Understand health
Inspect infrastructure, service readiness, configuration, and model access in the admin interface.
The Srasta platform
From private models to useful work.
Connect your applications to private AI, bring in selected company knowledge, and manage the system throughout its life.
Company knowledge
Give AI the context your team needs.
Upload text-based PDF, DOCX, RTF, and supported text files, or ingest a Git repository. Configure source roles and connect a compatible client to the retrieval API.
Retrieval supplies document context to the model. Review source references and evaluate answer quality against questions your team can verify.
Start with a bounded evaluation
- Use a representative document set
- Define who may access each source
- Compare answers with known examples
- Check behavior when relevant information is missing
Scanned documents requiring OCR and automatic synchronization of workplace app permissions are not implied by document upload support.
Read the ingestion guide →Coding and applications
Connect private models to the work you do.
Use a suitable model through a compatible coding client to evaluate code explanation, focused changes, and test suggestions. Your client supplies the editing experience and controls what code or context is sent.
Business applications can also use model endpoints for tasks such as summarization, extraction, classification, or drafting. Integration and output validation belong to the application workflow.
Plan a coding evaluation →Confirm the complete setup
- Model, runtime, and hardware support
- Client authentication and API compatibility
- Context handling and external connections
- Human review and relevant code tests
A model endpoint does not provide an IDE, repository awareness, or autonomous tool execution by itself. The planned agent harness is a separate roadmap direction.
Models and access
Connect through a controlled entry point.
Use OpenAI-compatible endpoints with models selected for supported hardware. Manage identity, roles, and model permissions at the gateway, and review recorded activity.
The admin interface includes diagnostic chat for checking the inference path. Your employee-facing experience depends on the compatible client you choose.
A clear deployment boundary
Run private inference and retrieval on your infrastructure. Choose external integrations deliberately and review their data flows.
Explore access and security →Day-to-day operations
Keep control after installation.
Review activity
Inspect audit activity and export evidence for your own review processes.
Manage change
Use upgrade, backup, verification, and recovery workflows with documented operating procedures.
See how it fits your workflow.
Discuss knowledge, coding, or an application workflow alongside access controls and operations. Bring your questions about infrastructure and evaluation. For future memory and agent capabilities, explore the planned product directions.