Everything you need to know before working with us.
Private AI means running AI models and agents on infrastructure you control — your own servers, your own network. Your data never goes to third-party cloud services for processing. This gives you data ownership, cost predictability, and no vendor dependency. We typically deploy using Ollama and open-source models like Llama or Mistral running in Docker on your servers.
It depends on the solution. Some agents and automations are designed to run fully autonomously — you just review the outputs. For more technical solutions, basic command-line comfort helps. We always provide documentation and training so you understand what's running and how to use it. If you can use a web interface, you can use our AI systems.
It depends on what you want to run. For lightweight agents and automations, a modest VPS or old desktop can work. For running full language models locally, you'll want a machine with a decent GPU (NVIDIA preferred) or access to a dedicated server. We can help you assess your infrastructure and recommend what you need. Some solutions can also use cloud GPU instances that you control.
Standard agent or workflow projects typically take 2-4 weeks from scoping to deployment. More complex custom development can take 6-12 weeks. The timeline depends on scope, complexity, and how many feedback loops we need. We give you a realistic timeline upfront and keep you updated throughout.
We fix anything that breaks due to our work. If the agent stops working as specified or the automation fails, we debug and fix it. We also answer usage questions and help you optimize. What we don't cover: new features outside scope, infrastructure issues (hosting provider problems), or issues caused by your modifications to the code.
Yes. We offer extended support agreements for ongoing needs — bug fixes, small adjustments, and priority response. Pricing depends on your needs. For many clients, once the system is running well, they don't need much ongoing support. But it's available if you do.
You're in control. Everything we build is containerized (Docker) and documented. Moving to a new server is straightforward — we can help with the migration or hand you the documentation and let you handle it. No lock-in, no proprietary formats.
Cloud AI services (ChatGPT, Claude, etc.) are convenient but come with tradeoffs: your data goes to their servers, per-query costs add up, and you're dependent on their availability and pricing. Private AI keeps your data local, has a fixed infrastructure cost, and runs on your terms. The capability gap is smaller than it used to be — open-source models are surprisingly capable for many tasks.
Usually, yes. We integrate with common tools via APIs, webhooks, and direct connections. If a tool has an API, we can work with it. For tools without APIs, there are often workarounds. During scoping, we identify your integration needs and confirm what we can connect before we commit to a project.
We don't start building until you approve the scope. If you're not satisfied with the scoping document, you can walk away with no obligation. For projects in progress, we handle concerns case-by-case. Our goal is a successful outcome for everyone — we're not going to take money for work that doesn't meet your needs.