What does the output contain?

Always an overall sentiment label and a polarity score from -1.0 to 1.0, plus a one-line tone summary. Optionally it adds the top emotions present, each with a 0–1 intensity, and the author's likely intent such as complaint, praise, question, or request — all as one JSON object.

Why JSON instead of prose?

JSON is easy to parse and store, so you can feed batches of reviews or messages through the tool and aggregate the scores. The prompt forces a single fenced JSON object with no commentary so the output drops straight into a spreadsheet or script.

How accurate is sentiment analysis?

Good on clear, longer text and weaker on sarcasm, irony, and very short messages where context is thin. The prompt tells the model not to over-interpret ambiguous input, but treat the scores as a guide, not ground truth, and spot-check edge cases.

Where does my API key go?

It stays in your browser tab and is sent directly to OpenAI or Anthropic with the request you trigger. It is never stored, logged, or routed through any Gera server, and refreshing the tab clears it.

Who pays for the API calls?

You do, on your own provider account. Each analysis is one real API call billed at your usage rate. The tool itself is free — your only cost is the tokens consumed on your key.

Sentiment & Emotion Analyzer (BYO Key)

Knowing whether feedback is positive or negative is only the start — the useful signal is which emotion is driving it and what the author actually wants. This tool runs a multi-dimensional sentiment analysis on any text using your own OpenAI or Anthropic key and returns the result as clean, parseable JSON.

How it works

Pick a provider and model, paste your API key, and choose whether to include an emotion breakdown and author-intent classification on top of the core polarity score. Paste your text and the tool builds a prompt that asks the model to return a single JSON object: an overall sentiment label, a polarity score from -1.0 to 1.0, a one-line tone summary, and optionally per-emotion intensities and the author’s likely intent. It is told to base every value on evidence in the text and return JSON only. The tool makes one direct request to the provider and shows the result to copy.

For Anthropic, the request includes the official direct-browser-access header so it works straight from the page.

Using the output

Because the result is structured JSON, you can run a batch of reviews or support messages through the tool one at a time and aggregate the polarity scores into a trend, or route messages by intent — complaints to one queue, questions to another. The emotion intensities are useful for triage: a review with high “anger” intensity deserves a faster response than a mildly negative one. Keep the schema consistent across runs by leaving the same dimensions selected.

Notes and limits

Sentiment is interpretive, not factual. Sarcasm (“oh great, another outage”), cultural context, and very short messages are the classic failure cases, and the model can read tone into neutral text. The prompt guards against over-interpretation, but you should still spot-check borderline scores and never make an automated decision that affects a person purely on a sentiment label. Use the numbers to prioritise human attention, not to replace it.