Software & Hardware Engineering · Zurich · since 2007

Built with AI.
Owned by engineers.

Electronics, embedded software and applications from a single source. Much faster and more affordable with AI – scalable, maintainable and compliant thanks to engineering.

How we bill

The honest answer to the AI question

Yes, AI writes code, designs circuits and builds prototypes in days instead of months. But a prototype is not a product.

What AI delivers today

Code, drafts and variants in minutes
Tests and documentation generated automatically
Research, analysis and data evaluation
Fast prototypes to validate ideas

Engineer

What still takes engineers

Scalable – an architecture that grows with your success
Maintainable – for years, not just until the demo
Manageable – operations, updates and support in the field
Compliant – FADP, GDPR, CE, EMC and accountability

That is inaffect: we build with AI – and remain the engineers who take responsibility for your product.

From idea to operations

One team. Board to app.

Understand

We clarify goals, value and constraints – in plain language and as equals. Afterwards everyone knows what will be built and why.

Build with AI

AI accelerates code, layout variants and tests. Our engineers review, decide and take responsibility.

Hardware

Schematics, PCB layout and prototypes from our in-house SMD assembly – in days instead of weeks.

Validate

Testing, EMC and certification requirements are part of the plan from day one. That saves costly loops at the end.

Operate

Support, monitoring and further development. So your product runs at your customer’s site – not just in the demo.

New: billing in tokens

One currency.
For AI and engineering.

Instead of opaque timesheets we bill in tokens – traceable across every discipline. There are AI tokens and engineer tokens, and for both we distinguish between input and output, just like modern AI models.

Understand the token model

Input tokenlow cost

Everything that flows into the project: requirements, data, alignment, feedback, tests and compute. You need a lot of it – and it decides whether the project succeeds.

Output tokenknow-how

What comes out as the finished solution: architecture, reviewed code, layout, documentation and acceptance. More valuable, because it bundles experience and accountability.

Expertise

Specialists, not generalists.

Hardware

Schematics, PCB layout, simulation, SMD prototyping, testing, series production, IoT, FPGA

Embedded & Software

Firmware, RESTful services, microservices, PWA, desktop and smartphone apps

AI solutions

LLM integration, agentic workflows, edge AI, predictive analytics, GDPR-compliant AI architectures

Architecture

Embedded systems, SoC, bus systems, cloud, data models, application design

Business Intelligence

Data warehouse, ERP, CRM, IAM, management platforms, Power BI

Projects & operations

Project management, risk management, certification, support, monitoring, further development

Founder & CEO

“The hard part of an AI project is not the model.”

A prototype shows that something can work. Whether it lasts is decided by operations, maintenance and liability. Thomas Joss has been building technology companies since 1997, has led inaffect since its founding in 2007 and teaches “Agentic Harness” at universities of applied sciences in Zurich.

His stance shapes how we work: use AI where it creates a real edge – and say so honestly when a problem needs a clean interface rather than a model.

Thomas JossFounder & CEO, inaffect AG

thomasjoss.net ↗

Clients