What can I build with Anvia?
You can build TypeScript application features that need one or more model calls, tools, memory, retrieval, structured data, or streamed interaction. Anvia is useful for both user-facing experiences and server-side workflows.
Common examples include:
- Support or operations assistants that call application tools.
- Search and question-answering over private documents.
- Structured extraction and classification pipelines.
- Chat, completion, and multimodal interfaces.
- Background workflows that combine deterministic steps with agents.
- Evaluation and observability workflows around model behavior.
- Multi-agent systems where the added coordination is justified.
You do not need an agent for every feature. A one-call summarizer may only need direct completions, while a schema-validated extraction flow may use structured output. Use agents when the behavior needs reusable instructions, tools, multiple turns, memory, or run-level controls. Use pipelines when the workflow has explicit stages and typed transitions.
Anvia does not supply your application data or product decisions. A customer-support agent still needs application-owned tools with authorization checks. A knowledge assistant still needs an ingestion strategy, storage, metadata policy, and source permissions. A production chat still needs an authenticated server route and a user interface.
For local inspection, add Studio. For production traces and evaluation workflows, consider Lens. Browse the package catalog to see the available provider and infrastructure integrations.