AI agents for your software
Your people chat with your own data and create reports by prompt, and AI assistants like Claude, ChatGPT or Copilot use the functions of your application. We build it into the software you already have.
Discuss your AI projectAgents that work inside your software
An AI agent understands a request in plain language, calls the right functions of your systems and returns a result. We build these agents straight into your existing applications.
Chat and reports from your data
Your people ask in plain language, for example "revenue by region from January to March", and get an answer or a finished report from your systems.
Your application as an MCP server
An MCP server makes the functions of your software usable for AI assistants like Claude, ChatGPT or Copilot, through an interface you control.
Workflows without friction
Agents connect the steps of a process with tool calling, from capturing data to the finished report. Fewer hand-offs and fewer clicks, and where it matters, a person approves.
You stay in control of your data
In your cloud, on your terms
Usually you run the model yourself, for example in Microsoft Foundry on Azure. We know the levers: choosing region and deployment type so data is processed in Switzerland or the EU, where Microsoft does not use it to train models.
Local models as an option
For experiments or particularly sensitive data, smaller models can also run locally, for example on a developer's machine or on the device, like the AI in Taperwise.
Rules decide, AI explains
Business-critical decisions are made by traceable rules in your code. The AI understands, summarizes and explains. That keeps every result verifiable.
People decide. Agents support.
With us you work directly with the people who build your software. We listen, understand your business and talk through every important decision with you. AI agents are a tool that supports us where it makes sense. Put simply, an agent delivers the average, but your problem is rarely average.
What people decide with us
- Personal exchange: we listen and understand your business
- Architecture, patterns, performance and security that fit your problem
- Responsibility for every line of code that goes live
Where agents support us
- Getting to clickable prototypes faster, which we then discuss with you
- Routine work in large migrations, such as .NET Framework to modern .NET
- An extra pair of eyes in tests and reviews
What this looks like in practice
Client project: reports and workflows by prompt
The goal was more than a chat with the data: an end-to-end process with less friction. Employees request, for example, the data for a given date range by prompt and get a finished report straight away.
Client project: MCP server for an existing application
We opened the functions of an existing application to AI assistants through an MCP server, without rebuilding the application.
Taperwise: our own app, AI on the device
Our running app for iPhone and Apple Watch. The AI runs on the device: no account, no cloud. It turns a goal in plain language into a filled-in plan form, while rules calculate the plan itself. We built it with AI agents as a tool.
See the Taperwise running appThis website speaks MCP
Connect Claude, ChatGPT or another AI assistant to our MCP server. Your agent can then look up our services, trainings and articles directly.
https://bitspire.ch/mcpFrequently asked questions about AI agents and MCP
What is an MCP server and does my company need one?
An MCP server (Model Context Protocol) is a standard interface through which AI assistants such as Claude, ChatGPT or Copilot use the functions of your software. It makes sense when your employees or customers should get tasks done in your application by chat or voice. You decide exactly which functions are exposed.
Can AI agents be integrated into our existing .NET software?
Yes. We add agent capabilities to existing .NET applications, for example a chat over your data or an MCP server. A rewrite is not needed.
Does our data stay in Switzerland?
That depends on your infrastructure. In client projects the client usually runs the model, for example in Microsoft Foundry on Azure. We help choose region and deployment type so data is processed in Switzerland or the EU.
Does this work with local models, without the cloud?
For some use cases, yes. Smaller models now also run locally, for example on a developer's machine or directly on the device, like the AI in our app Taperwise. Whether a local model is enough for your case is something we find out together in a prototype.
If AI agents can write code, why do we need you?
An agent writes code, but put simply it delivers the average of what it has learned. Your problem is rarely average. Which architecture and patterns fit, how the solution stays fast and secure and which requirement really matters are human decisions that take experience and a real conversation with you. Agents support us where it makes sense, and your contact is always a person.
How do we get started with AI agents?
The easiest way is a conversation: together we pick a use case with clear value and build a prototype for it. Then you decide whether and how to continue. We do not offer AI trainings or workshops; we build the solution together with you.
More .NET services
Development
ASP.NET Core, Blazor, Angular and React. From the first line to the application in production.
More on ASP.NET Core developmentArchitecture and technical leadership
Architecture reviews, decisions that hold, and a sparring partner for your team.
More on .NET architectureUX/UI and accessibility
Interfaces people understand, and applications everyone can operate.
More on UX/UI and accessibilityRather write to us directly?
Let's Build Something Great Together
Have a question or want to discuss a project? We'd love to hear from you. Fill out the form below and we'll get back to you as soon as possible.
You can also reach us directly at [email protected]