A Timerise productRStags: Zero hallucinated tags. Sub-second answers.
Send a piece of text and the tags you use. RStags returns the tags that fit, with a score from 0 to 1 for every tag you asked about, in under a second for a typical list.
Free plan: your first 10,000 tags, no credit card required.
You send
content
“The payment page froze, I was charged twice and I would like my money back.”
tags, your own list
- billing
- bug
- refund
- feature-request
- praise
You get back
- billing0.94
- bug0.88
- refund0.61
- feature-request0.07
- praise0.02
3 of 5 tags come back. Every tag has a score, so you can move the threshold later without a new request.
- 10,000
- Free tags, no card
- $50
- Per 100,000 tags
- 500
- Tags per request
- < 1 s
- Typical response
Why it exists
What goes wrong when a prompt picks the tags?
You can ask a language model to pick tags, and it usually works. The times it doesn't are the times you find out in production, from a report built on a tag nobody defined.
A text model writes until it decides to stop, and then you parse the output and hope it matches your taxonomy. An evaluation model answers a closed question with a probability. RStags is built on that difference, as part of the Timerise AI infrastructure.
How it works
How does a tag request work?
- 1
Send content and your tags
One POST request. The content can be a ticket, a review, an email, or a document, and the tags are the taxonomy you already use.
- 2
Every tag is scored in parallel
The model answers one yes-or-no question per tag, in batches of 50, and gives each answer a calibrated probability.
- 3
Tags above your threshold come back
Ranked best first and capped at five by default. You get the score for every tag, so you can change the threshold later without paying again.
One request with the content and the tags you use.
POST https://rstags.com/api/v1/tag
{ content, tags }What you get
Why can't this tagging API make up a tag?
It can't invent a tag
The model answers yes or no for each tag you supplied. At no point can it produce a new label, so you never get a misspelling, a near-synonym, or a sentence where you expected a tag.
Under a second because nothing is written
Nothing is generated. The model reads your content and returns probabilities, and large tag lists are split into batches of 50 that run in parallel.
Scores, not guesses
Every tag comes back with a probability. You pick the threshold, or you store the scores and re-rank later without calling the API again.
One endpoint
Send content and tags to one endpoint with a bearer token. There's no training set, no fine-tuning, and no prompt to maintain.
POST /api/v1/tagRules in plain English
Pass rules with your tags and each one attaches to the tag it names. “Tag billing whenever a payment method is mentioned” is the whole configuration.
Built on Jev
Jev is an evaluation model from TypeSafe AI. It returns typed decisions instead of text, which is where both the speed and the fixed tag list come from.
typesafe-ai/jevUse cases
What do people use it for?
Support tickets
Tag every ticket and chat the same way, whether it arrives at 2pm or at 2am.
Lead routing
Classify inbound leads by intent and send them on before anyone opens them.
Label backfills
Label a whole table of free text, with the same result on every run.
Guardrails and filters
Get a score your code can branch on for routing, filtering, and guardrails.
Pricing
What does it cost?
Pay for the tags you check. No subscription and no monthly minimum.
Free
10,000 tags
To start, with no credit card
Pay as you go
$50
Per 100,000 tags after that
- One tag decision is one tag checked against one piece of content, so a request with 50 tags counts as 50.
- Credits never expire.
- Up to 500 tags per request.
- If the model is unavailable, nothing is charged, and the response says so.
Security and privacy
What happens to my content?
- Content is sent to the model to be scored and is not stored
- No transcript archive of your requests on our side
- Scoped API keys, revocable from the dashboard
- Requests over HTTPS, authenticated with a bearer token
- Privacy policy published on rstags.com
FAQ
What do people ask?
Yes. RStags is part of the Timerise AI infrastructure, and Timerise Sp. z o.o. owns and operates the service: the API, your account and its keys, and billing.
One tag checked against one piece of content. A request with 50 tags counts as 50 decisions, because every tag is a separate question. Duplicate tags count once.
A tag outside your list cannot come back, because the model only answers yes or no about tags you supplied. It doesn't mean every score is right, which is why you get the score and set the threshold.
Under a second for a typical tag list, because nothing is generated and the batches run in parallel. We will publish a median once we have measured one over real traffic.
No. It is sent to the model to be scored and is not stored, and there is no transcript archive on our side.
The API answers with HTTP 402, the number of decisions the request needed, and your remaining balance. It never scores part of a tag list.
Get started
Tag your first piece of content today
Create an account, copy an API key, and send a request. The first 10,000 tags are free.
The company
Who runs RStags?
RStags is owned and operated by Timerise Sp. z o.o. in Wrocław, Poland. Timerise runs the API, your account and its keys, and billing.
Timerise Sp. z o.o.
ul. Księcia Witolda 49/15, 50-202 Wrocław, Poland
- KRS
- 0000895019
- NIP
- 8971890756
- EU VAT
- PL8971890756
- hello@rstags.com