AI assistant
It answers when you ask. You give it keywords or a text, it delivers a draft, and you decide what happens with it.
Example: our Mac app Aivy writes replies to emails and saves them as drafts in Apple Mail.
AI agents
A colleague who takes over the routine.
An AI agent handles tasks that consist of several steps: it reads, looks things up, decides within the agreed framework and prepares the result. You check and approve. Your team gains time for the work that needs people.
Assistant, agent or agent system?
The terms are often mixed up. We distinguish three levels.
AI agent
It pursues a goal over several steps and uses tools such as email, CRM or calendar to do so.
Example: you copy our email agent Qonnect into a message, it checks the calendar, suggests times and books the meeting including a video link.
Agent system
Several agents with different roles work together, for example one plans, one checks, one executes. This pays off for extensive workflows with many interfaces.
Example: AEGIS, our agent system, runs entirely on premises and keeps planning, checking and execution strictly separate.
What AI agents are good for in SMEs
- Enquiries and inboxSort emails by request, flag urgent ones, prepare replies.
- Quotes and ordersCombine data from the enquiry, price list and customer history into a draft quote.
- Documents and dataRead invoices, delivery notes or forms and transfer them into existing software.
- Research and reportsCollect information from several sources and turn it into a summary or monthly report.
- Appointments and follow-upTrack responses, remind about open quotes, suggest appointments.
You stay in control
An agent only acts within the rules we define together. Wherever money, customers or contracts are involved, a person checks before anything goes out. Every step is logged, so it stays clear what the agent did and why.
You decide which data it may see. For sensitive data the agent runs with a model on your own server, see Local AI.
How we proceed
Initial conversation
Free and without obligation. We clarify the task, the goal and the framework.
Map the workflow
We record how a task runs today and define where the agent takes over and where you approve.
Pilot
The agent first runs in a clearly defined area. After two to six weeks you see real results.
Roll out and support
Training for the team, adjustments from daily use, support and further development.
Costs
We set the exact scope in a clear quote after the initial conversation. Ongoing costs for models and operation are shown separately.
AI assistant
from €1,000
Drafts and answers on request.
AI agent
from €2,000
One multi-step task with tools and approval.
Agent system
from €5,000
Several agents with interfaces, open-ended depending on scope.
Matching products
Agents we have developed ourselves and see in daily use.
FAQ on AI agents
What is the difference between a chatbot and an AI assistant?
A chatbot usually sits on a website and answers general questions from visitors, often following fixed rules. An AI assistant works for your team: with your templates, data and workflows, and delivers drafts that you check and use. The next step is the AI agent, which completes a task on its own.
Which software can an agent work with?
With anything that has an interface or can be controlled reliably: email, calendar, CRM, accounting, inventory management, spreadsheets and document storage. Whether your software is among them is something we check in the initial conversation.
How long does it take until the first agent?
A clearly defined agent is usually in pilot operation after two to six weeks. What matters most are the interfaces and how well today's workflow is documented.
What happens if the agent makes a mistake?
That is what approval points are for. For tasks with external impact the agent suggests and a person decides. The log shows where something went wrong, so we can tighten the rules.
How does this fit with the AI Act?
Most agents in everyday office work do not fall into a high-risk category. Anyone using them must, however, ensure sufficient AI literacy in their team. Our workshops help with that (in German).
Further reading
From the blog (in German)
Terms explained
- Agentic AI AI that plans and acts on its own
- AI agent the term at a glance
- Human in the loop people as the checkpoint
More AI services
AI solutions at a glance
Consulting, assistants, agents, company knowledge, infrastructure and training at a glance.
Learn more →Company knowledge with AI
Answers from your own documents, with sources. Knowledge assistant from €1,000.
Learn more →Local AI
Openly available models on your own server or in an EU data centre. Setup from €1,500.
Learn more →