Skip to main content

Agentic coding workshop for banks and financial services

Banks want the productivity of AI coding agents, but every new tool touches ICT risk management, outsourcing rules and audit trails. Our workshop teaches your developers agentic coding in a way your risk, compliance and information security teams can sign off.

ISO 9001 & ISO/IEC 27001Certified quality and information securitySince 2014, made in GermanyDiese Seite auf Deutsch

Agentic coding inside a regulated IT landscape

In a bank, the question is rarely whether Claude Code, OpenAI Codex or Cursor can write good code. The questions are whether the provider counts as an ICT third-party service under DORA, where prompts and code snippets are processed, how the four-eyes principle applies to agent-generated changes and how an auditor can trace who approved what. A developer workshop that ignores these points produces enthusiasm in the team and a veto from the second line of defense.

neooutput, the AI education unit of jocapps GmbH, designs the workshop together with your engineering leads and, ideally, someone from information security. Developers practise on typical banking code: modernizing a legacy batch job, writing tests for interest or fee calculations, documenting an undocumented interface and hardening an internal API. In parallel we work out the guardrails that fit your environment, from approved models and data classes to logging and review rules. OnlineAppCreator is certified to ISO 9001 and ISO/IEC 27001, so these discussions are familiar ground for us.

What to watch out for

  • Assess the AI coding provider as an ICT third-party service before rollout. Under DORA this may require a register entry, a risk assessment and contract clauses on audit rights and exit strategies.
  • Never let agents see customer data, account numbers or production extracts. Test data should be synthetic or properly anonymized, and the working directory should contain nothing above the approved data class.
  • Agent-generated code is still your code. The four-eyes principle, secure coding standards and static analysis apply to it exactly as they do to manually written code.
  • Check model hosting regions and retention settings in the enterprise contract. Default consumer settings are usually not acceptable for source code that implements risk or payment logic.
  • Involve the works council early if agents change how work is distributed or measured. In German institutions the introduction of such tools is typically subject to co-determination.

What your bank gains

Guardrails that pass review

We translate your data classification and outsourcing policy into concrete agent settings, allowed repositories and forbidden data. Your information security team receives a description it can actually assess.

Legacy code becomes approachable

Agents are strong at explaining old code and writing characterization tests before anyone changes it. That lowers the risk of modernizing systems whose original authors left long ago.

Traceable changes

Developers learn to keep agent work in small, reviewable pull requests with clear descriptions. Change management and auditors see the same evidence they expect from manual work.

A path to governed AI use

For use cases beyond coding we show how a governed platform such as kalAIDoCreate fits. Business units then get AI access with policies and logging instead of private accounts.

How we run the workshop

  1. Scoping call

    We clarify stacks, sourcing restrictions, approved tools and who from risk or security should take part.

  2. Guardrail draft

    Before the session we prepare a draft of usage rules, data classes and review requirements for your setting.

  3. Hands-on days

    Developers work with the approved agent on sample or sanitized banking code, remote or on-site.

  4. Pilot plan

    We close with a pilot scope, success criteria and the documentation your committees need for a decision.

Matching products

neooutput

Agentic coding workshops and AI services. Hands-on agentic coding workshops for teams on Claude Code, OpenAI Codex and Cursor, remote or on-site, in English or German. Plus AI Code Engineers on demand who deploy, secure and scale AI-built apps.

Explore neooutput workshops

kalAIDoCreate

Governed AI platform for enterprises and the public sector. An all-in-one AI platform for enterprises and public institutions with compliance features, governance and custom APIs.

Explore kalAIDoCreate

Frequently asked questions

Can we run the workshop without cloud-based AI models?

The tools we teach rely on hosted models, but enterprise plans offer controls for data retention and regions. If your policy rules out external processing, we discuss which parts of the workflow can still be trained with sanitized sample code.

Which tool do you recommend for banks?

That depends on your existing contracts and IDE landscape. We compare Claude Code, OpenAI Codex and Cursor against your requirements and avoid a blanket recommendation.

Who should attend besides developers?

An engineering lead and a person from information security or IT risk make the biggest difference. They see the same exercises and can judge the guardrails from practice rather than slides.

Do you support the pilot after the workshop?

Yes, if you wish. Our AI Code Engineers can work alongside your team during the pilot, and neooutput offers follow-up sessions once the first results are in.

Let's talk about your project.

15 minutes with an AI Code Engineer, free and without obligation. Or call our team around the clock.