In short: Lio Technologies, which builds multi-agent systems for procurement, finance and logistics and raised a $30M Series A led by a16z, runs nightly agentic regression with Agentiqa. A defined suite of test plans runs unattended every night against real browsers, real logins and real data, and every finding comes with video, screenshots and step-by-step reproduction. In Lio's words, Agentiqa surfaced five real production bugs in the first week alone.
About Lio
Lio Technologies (lio.ai) — building the world's first multi-agent system for Procurement, Finance and Logistics for enterprises, backed by a $30M Series A led by a16z — is running nightly agentic regression with Agentiqa.
The challenge: does the product still do what customers expect?
Lio ships fast. The open question was never whether the code compiles. It was whether the product still does what a customer expects once a feature is finished. That verification is the work Agentiqa does.
What is nightly agentic regression?
Nightly agentic regression means a defined suite of test plans runs unattended every night, with an AI agent executing each plan in a real browser and reporting what it found.
What the engagement covers
- Nightly regression. A defined suite of test plans runs unattended every night against real browsers, real logins and real data.
- Deliverable verification. A developer who has just finished a feature triggers a check from their own terminal, watches the recording, and reads the finding. No local checkout needed.
- Findings ready to act on. Every finding ships with video, screenshots and step-by-step reproduction — ready to consume and fix.
- Managed execution. Runs execute on Agentiqa's managed cloud, against Lio's own environment and shared accounts — provisioned and managed by Lio's own admin. Lio's engineers keep their machines free: they trigger and review, they don't host the browser fleet.
Results, in Lio's words
"What impressed me most is how easy Agentiqa is to integrate. With their CLI, our agents build and iterate on complex end-to-end test plans autonomously, in no time. Running them daily, we proactively catch regressions across all our systems and agents. In the first week alone, Agentiqa surfaced five real production bugs we'd never have caught without a full-time QA team."
— Vladyslav Melnychuk, AI Software Engineer, Lio
Why independent verification
"AI-written code should not grade itself. Lio's team ships faster than any human QA cycle can follow, so the verification has to be independent and it has to run while everyone sleeps. We are not replacing their engineers' judgement — we are giving them something to judge."
— Radik Zagirov, Builder & Cofounder, Agentiqa
FAQ
What does Agentiqa do for Lio?
Agentiqa runs nightly agentic regression for Lio: a defined suite of test plans runs unattended every night against real browsers, real logins and real data, and a developer can trigger a check on a finished feature from their terminal. Findings arrive with video, screenshots and step-by-step reproduction.
Where do Lio's test runs execute?
On Agentiqa's managed cloud, against Lio's own environment and shared accounts, which Lio's own admin provisions and manages.
What is nightly agentic regression?
A defined suite of test plans runs unattended each night, with an AI agent executing each plan in a real browser and reporting what it found.
Related reading
- What is agentic testing?: the method behind nightly agentic regression
- Why agentic QA is more than AI + a browser: why the verdict, not the clicking, decides whether you can trust a run
- Should the AI that wrote the code write its own tests?: why the verification has to be independent of the change
- What is a quality gate for AI agents?: what a release gate checks and why it also runs on a schedule
- Why tests pass but production breaks: why "the code compiles" and "the product does what customers expect" are different questions
- How to test AI procurement agents: what to check on requisitions, purchase orders and invoice matching before a release
Shipping AI agents that write to customer ERP, CRM or billing systems? Agentiqa sets up release gates for AI agents with write access: our engineers gate the agent's critical write flows before production, with the first gates live within 7 minutes. Book a 20-Minute Vertical Agents Safety Review.
Contact: hi@agentiqa.com