UiPath: 70-80% of devs use AI but QA adoption sits under 50%

TL;DR:

• UiPath CEO Daniel Dines and testing SVP Gerd Weishaar say 70-80% of developers now use AI tools while QA adoption of AI sits well under 50%, creating a testing bottleneck just as DORA and the EU AI Act tighten enforcement.
• UiPath has launched UiPath for Coding Agents, an orchestration platform that integrates Anthropic’s Claude Code and OpenAI Codex with enterprise governance, audit trails, role-based access control and runtime guardrails.
• Customer reference points include Bank Dhofar, Bank of East Asia, First Abu Dhabi Bank and Sweden’s Ikano Bank; UiPath explicitly positions QA teams as governance and release-approval gatekeepers rather than test-case writers.

This is the operational consequence of the agent-failure pattern showing up in adjacent stories this week. The Independent’s PocketOS write-up illustrated what happens when agentic AI hits production without governance scaffolding; UiPath’s pitch to banks is to sell that scaffolding. For UK fintech leaders, the picture is now squarely about how to add governance around AI code velocity, not whether AI codegen is real.

The bottleneck and the platform response

Weishaar’s framing is direct: traditional testing is already “the biggest bottleneck to delivering new innovations” in regulated sectors; AI-generated code at the front of the pipeline only makes the bottleneck worse. UiPath’s response is to bring the governance to the agent rather than the agent to the governance team. The orchestration layer connects Claude Code, OpenAI Codex and (eventually) others into enterprise CI/CD, with audit trails and policy enforcement applied automatically.

Dines framed it as a structural shift: “Now, anyone can describe what they want, direct a coding agent to produce it, and carry through every stage to production. It lowers the barrier to who can build, crossing the line from idea to execution.” That promise – broader builder pool – is the productivity case. The countervailing point is that the same shift makes governance a load-bearing function, not a compliance add-on.

Why this matters for UK banking

Two UK-relevant pressures sit behind the bottleneck story. DORA – the EU Digital Operational Resilience Act – applies to UK financial entities with EU exposure, and its requirements on testing, third-party risk and incident reporting cover AI-generated code paths even when those paths are not explicitly named. The FCA has separately signalled, in this month’s “evidence not theory” push, that it wants banks to demonstrate concrete AI testing practices. Weishaar’s blunt admission – “Seeing huge or large amounts of code being generated in front of you and being responsible that this code does exactly what it’s supposed to do is a big burden… I don’t think it’s easily doable” – is the regulator-facing risk in two sentences.

Looking forward

Expect UK banks to push testing budgets up faster than headline IT spend in 2026 as the gap between developer-AI adoption and QA-AI adoption closes. The interesting commercial signal is which orchestration layer wins – UiPath has first-mover positioning, but Microsoft, Atlassian and the cloud providers all have credible plays. For UK SMEs supplying QA, audit-trail or governance tooling to banks, the market is now framed by the developer-vs-QA adoption gap.