WEBVTT

00:00:00.200 --> 00:00:01.680
Your agent moves fast.

00:00:01.680 --> 00:00:03.740
But who challenges the answer?

00:00:03.740 --> 00:00:05.000
Meet MOMM.

00:00:05.000 --> 00:00:06.640
Mixture of Model Modality.

00:00:06.640 --> 00:00:08.600
Keep the agent you trust to do the work.

00:00:08.600 --> 00:00:10.420
Bring in the tools you choose to challenge it.

00:00:10.420 --> 00:00:11.660
One request.

00:00:11.660 --> 00:00:13.480
Parallel reviewers.

00:00:13.480 --> 00:00:15.080
Less copying between chats.

00:00:15.080 --> 00:00:17.520
More questions worth asking before you ship.

00:00:17.520 --> 00:00:18.600
Not a vote.

00:00:18.600 --> 00:00:19.760
Not blind trust.

00:00:19.760 --> 00:00:21.780
Your agent investigates.

00:00:21.780 --> 00:00:23.160
Tests the changes.

00:00:23.160 --> 00:00:26.880
Records what it accepted, what it rejected, and why.

00:00:26.880 --> 00:00:29.240
Three reviewers challenged our own introduction.

00:00:29.240 --> 00:00:31.160
We clarified the setup.

00:00:31.160 --> 00:00:33.440
Rejected a claim that didn't match the code.

00:00:33.440 --> 00:00:34.980
Recorded why.

00:00:34.980 --> 00:00:38.418
Real decisions you can inspect, not a promised productivity

00:00:38.418 --> 00:00:38.800
multiplier.

00:00:38.800 --> 00:00:40.200
Choose your reviewers.

00:00:40.200 --> 00:00:41.060
Keep one writer.

00:00:41.060 --> 00:00:43.020
Own the local decision trail.

00:00:43.020 --> 00:00:44.080
Open source.

00:00:44.080 --> 00:00:45.020
Account based.

00:00:45.020 --> 00:00:46.660
Built for the way you already work.

00:00:46.660 --> 00:00:48.900
Start with one meaningful change.

00:00:48.900 --> 00:00:49.400
MOMM.

00:00:49.400 --> 00:00:50.640
Keep your agent.

00:00:50.640 --> 00:00:51.387
Add the review team.