Common questions

Do I need my own servers?

No. For many tasks, established AI services with a proper data processing agreement are enough. If data must not leave your premises, a model can also run locally on your own hardware. What makes sense depends on data, volume and budget.

What happens when the AI gets something wrong?

That is why we measure first. An assistant should decline when information is missing instead of guessing, and decisions with consequences stay with your team. Critical steps get an approval step.

What does an AI assistant cost?

It depends on data volume, integrations and operation. After an initial review, you receive a quote with a defined scope. Development and ongoing model costs are listed separately.

Which models and tools do you work with?

Models from Anthropic (Claude) and OpenAI, as well as open models running locally via Ollama or llama.cpp. Connections to existing software are built with the Model Context Protocol (MCP) and custom interfaces. The right mix depends on the task, data protection and cost.