Annsa reads each piece of feedback four ways: intent — Bug, Feature, Improvement or Praise — the job it relates to, urgency, and sentiment. The four readings are what let one ranked list hold a bug report, a feature request and a compliment without flattening them into each other.
01Four readings of every signal
02The four intents
| Intent | What it captures |
|---|---|
| Bug | Something is broken |
| Feature | An explicit request for something new |
| Improvement | Friction in something that exists |
| Praise | Something worked — worth knowing, nothing to build |
When a message is genuinely ambiguous, Annsa leans to Improvement rather than crying Bug.
03Grouped by meaning, not wording
"Export is slow" and "I can't get my data into a board deck" are the same problem wearing different words — semantic grouping folds them into one theme, so a theme's size reflects the problem, not the phrasing. Separately, each signal carries which job it relates to — deciding what to build, understanding a customer, closing the loop — so the list can also be read by the work it serves.
0416 languages
English, Spanish, French, German, Portuguese, Japanese, Korean, Chinese, Italian, Dutch, Arabic, Hindi, Russian, Turkish, Polish and Swedish. Feedback in Japanese ranks alongside feedback in English on the same list. Very short fragments — under about 10 characters — default to English rather than guessing.
05One message, several points
A message that raises two things becomes two signals — the bug report and the feature request inside one paragraph each land where they belong, instead of the louder one swallowing the quieter one.
06Duplicates don't double-count
The same customer saying the same thing twice is one voice, not two. Deduplication keeps volume honest, so a copy-pasted complaint can't buy a theme extra rank.
07Correct it, and Annsa remembers
Reclassify anything — move a signal, rename a grouping, correct an intent. The correction sticks and feeds Memory, so the same mistake gets rarer, not repeated.
08How does Annsa categorize feedback?
Every signal is read four ways: intent (Bug, Feature, Improvement or Praise), the job it relates to, urgency and sentiment — and semantic grouping folds different wordings of the same problem into one theme. Multi-point messages are split, duplicates are folded, and corrections you make feed Annsa's memory for the next classification.
09What languages does Annsa support?
16 for feedback classification: English, Spanish, French, German, Portuguese, Japanese, Korean, Chinese, Italian, Dutch, Arabic, Hindi, Russian, Turkish, Polish and Swedish. Feedback is classified and ranked on one list regardless of language, so a Japanese bug report weighs the same as an English one.