Choose the next capability to understand
Start with the knowledge-base walkthrough to learn how to upload documents, verify indexing, and query a reference collection. The other guides cover the neighboring capabilities that turn model access into a reviewable application.
For a new implementation, use this order: access → structured extraction → reference ingestion → retrieval → exact tools → evaluation → interface. Evaluate the examples against the quality criteria for your own application.
Model access, chat & conversations
Connect Python to GenAI Studio, compare models, stream drafts, and manage conversation history.
1,177 words · Code examples and source links
Structured extraction & evidence
Turn résumé text into typed facts, check supporting passages, and preserve human corrections.
1,135 words · Code examples and source links
Create & use a knowledge base
Upload documents, ingest Markdown bundles, verify indexing, and reuse a saved reference collection.
1,518 words · Code examples and source links
Retrieval, embeddings & sources
Inspect retrieved passages, understand vector similarity, and choose built-in or custom retrieval.
1,183 words · Code examples and source links
Tools & database lookups
Expose narrow Python functions for exact crosswalk queries and understand how tool calls execute.
1,240 words · Code examples and source links
Workflows, tracing & evaluation
Coordinate bounded stages, inspect traces, compare repeated runs, and evaluate model-based review.
1,117 words · Code examples and source links
Build a Streamlit review interface
Connect a deliberate submit action to GenAI Studio, handle drafts and errors, and plan persistence.
1,196 words · Code examples and source links
No guides match. Clear the search or try a broader term.
What every guide provides
- A plain-language explanation and a small workflow diagram.
- Source-checked SDK methods, examples, and setup assumptions.
- Copyable code with explanations of inputs, outputs, and failure cases.
- Practical checks for understanding behavior and verifying results.
- Descriptive resource links with visible full URLs for copying into a console.
How to use the examples
Configure credentials privately and choose currently available models through the access guide. Code blocks are starter scripts or explicitly labelled continuations. The HTML pages do not execute Python or call GenAI Studio. The knowledge-base guide documents a readiness-check mismatch found in the reviewed SDK; it does not silently patch the repository.
Shared records, exact mappings, and reviewer edits belong in deliberate database storage. The reference collection supplies context. Preserve that distinction throughout the application.