canonical: https://jentic.com/apis/geneea.com/geneea

# Geneea Natural Language Processing

Geneea provides a multilingual natural language processing API focused on Czech and other Central European languages alongside English. The 12 endpoints cover spelling and grammar correction, named entity extraction, lemmatization, sentiment analysis, and topic detection through the /s1 namespace, plus account info and service status. It is well-suited for content moderation, search indexing, and analytics over text written in languages where global LLM coverage is uneven. The OpenAPI spec was sourced from APIs.guru and may lag the live service in minor details.

## For AI agents

Run NLP tasks - entity extraction, sentiment, topic detection, lemmatization, and spelling correction - on text in English and Central European languages.

## Scope

Does not handle text generation, translation, or speech - use for entity extraction, sentiment, topic detection, lemmatization, and correction only.

## Capabilities

- Extract named entities (people, organisations, places) from text via `/s1/entities`
- Score document-level sentiment with `/s1/sentiment` for review and feedback analysis
- Detect dominant topics in a document through `/s1/topic` for tagging and routing
- Correct spelling and basic grammar with `/s1/correction` for user-submitted text
- Reduce inflected words to their base form via `/s1/lemmatize` for search indexing
- Read account quota and service status from /account and /status

## Use cases

### Multilingual Sentiment Over Customer Feedback

Score sentiment on customer reviews and survey responses in Czech, Slovak, German, and English. The agent posts each text to `/s1/sentiment` and routes negative scores to a support queue. Captures churn signal in markets where English-only NLP underperforms.

Example prompt: POST `/s1/sentiment` with the review text, read the score and label, and route any negative score below -0.3 to the support queue

### Named Entity Extraction For News Indexing

Extract entities from incoming articles to build per-entity feeds (people, organisations, places). The agent posts to `/s1/entities`, indexes the canonical entity ids, and links articles into topic pages. Supports newsroom search and recommendation systems.

Example prompt: POST `/s1/entities` with the article body, then upsert each returned entity (type, text, lemma) into the search index keyed by article id

### Spelling Correction For User-Submitted Content

Run inbound user text through Geneea's correction service before it hits a search index or downstream model. The agent posts each blob to `/s1/correction`, reads the corrected version, and stores both the original and the cleaned form. Improves search recall and downstream ML quality.

Example prompt: POST `/s1/correction` with the raw text and store the corrected output alongside the original; index only the corrected version

### Agent-Driven NLP Workflows Through Jentic

AI agents discover Geneea via Jentic when they need NLP for Central European text without standing up models locally. Jentic returns the right operation per intent and isolates auth at execution time. Useful for content-moderation and multilingual analytics agents.

Example prompt: Use Jentic search query 'extract entities from text' to load the geneea_extract_entities tool and execute with the document text

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | `/s1/entities` | Extract named entities from text |
| POST | `/s1/sentiment` | Score document sentiment |
| POST | `/s1/topic` | Detect dominant topics in text |
| POST | `/s1/correction` | Correct spelling and grammar |
| POST | `/s1/lemmatize` | Lemmatize words to base forms |
| GET | `/account` | Read account info and quota |
| GET | `/status` | Service status check |

## Key resources

- **Entities** — Named entity extraction from text
- **Sentiment** — Document and sentence-level sentiment scoring
- **Topic** — Dominant topic detection for documents
- **Correction** — Spelling and grammar correction
- **Lemmatize** — Reduce inflected forms to base lemmas
- **Account** — Account info and service status

## Why Jentic

- **Setup:** Wiring Geneea Natural Language Processing by hand means constructing the POST bodies for each analysis endpoint and routing entity, sentiment, topic, correction, and lemmatization calls to the right paths yourself. Through Jentic you install once, import Geneea from the API Directory, store any required credential once, and your agent calls it.
- **Permission scoping:** Geneea's analysis endpoints like `/s1/entities` and `/s1/sentiment` take the text in the request body rather than the URL path, so scope the agent to the operations it needs, such as entity extraction and sentiment. You choose the operations it may call, so functions like topic detection or correction are not included unless you add them.
- **Credential handling:** Any Geneea credential is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'extract entities from text' or 'score sentiment', and Jentic returns the matching Geneea operation with its input schema, distinguishing `/s1/entities` from `/s1/topic` so the agent calls the right endpoint without browsing the reference docs.

## Related APIs

- **OpenAI API** — OpenAI offers general-purpose LLM-based NLP across many languages; Geneea provides purpose-built Central European NLP endpoints with predictable structured output.
- **Cohere API** — Cohere provides multilingual classify, rerank, and embed endpoints; Geneea targets specific tasks like entity extraction and lemmatization.
- **Hugging Face API** — Hugging Face hosts custom NLP models; pair with Geneea when a specific model fills a gap Geneea does not cover.

## FAQ

### What authentication does the Geneea API use?

The OpenAPI bundled in this catalogue does not declare a security scheme; the live service uses an API key as documented at help.geneea.com. Through Jentic any required header or query credential is stored encrypted in the vault and injected at execution time.

### Can I extract named entities with the Geneea API?

Yes. POST `/s1/entities` with the document text and the response includes detected entities with type (person, organisation, location), surface form, and lemma. Suitable for indexing news, support tickets, or research papers.

### Does Geneea support languages other than English?

Yes. Geneea's strength is Central European coverage - Czech, Slovak, Polish, German, plus English. The /s1 endpoints accept a language hint for content where auto-detection would be ambiguous.

### What are the rate limits for the Geneea API?

The OpenAPI spec does not declare rate-limit headers - limits are tied to your Geneea plan and visible at /account. Back off on HTTP 429 and check /account before launching large batch jobs.

### Can I check Geneea's status before a batch run?

Yes. GET /status returns the live service status. Run it as a preflight check before launching any large /s1 batch so you fail fast on planned outages.

### How do I run NLP through Jentic?

Install with pip install jentic, search 'extract entities from text', load the schema for the geneea_extract_entities operation, and execute with your document text. Jentic handles credential injection if required by the live API.

### Can I limit what my agent is allowed to do with the Geneea Natural Language Processing API?

Yes. Because you run Jentic One yourself, your own rules decide which Geneea operations and credentials the agent may use. Geneea's analysis endpoints such as `/s1/entities` and `/s1/sentiment` take the text in the request body rather than the URL, so you can scope the agent to just the operations it needs, for example entity extraction and sentiment scoring. Operations you do not grant, such as `/s1/topic` detection or `/s1/correction`, stay off limits until you add them.