There is work that needs a person. And work that just needs to get done. We help you figure out which is which. So your people have more time for what people are genuinely good at.
Give us 30 minutesWe're a small team. On purpose.
Whatever can be automated, we automate ourselves too. That leaves more time for what people do best: listen, understand, decide and shape.
Ann-Christine Vossberg-Straw
systems & discovery
and today works on making complex systems understandable and changeable. At Curious, she listens, asks the right questions, and finds out where change actually helps.
ETH Zurich · astrophysics · systems thinking
Pasqual Vossberg
product & delivery
— and turns complex requirements into solutions that work in daily use. At Curious, he makes sure a good idea becomes something that actually gets used.
business · design thinking · digital products
Jenny Korzun
brand & communication
turning complex technology into a story people understand. At Curious, she keeps an eye on how a good solution also becomes one that finds its market.
brand · communication · market presence
Alex Oleshkevich
technology & architecture
Alex thinks about systems from the ground up: architecture, backend, frontend, infrastructure and security. At Curious, he makes sure an idea becomes robust technology.
software engineering · architecture · open source
we start with questions
What's annoying?
What do you do over and over again?
Where would you love an assistant — someone you just hand it to and never think about again?
What gives you a stomachache before you've even started?
Where do the same mistakes keep happening?
What eats up time even though it should be simple?
What does nobody actually want to do?
Where does information get lost?
That's exactly where we get curious.
Focus
We look for the places where good people get slowed down: tedious work, postponed tasks, repetitive processes, and the things nobody wants to do.
No AI consulting call. No chatbot. No dashboard. No new app.
Solution
There, we build AI assistants that use your existing tools and take over or prepare concrete work.
So people get back to the work that actually moves their organization forward and makes it stronger.
Not every good AI idea starts with a new tool. It often starts with a place where work keeps getting stuck.
We look for that friction. Then we build assistance right there: small, concrete, and with the tools your team already uses.
It is almost always one of these.
Then you are checking whether we are real. This page is the answer, and the four names above are the whole company.
You should not, yet. Take the free thirty minutes and judge the written answer.
It stops and asks a named person. Being wrong quietly is the failure we design against.
No. The mess is the job.
You do. Your repository, your infrastructure, nothing to license.
A developer builds what you specify. Here the hard part is the specification.
Good. We take the process nobody has had time to finish. If your team has the time, do not call us.
Yes. Four. You will meet all of them in week one.
Then it is a bad fit, and we will say so on the call.
Eight kinds of stuck work that look the same in almost every company. Open one to see what fits AI assistance — and what stays with your team.
what gets stuck
Availability, absences, qualifications, locations and last-minute changes live in different heads, chats or lists.
what AI assistance prepares
Collects constraints, detects conflicts, drafts a plan, marks unclear cases.
what stays human
Fairness, final approval, exceptions, relationship decisions.
what gets stuck
Hours, expenses, surcharges and edge cases for monthly and final billing come from different sources and get merged by hand.
what AI assistance prepares
Pulls the numbers together from the sources, checks them against the rules, marks what needs a human look.
what stays human
Approval, edge cases, the conversation when something does not add up.
what gets stuck
Requests, complaints and internal messages land in one inbox, and someone has to decide by hand what comes first.
what AI assistance prepares
Reads every message, sorts it by urgency and topic, drafts a first reply.
what stays human
The tone on sensitive cases, the final call, what is allowed to wait.
what gets stuck
New enquiries, old contacts and open follow-ups sit side by side, with no clear read on who comes next.
what AI assistance prepares
Sorts the list by recency and likelihood, suggests the next calls to make.
what stays human
The call itself, the tone, the decision who makes it.
what gets stuck
The same information sits in several spreadsheets, written slightly differently, and nobody is sure which version is right.
what AI assistance prepares
Reconciles the spreadsheets, flags the discrepancies, proposes a cleaned-up version.
what stays human
The call on which version is right when the sources disagree.
what gets stuck
What one shift knows does not fully reach the next — context gets lost between team, office, site, workshop or practice.
what AI assistance prepares
Summarises the state, gathers open points, prepares a clean handover.
what stays human
The conversation on complex cases, responsibility for what gets handed over.
what gets stuck
Every document has to be checked against the same list, and the check eats time without being hard.
what AI assistance prepares
Checks each document against your list, flags what is missing or off.
what stays human
Judging the exceptions, the final sign-off.
what gets stuck
Regulatory documentation has to be complete, current and traceable, and checking that by hand costs time every week.
what AI assistance prepares
Checks the documentation against the requirements, flags gaps, keeps the audit trail.
what stays human
The professional judgement, the signature, accountability to the regulator.
Not quotes from named clients — the lines that keep coming up, before a project starts and once it's underway.
Before
We know this could be easier. We just don't know where to start.
After
We found the one process worth fixing first.
Before
Our best people spend too much time searching instead of deciding.
After
The searching is gone. That time goes into deciding now.
Before
We didn't want another tool. We wanted our own tools to talk to each other.
After
Nobody sold us AI. They found the one workflow that was slowing us down.
Plenty of engagements stop after step two. The spec is yours either way, and you can take it to somebody cheaper.
Describe one process that costs your team time. We will tell you whether we can help and how long it would take.
Something was missing. Check the fields and send it again.