessions	Limited; UI for most workflows
Signature	Windows desktop client	Limited; UI for most workflows
DNA	Windows desktop (teller, lending, back office)	Some; UI for the daily screens
Cleartouch / Precision	Hosted screens (folding into CoreAdvance)	Limited; UI for most workflows
CoreAdvance	Modern infrastructure on Premier base	More API surface than legacy cores

Platform names and segments verified against Fiserv's own platform pages on June 17, 2026. The pattern holds across the lineup: the daily teller, maintenance, and exception screens live in a UI, not behind a clean web service. That is the layer you actually have to automate.

What the agent targets: the accessibility node, not the pixel

Here is the part that makes Fiserv core automation hold up over time. Every Windows control, including the fields inside Premier's terminal emulator and DNA's teller window, publishes an accessibility node: a Name, a Role/ControlType, and an AutomationId. Screen readers use these. So does Mediar. You can see them yourself by pointing Windows Inspect.exe at any core screen.

post-transaction.ts

That selector is the anchor of the whole approach. A pixel-matching bot keys off where the "Post" button sits on screen. A recorded-selector bot keys off a fixed path that the next release can invalidate. The accessibility selector keys off what the button is. The engine that does this, Terminator, is written in Rust, MIT-licensed, and open source at github.com/mediar-ai/terminator, so your own RPA center of excellence can inspect exactly how it resolves a node before it ever touches the core.

One engine, every core, the workflows that actually matter

The agent watches a workflow once, learns it, then runs it against whichever core screen it was trained on. The interface technology on the left does not change how the work on the right gets done.

How a Fiserv core workflow flows through Mediar

8 wks → 2 wks

“Bank onboarding workflows that took eight weeks to stand up on legacy RPA have gone live in two, at roughly a fifth of the build cost.”

Mediar deployment numbers, banking core workflows

of a comparable UiPath build cost

per minute of runtime, no per-seat license

selectors or pixel matchers to maintain

Why this matters more on a core than anywhere else

Cores are not static targets. Fiserv pushes releases that move panels, rename fields, and reflow screens. On a recorded-selector bot, every one of those changes is a maintenance ticket: the run fails overnight, someone notices the queue did not clear, and a developer re-records the workflow. That maintenance tax is the real reason bank RPA programs stall, not the cost of the first build.

When the agent resolves the "Post" button by its role and name instead of its position, a moved panel usually does not break the run. A renamed label degrades gracefully and re-resolves. The class of failure that fills an RPA backlog after every core release mostly disappears. That is the difference between an automation you babysit and one you can leave running against a production core.

If you want the longer version of how this plays out across Jack Henry, Fiserv, and FIS cores, with the specific workflows banks automate first, the bank RPA breakdown covers card issuance, token deactivation, and CD renewals in detail.

Each core renders a Windows UI for its daily workflows. The accessibility tree exposes that UI, so the same agent approach covers all of them.

The Fiserv core lineup, and where automation fits

Premier

Host-based platform reached through terminal / green-screen sessions. The most widely deployed core in the US.

Signature

Windows desktop platform for larger and more customized institutions.

DNA

Windows desktop client with built-in teller, lending, and back-office screens (the former Open Solutions core).

Cleartouch

Hosted platform whose features are being folded into CoreAdvance.

Precision

Hosted core for community banks, also being merged into CoreAdvance.

CoreAdvance

The newer Premier-based platform; modern infrastructure, more API surface than the legacy cores.

Bring one Fiserv workflow that keeps breaking

Show us a teller, maintenance, or exception workflow on Premier, Signature, or DNA, and we will demo a watch-once agent driving it on a call.

Fiserv core automation, answered

Frequently asked questions

Can you automate Fiserv core banking workflows?

Yes. Most teller and back-office workflows on Premier, Signature, and DNA have no public API you can call, so automation has to drive the user interface. An accessibility-API agent reads the same Name, Role, ControlType, and AutomationId properties a screen reader sees, which lets it click, type, and read fields on cores that expose nothing over HTTP. That is the only honest way to automate a green-screen Premier session or a DNA teller screen end to end.

Does Fiserv expose an API for these core workflows?

It depends on the core and the workflow. CoreAdvance and Finxact are newer and carry more API surface. The legacy cores, Premier through terminal sessions and Signature and DNA as Windows desktop clients, were built before API-first design, and the day-to-day teller, maintenance, and exception screens are reachable only through the UI. That UI layer is exactly where accessibility-tree automation works and where browser-only AI agents do not.

How is this different from traditional Fiserv RPA bots?

Traditional RPA records coordinates or brittle selectors against the screen. When Fiserv ships a release that moves a field or relabels a button, the recording breaks and someone rebuilds it. Mediar targets the accessibility node by its Name and Role, so a label change degrades gracefully and the agent re-resolves the element instead of failing. There is no pixel matcher and no recorded selector to maintain.

What happens when Fiserv pushes a quarterly core release?

Core releases routinely move panels and rename labels, which is the single biggest maintenance cost of bank RPA. Because the agent resolves elements by accessibility properties rather than fixed positions, a moved panel or renamed field usually does not break the run. When a change is large enough to matter, the agent flags it and you adjust one step instead of re-recording the whole workflow.

Is this safe to run against a production core?

Mediar is SOC 2 Type II certified and HIPAA compliant, runs on-prem or in the cloud, and produces audit logs and validation rules for every run. You can scope an agent to a single workflow, run it against a test or training environment first, and require validation checkpoints before anything posts. It drives the same screens your tellers use, under the same permissions.

What does it cost compared to UiPath?

Pricing is $0.75 per minute of runtime with no per-seat licensing, plus a $10,000 turn-key program fee that converts to credits. There is no six-figure platform license and no per-bot tax. Across deployments that has worked out to roughly 20% of a comparable UiPath build, and bank onboarding workflows that took eight weeks to stand up have gone live in two.

Which Fiserv cores does this work on?

Anything that renders a Windows UI: Premier through its terminal / green-screen sessions, Signature and DNA as Windows desktop clients, and the hosted Cleartouch and Precision screens. If a teller or back-office user can see and click it on screen, the accessibility tree exposes it and the agent can drive it.