Give your team their week back.
Every company has work that runs on copy-paste: invoices retyped into a system, support mail sorted by hand, reports rebuilt every Monday. We find the ones worth automating, build them properly, and put them live. Not a demo, not a chatbot bolted on the side. Your team keeps the work that needs judgement, and stops doing the rest.
Sound familiar?
You're here because one of these is true.
Your best people spend their days on copy-paste
Retyping invoices, sorting mail, rebuilding the same report every Monday. It's slow, it's easy to get wrong, and it's not why you hired them.
The AI demo worked, then it didn't
Something clever in ChatGPT never became something your team can rely on. Automations need error handling, logging and someone accountable when they break.
You don't know which process is worth automating
Ten candidates, no obvious winner. We look at volume, cost and how often it goes wrong, then start with the one that pays for itself first.
What we do
The repetitive work eating your team's week, handed to AI: documents, inboxes, reporting, data entry, built as real software and running in production.
AI automation means handing repetitive work to software that reads, decides and acts on its own. We start by finding the process where that actually pays off, then build it like any other production system: connected to your tools, tested against real cases, monitored, and documented. You get something that runs quietly in the background, so your team spends its time where a person actually makes the difference.
The process
How an automation runs.
Light on process, transparent, senior from day one.
- 01
Map
A few days. We follow the process end to end and cost out what it's worth.
- 02
Build
1 – 2 weeks per workflow. Wired into your tools, tested on your real cases.
- 03
Pilot
It runs alongside your team, with a human check, until the numbers hold up.
- 04
Hand over
Monitoring, docs, training. Then it's yours to run.
What you get
What's in. What's out.
Built for
Teams losing days a month to repetitive, rules-based work · Companies sitting on documents, mail or forms someone processes by hand · Founders who tried an AI tool, saw the potential, and want it running for real
Not the right fit if
Teams still deciding where AI fits at all (see AI Masterclass) · Processes that run a handful of times a year
Every tier includes
- Process mapping and a written case for what to automate first
- The automation built as proper software, in your own repository
- Connections to the tools you already use (mail, CRM, ERP, storage)
- Testing against your real historical cases
- Monitoring, logging and alerts when a case needs a human
- Model and provider choices explained, including what they cost to run
- Plain-English docs plus 2 weeks of free support after go-live
Intentionally not included
- Running costs for AI providers and third-party tools (billed to your own accounts)
- Automating processes nobody has agreed on yet
- Replacing your ERP, CRM or core systems
- Long-term maintenance beyond Run & improve (see Monthly Support)
- Data cleanup and migration of years of old records
- Training your staff on tools we didn't build
Plain pricing.
- 2 – 3 weeks
- one process live in production
- 6 – 10 weeks
- 3 – 5 connected workflows
- usage dashboard
- Optional
- monitoring, model upgrades and new steps
Common questions
Everything else.
Does our data get sent to OpenAI or Anthropic?
Only if you want it to. We map what leaves your systems before we build anything, and we can run models in Swiss or EU regions, or on your own infrastructure, when the data calls for it.
What if the AI gets something wrong?
It will, sometimes. That's why every automation has a confidence threshold: below it, the case goes to a human instead of through. We tune that during the pilot, using your real cases.
What does it cost to run once it's live?
Usually tens to a few hundred francs a month in model and infrastructure costs, depending on volume. You get the estimate during the mapping phase, before you commit to the build.
Do we need clean data first?
No. Messy input is normal, and reading messy input is exactly what these models are good at. If something is genuinely unusable, we'll tell you during mapping rather than after the invoice.
Book a scoping call.
No slides, no pitch. Just a quick conversation to see if we're the right fit.
Start a conversationOther ways we help
You may also need.
Development
Websites, web apps, mobile apps, platforms & MVPs, designed and launched as one senior team, built to grow instead of rewrite.
Rent-a-CTO
Part-time senior tech lead, plan what to build, review what you have, hire engineers, unblock the team, and speak to investors like a CTO.
Monthly Support
Steady monthly care for live software, bugs, speed, security, small features. WordPress & Laravel welcome.