How it works
From one question to one live system
Every engagement starts with a clear, written scope. Each step ends with something you can decide on.
How the work goes
Two weeks of proof. Four to six weeks in production. Measured every month, and each round feeds the next.
What every engagement has in common
Written scope
Scope is fixed in writing before any code.
Real tests before go-live
Real test examples that the system has to pass, and an automatic gate (CI gate) that blocks changes that make the system worse.
Your environment
AWS, Azure, GCP, GleSYS or on-premise, with audit logs, and no training on your data without written consent.
You own the result
You own the code, the prompts and the data.
An operating guide and a written yes or no
An operating guide (runbook) for running the system, and a written yes or no before you scale.
Who delivers
Telchyn is built by an engineering team in Gothenburg and Malmö.
The team started in 2024.
The founders still work on client projects themselves.
- Founded
- 2024
- people
- 10
- Office, also working from Malmö
- Gothenburg
We have been part of teams at Volvo Group Connected Solutions, Polestar, Porsche, SKF and Coop, among others.
Telchyn team
What clients say about our consulting work
References from earlier consulting work. Not Telchyn AI deliveries.
Johan and Elias worked with our team for over a year. Their technical depth, mentorship, and ability to communicate clearly across business and engineering made a real difference.
Maria Liess, Manager IT Portfolio · Smart
The market
Half of mid-sized Swedish firms did not use AI in 2025
50%
of Swedish firms with 50-249 employees used at least one AI technology in 2025 (margin of error ±4 points).
Note: Our calculation: 100% − 50% = 50% of these firms used no AI technology.
49.6% vs 30.4%
Sweden against the EU27 average: enterprises with 50-249 employees using at least one AI technology, 2025.
9%
of firms with 50-249 employees give lack of relevant expertise as a reason for not using AI. It is the largest reason in every size class.
Note: Share of all firms in the size class, not only firms that did not adopt AI.
52%
of medium-sized firms see AI as a very important growth measure.
Common questions
AI in your process, in short
How much time does it take us?
The call is 30 minutes. You describe one process and where its data lives.
Do you train on our data?
No. We do not train on customer data without written consent.
What do we own?
You own the code, the prompts and the data.
Where does the system run?
In your environment: AWS, Azure, GCP, GleSYS or on-premise, with audit logs.
What if the answer is no?
You still get a written report: what works, what does not, and the gaps in the data.
All offers
Next step
Choose the step that matches your question
Not sure which one? Tell us the process and where its data must stay.