Ollama

In this catalog it is the Serving option when you need to run model inference at usable latency and cost. Keep Ollama on the observe/propose side of the loop: useful signals and drafts, with human or policy authority still required before anything irreversible runs.

Ollama wraps local model download + serve behind a simple CLI/API for laptop and small-server agent backends.

1. ollama pull fetches a model image.
2. ollama run / API loads it into the local server.
3. Apps send chat/generate requests to localhost.
4. Responses return without a cloud GPU dependency.

Related: llama-cpp · vllm · topics/04-data

• Use when: Ollama packages local model download and serving behind a simple API/CLI for developers matches the bottleneck and you specifically need to run model inference at usable latency and cost.
• Skip when: another pack already covers the need with less ops surface, or Ollama would become a second place people retype truth.
• Prefer instead: compare llama.cpp, vLLM before locking the cell.

• Ops: confirm license, hosting, and release cadence for Ollama (ollama/ollama) before relying on it in a live corridor.
• Integration: bad credentials, scopes, or schema drift produce empty retrievals or runaway tool/model spend without an obvious UI alarm.
• Scope: Serving components support agents; they are not the company ledger or the approval authority.