Atsikratykite rankinio darbo: two days of AI work in Kaunas
On 16 and 17 June 2026, ChangeAI ran its own two-day intensive, Atsikratykite rankinio darbo ("Get rid of manual work"), at DOKAS in Kaunas. Odeta Isevičiūtė designed and delivered it for operations managers and COOs.
The event
This was ChangeAI's first two-day workshop: a compressed, hands-on version of ChangeAI's four-week programme, built for people who run operations and see repeated manual work in their teams every day. It was a small group, taught in Lithuanian, with every participant working on their own laptop and their own processes.
The promise was concrete. Each participant builds three AI assistants and at least one automation with Claude Cowork and Make.com on their own work, and leaves with a 90-day plan for their team. Two group support calls follow in the weeks after.
What we did there
Before the workshop, each participant sent in three repeated processes from their own work, examples of how they already use AI, and the tools they have. Odeta used those to shape the two days.
Day one covered prompting for content, data analysis and presentations, then working with search and large documents in NotebookLM and Claude Projects. In the afternoon the whole group built an email assistant together, and each participant then built a second assistant in small groups around their own process. Day two moved to automations: when an assistant is enough and when you need an automation, a live Claude Cowork demo on a real business process, then Cowork practice and a first full automation with Make.com. The day ended with a third assistant and each participant's 90-day plan.
Three takeaways from the workshop
1. Start from your own repeated process
Every exercise used a process the participant had brought in from their own work. Odeta opened the workshop by asking each person to name the biggest manual work pain in their team in one sentence. Starting there means whatever gets built goes back to work with them.
2. Give the assistant context and data
One slide summed up most of day one: "Kontekstas = duomenys = atsakymų kokybė" (context = data = quality of answers). AI doesn't read your mind and doesn't know your problems or their nuances. Give it no context and you get generic answers. So the email assistant was built with real instructions, real rules and 5–10 emails each participant had written themselves. A second slide set expectations: "AI negali padaryti visko, ką mes sugalvojam" (AI can't do everything we come up with). Language models predict the next word, and AI companies add tools on top, like search, app connections and code. If AI won't do what you imagine, ask whether it actually can.
3. Automations come after the process is clear
An assistant works when you start it. An automation works without you: an email arrives and it gets sorted, an order comes in and it gets processed. That difference shaped day two. Participants picked their third assistant only after seeing what Cowork and Make.com can do, so each of them could tell which processes fit an assistant and which are worth automating. An automation repeats a process exactly as it is, so the process has to be clear first.
Upcoming ChangeAI workshops are listed on the workshop page.
Sources and notes
- ChangeAI's own event, with no external event page. Programme details come from ChangeAI's internal workshop programme. Participants and their companies are not named.