02 — Capability · BC-770.30
Generative AI & Agents
Build assistants, content generators and agent workflows on foundation models, grounded in enterprise knowledge, evaluated for accuracy and fenced by guardrails appropriate to what they are allowed to do.
- GenAI
- AI Assistants
- Agentic Workflows
In scope
- Prompt and context design
- Retrieval and grounding on enterprise content
- Agent orchestration and tool access
- Output evaluation and guardrails
Out of scope
- Knowledge article authoring itself (see BC-740)
- Customer-facing conversational service design (see BC-450)
Realized by · 2
- GitHub · GitHub Copilotstrong
- Snowflake · Snowflake Cortexemerging
Used in · 0
- Not yet placed on a value stream.
Build it · 0
- Nothing in the library points here yet.
Decomposes into · 4
- BC-770.30.10Prompt & Context DesignCraft the instructions, examples and context a model receives so it produces consistent, on-task output, and version them like code.
- BC-770.30.20Retrieval & GroundingConnect models to governed enterprise content so answers cite what the organization actually knows rather than what the model guesses.
- BC-770.30.30Agent OrchestrationCompose multi-step workflows in which models call tools, hand off to each other and ask a human when the stakes require it.
- BC-770.30.40Output Evaluation & GuardrailsTest generated output for accuracy, safety and policy fit before and after release, and block or flag what fails.