[01] Core engine · Tagma-1
Cards cut the way a debater cuts them.
Search, Mine, and Counter find evidence in different ways. Tagma-1 is what turns it into a card: the tag, the highlighting, and the warrant, done the same careful way every time.
[02] What it does
One pass from raw text to a card you can read.
Reads the whole passage
Tagma-1 reads the full stretch of source text behind a card before deciding anything, so the cut follows the author's reasoning instead of the first sentence that sounds right.
Writes the tag
Tags name the argument, not the author. One clear sentence that says what the card proves, written in the tag style and length you picked.
Picks what you read aloud
Short fragments that carry the claim and the warrant, not whole highlighted paragraphs. The card reads as one clean line of argument when spoken.
Checks every word
Before the card is returned, every highlighted word is matched back to the original source. Anything that does not match is fixed or removed.
Every card in the product goes through this same step on our servers, whether it came from Search, Mine, Counter, or your AI assistant. There is no lighter version anywhere, so a card cut at 2am from your phone reads the same as one cut at practice.
[03] The guarantee
It points at the author’s words. It never writes them.
Most AI writing tools will happily paraphrase a source and hand it back looking like a quote. In debate, that is a fabricated card. Tagma-1 is built so that cannot happen: it can only choose words that are already on the page.
Every highlight is verified against the source text. If a fragment comes back even slightly reworded, it is snapped to the author’s real wording or dropped. If there is nothing worth reading in a passage, the card is dropped instead of padded. What you read in round is what the author wrote.
[04] Your style
Show it your cards. It cuts like you.
Every team cuts differently. Some read long, complete sentences. Some read three words at a time. Tags range from a short label to a full argument. A generic highlighter gets all of that wrong.
We learned that examples shape a cut far more than written rules do. So instead of a settings page, Tagma-1 learns from your own cut cards. How much you read, how short your fragments run, and how your tags sound all come from the examples you give it.
Upload a docx of your cards
Any file your team has already cut in Word or Verbatim. Tagma-1 reads what you actually read aloud, not just what you underlined.
Save it as a cut style
Your style shows up next to the built-in ones in the cut style picker. It belongs to your account.
Use it everywhere
Pick it for any Search, Mine, or Counter run, or from your AI assistant, and every card follows it.
[05] Density
Base or tight, per run.
Base
A complete card that is no longer than it needs to be. The argument and its warrant read cleanly out loud, with nothing extra.
Tight
The fewest words that still carry the argument. Built for speed-heavy rounds where every second counts more than polish.
[06] How we test it
Measured against real debaters, not vibes.
Real cards as the answer key
Our benchmark is built from thousands of cards cut by competitors across seven divisions, not from what we think a card should look like.
Graded blind
When we change how Tagma-1 cuts, the old and new versions are graded side by side without knowing which is which.
Checked for consistency
The same passage should come back cut the same way. We track how stable the cut is across repeated runs, not just how good the best run looks.
Read, not just scored
Numbers can look right while cards read badly. Every change is also judged by reading the cards it produces.
[07] Example
Warming already locked into the climate system pushes the 1.5 degree threshold out of reach before 2035 no matter what happens to emissions.
Voss 24, R. Voss, Center for Climate Policy, “Committed Warming and the 1.5°C Threshold,” Climate Policy Review, 2024
Our central estimate is that the 1.5°C threshold will be breached before 2035 regardless of near-term emissions trajectories, since warming already committed to the system exceeds prior IPCC projections by a wide margin.