This AI Startup Just Made Workflow Training 10x Faster

Jan de Vries · 2026-08-20

If you've ever had to create a training video for a new software tool, you know the pain. Record your screen, edit out the mistakes, add captions, and pray your colleagues actually watch it. Most of the time, they don't. They just ask you directly or open a support ticket instead. That's exactly the problem Vilnius-based Guideless is trying to solve. The AI-powered software training platform just raised $1.06 million in a pre-Seed round to help companies slash the time and money spent on building and maintaining workflow training materials. ### Who's backing Guideless? The round was led by Superhero Capital, with participation from FIRSTPICK VC and a group of notable angel investors. That list includes Thomas Plantenga (Group CEO of Vinted), Vytautas Atkočaitis (Vinted Go CEO), and Mantas Mikuckas, who co-founded Vinted. When seasoned operators from a unicorn like Vinted put their money behind a tool, it's worth paying attention. "Workflow and training automation is going to be a huge market over the next decade, but most of the current tech isn't forward-thinking enough to get there," said Audrius Milukas, Partner at Superhero Capital and lead investor. "We backed Guideless early because they do more than eliminate the grunt work of creating and updating training videos — they're fixing problems companies don't yet realise they have." ### How Guideless actually works Founded by Evaldas Bieliūnas and Dovydas Remeika, Guideless soft-launched in late 2025 and officially incorporated in June 2026. The premise is simple: instead of recording your screen and hoping for the best, you just perform the workflow while the platform runs in the background. The AI captures each step, writes the text instructions, adds visual highlights and zoom effects, generates a natural-sounding narration with captions, and applies your company branding. The result? Training guides that are up to 10 times faster to create compared to traditional methods. And since every step is editable, you can update a single step without re-recording the entire video. That's a game-changer for teams that constantly tweak their processes. ### The hidden cost of bad documentation Here's the irony: companies spend a fortune creating extensive training materials, yet employees often ignore them. They'd rather ping a coworker or open a ticket than dig through a 20-minute video. This isn't just annoying — it's expensive. The market for training, knowledge management, and AI-driven workflow automation is projected to hit $21.5 billion by 2031, according to a report cited by the company. Bieliūnas, who spent over a decade leading commercial teams and onboarding employees across multiple countries at companies like Vinted Go, knows this pain firsthand. "Another standard documentation tool wasn't going to solve this," he said. "We designed Guideless to convert every workflow it captures into structured operational context, eventually forming an evolving memory layer that preserves institutional knowledge." ### Why this matters for AI agents too Here's the clever part: because each guide is captured at the step level rather than as a single fixed recording, the same operational context can serve as reliable, up-to-date input for AI agents. That means an AI assistant could one day understand exactly how your company handles refunds or customer onboarding, based on your actual workflows — not a generic template. The platform currently supports narration in over 40 languages, with more on the way. It combines browser-based capture, computer-vision step detection, generative AI, and multilingual text-to-speech into a single automated pipeline. ### The bigger picture Guideless isn't just building another training tool. It's positioning itself as the default workflow intelligence layer for software-driven organizations. As teams, tools, and processes change, the platform preserves institutional knowledge in a structured, usable format. That's a bold vision — and with backing from some of Europe's most successful operators, it just might work.