chifty Private AI
On-premise AI for your company
Your own infrastructure · Supported local AI models · Individually planned
What does chifty On-Premise mean?
chifty runs on your infrastructure. For Private AI, the AI models and all services required for the agreed use case are also configured locally. Depending on the function, this includes embeddings, document processing and OCR. Connected cloud mailboxes, sources and external model endpoints remain external services and need separate review.
chifty is in development. We closely support the first rollouts and agree the available features, prerequisites and schedule for your project.
Define the operating scope
Choose the products, sources and roles for your starting scope. Your IT provides the agreed environment; together we define setup, acceptance criteria and operational responsibilities. Smaller teams can also discuss an individually planned Private AI project with us.
Backups, updates, monitoring and incident handling need an owner. Specify what your team handles and what support is agreed with chifty.
Separate the application from model processing
A locally hosted application can still send requests to an external model endpoint. Review model selection, routing, processing region and retention for each enabled task.
Fully local inference needs supported models and suitable hardware. Include embeddings, document processing and OCR where needed. Test quality and response times with your sources; functions that depend on external services are agreed separately.
Start with a measurable scope
Use a bounded source set and representative tasks. Include permissions, source changes and questions without supporting information in acceptance testing.
Expansion follows the agreed review. Document which connectors, models and workflows were accepted and repeat relevant checks after changes.
How the rollout works
We plan the application, local model processing and rollout with you and your IT team. The scope follows your tasks and infrastructure.
First conversation
You reach out via our contact page. Together we work out whether and how on-premise fits your environment.
Joint planning
We agree sources, users, supported models and operational responsibilities. Hardware, storage, setup and ongoing operation are included in the planning.
Setup
You set chifty up on-site, and we guide you through it and answer any questions along the way.
Acceptance and operation
Review the agreed tasks and assign owners for updates, sources and support.
Private AI for small and medium-sized businesses
Your AI. On your infrastructure.
Search company knowledge, chat with documents and prepare email replies. Together, we plan chifty on your infrastructure – with suitable local models, hardware and implementation.
Knowledge for your team
Knowledge, Chat and Office for company search, documents and emails.
Models on your infrastructure
Supported AI models, embeddings and document processing tailored to your project.
Plan implementation together
We agree on hardware, setup and responsibilities for ongoing operation with you.
Fully local processing requires locally configured AI models, embeddings and the document processing you need. Enabled external models, knowledge sources and email services retain their own data flows.
Funding support for your project with chifty
Funding for your chifty project.
Together, we explore which funding programmes suit your chifty project – for our software, our services and, where applicable, the server hardware needed for the project.
For this project, we help prepare the project description, the required documents and the funding application.
Eligibility, eligible costs and approval depend on the programme and the responsible authority. Our support does not guarantee funding.
Security isn't an add-on
Local AI, tenant isolation, access control, and concrete technical measures: find all the details on our security page.