auxfirst is an agentic experience design practice founded in Warsaw, Poland, by Emil Krzemiński, working with clients across Europe in English and Polish.
We exist because the hardest problem in agentic AI is not making agents capable. It is making delegation to them safe, legible, and worth trusting.
Every organisation adopting AI agents faces the same set of questions: What should this agent be allowed to do on its own? When must it ask? How does it explain what it did? Who is accountable when it acts? These are not engineering questions. They are experience design questions — and almost nobody is answering them deliberately.
auxfirst designs the human side of agentic systems: how people delegate work to agents, how autonomy is scoped and escalated, how memory and identity are made visible, and how trust is built through interaction rather than asserted through marketing. We call this discipline Agentic User Experience — AUX.
Our work produces concrete artifacts, not slideware:
These are things your product and engineering teams can build against. The methodology behind them — the auxfirst AUX Framework, including the Trust Architecture, the Action Heat Ladder, and the open-source TrustKit — is documented openly across this site.
Founder-led
Every engagement is led personally by Emil Krzemiński. Strategy, workshops, and the design decisions that matter are never handed to a junior team you didn't meet.
Partnered delivery
For engineering, research, and delivery at scale, we work with a defined network of specialist partners, assembled per engagement based on scope. You will always know exactly who is doing what before work begins.
Fixed-scope
Engagements are fixed-scope wherever possible — a clearly defined audit, sprint, or design engagement with named deliverables over open-ended retainers. It keeps incentives clean on both sides.
Where to start depends on the question: an Agent Experience Audit when the agent exists but people won't trust it, an Agent Operability Audit when the question is whether a workflow is ready for one, or a Blueprint Sprint when you're designing from scratch. The full menu is under Services.
Before founding auxfirst, Emil spent 12 years working at the intersection of enterprise technology, revenue operations, and product — including engagements involving IBM, Procter & Gamble, Coca-Cola, and Danone, via roles at Netguru and Listonic.
That background shaped a specific conviction: enterprises don't fail at AI because the technology is immature. They fail because nobody designed the moment where a human decides whether to trust the machine.
AUX is the practice of designing that moment — and everything that surrounds it.
We are based in Warsaw and work across Poland, the United Kingdom, and the wider EU. Delivery is remote-first, with on-site workshops available across European time zones. We work in English and Polish.
Formal registration details are being finalised for publication.
A founder-led design practice. Strategy and design are led by the founder; engineering and research are delivered with specialist partners scoped per engagement. We use "agency" as shorthand, but you're hiring a practice, not a bench.
Warsaw, Poland. We work with clients across Poland, the UK, and the EU, remotely and on-site.
Emil Krzemiński, the founder, leads every engagement personally.
English and Polish.
The design discipline concerned with how humans delegate to, supervise, trust, and correct AI agents. AUX is an emerging field; the auxfirst AUX Framework is our methodology within it, drawing on human–AI interaction research, trust calibration, automation bias, and agent safety work.
That's the whole practice. Tell us the workflow or the product, and we'll show you where trust breaks — and what to design first.
Start a conversation