Build portable agents end-to-end
The Generic target produces a platform-neutral system prompt that deploys to any LLM platform — the fastest path to a working agent when the target isn’t decided yet or the task is simple.
The platform
Generic is the default when there’s no reason to commit to one vendor. The artifact is a well-crafted system prompt that runs on any compatible model endpoint, with a few common deployment shapes:
- System prompt only — the bare prompt, pasted into any model.
- Simple RAG — the prompt plus retrieval context injected at runtime.
- ChatGPT Custom GPT or a Gemini Gem — the same prompt wrapped as a vendor agent.
It is deliberately thin: no platform schema, no managed state, no tool contracts. Output quality lives entirely in the prompt.
The build
AESOP runs the generic pipeline and emits a single portable prompt. It is the right call when the agent is simple — Q&A, summarization, classification, text generation — or when the platform decision is still open and the team wants a prototype before committing.
It is the wrong call when the workflow needs tool use, multi-step reasoning, or access to an enterprise knowledge base; those belong on a platform-specific target instead.
Deploy & evaluate
Deployment means pasting the prompt into the chosen model, GPT, or Gem. The deployed agent is then run through AESOP’s Evaluate stage and scored for quality against the PRD.