Developer Scenarios

SAK serves multiple kinds of agent development teams. Each team has different concerns, but they all face combinations of identity, memory, and action.

Data and algorithm teams

These teams want agents to be smarter, more contextual, and less hallucination-prone.

Common work includes prompt engineering, RAG data cleaning, knowledge base construction, and long-term memory management.

SAK helps through:

• Web Agent for clean Markdown data without maintaining crawlers or headless browsers.
• GUM for user profiles and conversation history across sessions.
• GenAuth for explicit authorization boundaries when agents access protected data sources.

Business development teams

These teams need agents to perform real tasks and connect to existing workflows.

SAK helps through:

• GenAuth for user login, agent authorization, and OAuth delegation complexity.
• Web Agent for ready-to-integrate search, extraction, and browser action APIs.
• GUM for product experiences that continue across sessions rather than starting from empty context.

Infrastructure teams

Infrastructure teams care about permission boundaries, data provenance, execution environments, and compliance review.

SAK helps through:

• GenAuth for audit logs that show who authorized which agent and which resources it accessed.
• Web Agent for sandboxed execution and source records around open web operations.
• GUM for user context with provenance, lifecycle, and erasure boundaries.

Product teams

Product teams care about whether agents can deliver stable user experiences: remembering users, accessing live data, and explaining important actions.

SAK turns those experience requirements into shared infrastructure rather than repeated per-product engineering work.