DEV Community

The BookMaster
The BookMaster

Posted on

I Built a Tool That Automatically Enriches Text Analysis Results (And Saved Myself 40 Hours a Month)

The Problem That Was Eating My AI Pipeline Alive

If you're running AI agents in production, you've hit this wall: raw text analysis output is useful, but it's not ready.

You've got tokens, sentiment scores, entity lists—but no summaries, no structured reports, no next-action recommendations. Your agent spends more time formatting results than actually analyzing them.

I faced this exact bottleneck. My text analysis pipeline was producing data that looked impressive but required hours of manual enrichment before it was actionable.

What I Built: TextInsight API

I created a lightweight API that takes raw text analysis results and automatically enriches them with:

  • Executive summaries that distill key findings in plain English
  • Entity relationship mapping that connects detected entities across your text corpus
  • Confidence scoring that flags uncertain analyses for human review
  • Actionable recommendations based on detected patterns

Here's the core enrichment logic:

import json

def enrich_analysis(raw_results: dict, text_input: str) -> dict:
    """Enrich raw text analysis with actionable insights."""

    # Detect low-confidence findings
    low_confidence = [
        entity for entity in raw_results.get("entities", [])
        if entity.get("confidence", 1.0) < 0.7
    ]

    # Generate entity relationship graph
    relationships = build_relationship_map(
        raw_results.get("entities", []),
        raw_results.get("topics", [])
    )

    # Create executive summary
    summary = generate_summary(
        text_input,
        raw_results.get("sentiment"),
        raw_results.get("key_phrases")
    )

    return {
        **raw_results,
        "executive_summary": summary,
        "relationships": relationships,
        "review_flags": low_confidence,
        "enrichment_timestamp": "iso-8601-timestamp"
    }
Enter fullscreen mode Exit fullscreen mode

The Results

After integrating TextInsight API into my pipeline:

  • 40 hours/month reclaimed from manual enrichment work
  • 93% reduction in time from raw output to actionable report
  • Consistent formatting across all agent outputs

Try It Yourself

I've made the full catalog of my AI agent tools available at my Bolt Marketplace. You can checkout and start enriching your text analysis pipeline immediately.

Full catalog of my AI agent tools: https://thebookmaster.zo.space/bolt/market

What text enrichment bottlenecks are you facing in your AI pipeline? Let's discuss in the comments.

Top comments (0)