How it works
Two modes. One memory.
The tech behind FluentC's batch and real-time translation, and how agents and workflows plug right in.
Pick a mode
Real-time for the moment. Batch for the library.
For agents, chat and support
Real-time
Fast, live responses, inside the request that needs them.
- Instant response with translation memory in front of the model
- No polling: one call, one answer
- Token-based usage
For libraries and catalogs
Batch
Big jobs in the background, delivered when they're done.
- Unlimited usage, built for content libraries, CMS and UI copy, and product catalogs
- Processed in the background with the same smart memory cache
- Delivered by polling or webhook
A request, step by step
What happens after you hit send.
Real-time
- Send text, HTML or JSON with your API key.
- FluentC hashes the content (SHA-256) and checks translation memory.
- On a hit, the stored translation comes back in 7 to 11 ms with no model call.
- On a miss, it translates, stores the result and returns it.
Batch
- Submit up to 50,000 rows. A job_id comes back right away.
- Thirty-two workers translate in parallel, memory first.
- JSON is walked and reassembled in its original shape.
- Poll /ai_agent/results. The response tells you when to ask again.
Translation memory
Why the second pass is cheaper than the first.
Exact, not fuzzy
Memory is keyed by a hash of the content, so a match is a match. You never get a near-miss translation standing in for yours.
The second pass is cheaper
Repeated strings are served from memory, so reruns of a catalog or a UI are faster and don't pay for the same words twice.
Included, not upsold
Translation memory is part of both plans. There's no add-on tier for it.
Which mode?
Match the mode to the job.
| The job | Mode |
|---|---|
| A chatbot reply, in the user's language | Real-time |
| A support ticket as it arrives | Real-time |
| An agent's translate tool call | Real-time |
| A product catalog, nightly | Batch |
| UI strings and CMS content | Batch |
| A newsletter, before it sends | Batch |
Start translating with FluentC.
Real-time speed and batch scale.