Services · AI agents
AI agents:
employees that never sleep.
An AI agent is software that thinks and acts on its own: it researches, answers emails, maintains data and makes simple decisions — around the clock. We build custom agents and multi-agent systems for Swiss companies.
Unlike a chatbot, an agent doesn't wait for questions. It gets a goal, plans the steps itself and uses your systems — email, CRM, calendar, databases — like a trained employee. With clear boundaries and escalation to humans where it matters.
What's included
What you get.
Custom agents
Agents that take over exactly your task — from lead qualification to research, built on your workflows.
LLM integration
The right language models, cleanly connected to your systems — Claude, GPT or local, depending on requirements.
Multi-agent systems
An orchestrator distributes work to specialised agents — research, data and communication run in parallel.
Guardrails & control
Clear boundaries, logs and escalations: the agent acts autonomously, you keep oversight.
Use cases
Where agents
already work today.
Lead qualification
The agent researches incoming enquiries, enriches them with company data and prioritises them against your ideal profile — before you pick up the phone.
Inbox triage
Read, sort, answer the standard cases, escalate the exceptions: your inbox does the prep work instead of piling up.
Research & market watch
Competitors, prices, tenders: the agent monitors continuously and reports only what matters to you.
CRM upkeep
Updating contacts, merging duplicates, adding call notes — your data stays current without appeals to discipline.
Support relief
The agent answers recurring questions from your knowledge base and hands complex cases to your team — context included.
Reports & summaries
The weekly report writes itself from numbers, emails and notes. The agent writes, you read.
Our approach
Autonomous doesn't mean
out of control.
Every agent starts with a narrowly defined mandate and few permissions. Only once it proves itself does it get more responsibility — like a new employee on probation.
You can always see what your agents are doing: every action logged, every decision traceable. Sensitive cases go to a human automatically — you define where that line is.
After a free initial consultation you get a fixed quote — operations and monitoring included.
References Built by Jubati: tap-print.ch. Every project with hosting and care from one source.
In depth
How an agent
actually works.
At the core is a language model — Claude, GPT or a locally hosted model — connected to your systems: email, CRM, calendar, databases. The agent gets a goal, plans its own steps and uses those tools until the task is done.
The real work isn't the model but the integration: which systems may the agent use, which actions may it take, when must it ask a human? We design these guardrails with you — and enforce them technically, not just on paper.
For bigger processes we orchestrate several specialised agents: one researches, one maintains data, one communicates — coordinated by an orchestrator, in parallel instead of one after another. Knowledge work scales without growing your team.
How we work
From idea to live —
in days, not months.
Briefing
A 5-minute questionnaire. We understand your goal — and you only explain it once.
Concept & build
Our team ships version 1. Web or AI — fast and without compromise.
Feedback & iteration
You review, test and comment. We refine it until it is right.
Launch & beyond
Go live. Ongoing support, updates and further development included.
Common questions
Quick answers.
What's the difference between a chatbot and an AI agent?
A chatbot answers when asked. An agent works towards a goal: it plans steps, uses tools and systems and completes tasks on its own — even when nobody is watching.
What do companies use AI agents for?
Typical uses: researching and qualifying leads, triaging and answering email inboxes, keeping CRM data current, market and competitor monitoring, producing reports. Anywhere knowledge work becomes repetitive.
How do we stay in control?
Agents get clear guardrails: defined permissions, logged actions, automatic escalation to humans when uncertain. You decide what an agent may do — and see what it does.
What does an AI agent cost?
Projects start from CHF 499; the exact price depends on scope. After the free initial consultation you get a fixed quote — operations and monitoring included.
Which language model do you use?
It depends on the task: Claude, GPT, or a locally hosted model when data must not leave the building. The model is interchangeable — the value is in the clean integration with your systems.
Do the AI providers see our data?
The agent sends the model only what the task requires. The major providers do not use API data for training by default; for particularly sensitive data we use local models. All nDSG-compliant and clarified transparently up front.
Does an agent replace employees?
In practice it takes over the work nobody ever has time for: research, data upkeep, routine communication. Your team gains hours for what needs judgement and relationships — selling, advising, deciding.
More services
What else we build.
Web development
Websites, shops and web apps: lightning fast, SEO-optimized and low-maintenance. Not templates — something that is yours.
/02AI automation
Repetitive processes become workflows that run themselves. More output, less manual work.
/04Tools & dashboards
Custom internal tools, AI dashboards and analytics platforms that bring your data to life.
Ready when you are
Let's build yours.
Website, AI automation or agent development — tell us where you want to go, and we make it happen.