Text Summariser
Condense long documents to their key sentences using extractive TF-IDF summarisation
/v1/text/summarize
How to Use
1. Send a POST request to `/v1/text/summarize` with the `text` to summarise. 2. Set `num_sentences` to control the summary length (default: 3). 3. The response includes the `summary` string and the `original_sentence_count` for context. 4. For best results, provide at least 5 sentences of source text.
About This Tool
The Summarise endpoint uses extractive summarisation to condense long documents to a configurable number of key sentences. It uses TF-IDF scoring to rank sentences by importance — no LLM or external service required.
Unlike abstractive summarisation (which generates new text), extractive summarisation selects the most relevant sentences directly from the source — making it fast, deterministic, and free from hallucination.
Use Cases
Article previews
Generate a 2–3 sentence preview for email newsletters or social media
Document digest
Condense reports and contracts to their key points for quick review
LLM pre-processing
Reduce token count by extracting relevant sentences before sending to an LLM
News aggregation
Auto-generate article summaries for a content aggregation feed