AI Adoption in 2026: What Worked, What Didn't (Design Studio POV
Hosted by Motka (Phil Smirnov) / Portugal Tech Week
Talk at PTW Palace on what worked and what didn't when a design studio rebuilt its workflows around LLMs.
20 years of a design studio's work sat in three disconnected places: a web archive for the earliest years, email from 2008, Slack from 2014. Over the past six months we rebuilt Motka, a product design studio running since 2005, around LLMs. Not a pilot on the side: the studio's knowledge now lives in a curated wiki of roughly 140 pages that an agent reads before it writes a proposal, a case study or a post, and the whole team works from it. This talk is the honest ledger. What you'll leave with The order the steps have to happen in, and the one most people skip and then stall on Practical context strategies: how to collect it, how to QA it, how to retrieve it Three layers of company data: input, facts, calculated How to organize and maintain an internal wiki, and where to start one Automated bans, clarifications and edits that keep a self-written knowledge base from drifting The main chokepoints and the ways around them What clients paid for, and what broke along the way Who it's for Founders and operators of small service and creative companies: agencies, studios, consultancies, small product teams. People with an archive nobody can use and no AI department coming to help. Speaker Phil Smirnov, founder of Motka design studio (Ljubljana). 300 projects since 2005, data-heavy B2B products: trading platforms, media and analytics systems, EdTech. Half of the studio's current work is designing AI products. Format A 45-minute talk in English plus questions, with one short screen…
