Te
See the demo Home
Unedited test run · 19 Jul 2026

We ran 10 creatives through Telll. In five dialects.

Every status tool demo you have ever seen was scripted in clean English. This one was not. These are synthetic agency creatives replying in Jordanian, Gulf, Saudi, Franco-Arabic and English–Arabic code-switching — put through the real engine, not a mockup. Including the three things it got wrong.

10 users × 3 rounds 20 replies 5 dialects isolated in-memory critic: opus

What the team wrote back

0 / 16
Press Replay the run to watch it happen.

Command Center

fills itself in
0Blocked
0At risk
0On track
0No update
Shared bottlenecks
Clusters appear as replies land.

What it got wrong

auto-detected
Findings appear at the end of the run.

Why show you the bugs?

Because you are being asked to point this at your team, and every competitor page in this category shows a perfect scripted English demo that tells you nothing about whether it survives contact with how your people actually write.

Your creatives write in dialect. They code-switch mid-sentence. They say "خلصت تقريباً" and mean 70%. If the parser cannot hold that, nothing else matters — and you should be able to check before you commit, not after.

Run swarm-2026-07-19T17-54 · replayed verbatim from the report file.
Synthetic users, isolated in-memory store, no production data, no customer data, no Trello writes.
Agents always reply. Real people do not — which is what the audit is for.