AIEnrichMiddleware¶

Use this when: an LLM should add, rewrite, or infer fields on each record.

AIEnrichMiddleware calls a provider once per record, parses a JSON response, and merges the returned fields back into the record.

What it does¶

• renders a prompt from the current record
• calls the configured AI provider
• parses the response as JSON
• merges selected fields back into the record

When it is a good fit¶

• summaries, tags, categories, or descriptions
• filling in missing structured fields from unstructured context
• record-by-record enrichment where the provider decision depends on the whole record

How it behaves¶

• works with dicts, dataclasses, and Pydantic models
• output_fields can whitelist only some response keys
• caching is optional but recommended for repeated runs
• on_error controls whether failures pass through, drop, or raise

Example¶

from agora.ai import SQLiteLLMCache
from agora.middlewares.ai.enrich import AIEnrichMiddleware


pipeline.pipe(
    AIEnrichMiddleware(
        provider=my_provider,
        prompt_template=(
            "Summarize this product: {name}. "
            'Return JSON: {"summary": "...", "price_level": "..."}'
        ),
        output_fields=["summary", "price_level"],
        cache=SQLiteLLMCache(".cache/llm.db"),
    )
)

When to choose something else¶

• use AIExtractMiddleware when the main job is extracting structured fields from one text field
• use AIClassifyMiddleware when the output is one label from a fixed category set
• use AIBatchMiddleware when provider cost dominates and records can be processed in grouped prompts