Prep
Bring a case, a block, or a half-formed idea. Cross names the missing warrant and describes the card that would fill it.
Argue like the room is watching.
An AI debate coach that hands back your own judgment, sharper, not a script to read.
Built for the round you are actually about to debate.
Cross reads your case, your flow and your judge, then coaches from what it has internalised about real rounds. It argues with you about strategy instead of handing you a paragraph to read aloud. Every recommendation carries its reasoning, because the reasoning is the part that transfers to the next round.
- PolicyCX · CEDA · NDT
- Lincoln-DouglasTRAD · CIRCUIT
- Public ForumLAY-ADJACENT
- ParliamentaryIMPROMPTU
What Cross does.
Drill
A generated round, an opponent with a tendency, a judge with a quirk. Write or speak the next speech, then get graded.
Judges
The paradigm for the judge you actually drew, read into what it changes about speed, theory and how the ballot gets written.
Coach
Ask anything mid-prep. Pre-Round mode adds the opponent’s disclosed positions from your own linked Tabroom account.
Three steps, one sightline.
1
Bring your round
Upload the case, paste the flow, record the speech, or just describe the pairing you were handed.
2
Cross reads it
Your material, your judge’s paradigm, and what real rounds on this kind of argument tend to do.
3
You get the ledger
Options with tradeoffs and a visible score, never a verdict you are asked to trust blind.
Every number shows its work.
Drill grading starts from an even round and adjusts in log-odds, one factor at a time. You see the whole ledger, so you can disagree with a weight instead of arguing with a black box.
Synthetic example
Prior: an even round
50%Extended the dropped turn cleanly
+0.62Weighed magnitude against timeframe
+0.41Preempted the opponent's likely 2NR collapse
+0.35New argument in the final speech
−0.28No judge-specific framing (the paradigm asks for it)
−0.1971%Structured judgment for comparing moves, not a claimed win rate.
The people behind it.
First Last
Product & engineering
Builds the coaching loop, the corpus pipeline, and everything that talks to Tabroom.
First Last
Debate lead
Sets what good feedback sounds like across Policy, LD, PF and Parli.
First Last
Evidence & corpus
Cuts, tags and ingests the material Cross reasons from, none of it ever quoted back.
First Last
Design
Keeps the interface out of the way of the round you are about to debate.
Placeholder names, roles and portraits
Built for competitive debate.
Your corpus stays yours
Cross reasons from the material you ingest and never names a file, a case or a cite back at you.
Real paradigms, your account
Judge paradigms and disclosed positions are pulled with your own linked Tabroom account, or not at all.
No invented evidence
When you need a card, Cross describes the card to go cut. It will not fabricate a cite.
