WEBVTT

00:00:00.000 --> 00:00:04.019
 So maybe your AI coding assistant is brilliant at writing code, but it has no idea why you

00:00:04.019 --> 00:00:08.359
 chose Supabase over Firebase, why you avoid complex permissioning, or why you decided

00:00:08.359 --> 00:00:10.840
 against that third-party integration last week.

00:00:10.840 --> 00:00:15.039
 Every chat session you're re-describing architectural decisions in product trade-off!

00:00:15.039 --> 00:00:19.020
 One of our users, Matt, told me he feels like a shit translator between what customers say

00:00:19.020 --> 00:00:20.399
 and AI builds.

00:00:20.399 --> 00:00:21.600
 Brief fixes this.

00:00:21.600 --> 00:00:22.600
 Let me show you.

00:00:22.600 --> 00:00:26.719
 This is Brief, where tactical teams record the why behind their code.

00:00:26.719 --> 00:00:29.639
 These are actual decisions we've made while building Brief itself.

00:00:29.639 --> 00:00:35.520
 No complex permissions in our MVP. Or the use of Supabase, with RLS. Or Workspace Isolation.

00:00:35.520 --> 00:00:39.959
 We drop customer interview transcripts, wireframes, anything the AI needs to know to understand

00:00:39.959 --> 00:00:40.959
 our product.

00:00:40.959 --> 00:00:42.959
 Here's where it gets interesting.

00:00:42.959 --> 00:00:46.400
 I'm going to ask my AI to add a role management system.

00:00:46.400 --> 00:00:48.599
 Watch what happens.

00:00:48.599 --> 00:00:51.159
 Brief just told my AI about decision 7.

00:00:51.159 --> 00:00:54.360
 We specifically decided against complex permissions for the MVP.

00:01:00.639 --> 00:01:06.559
 It suggests simple role assignment and said, no back and forth, no rebuilding what we already

00:01:06.559 --> 00:01:07.559
 decided against.

00:01:07.559 --> 00:01:10.040
 The AI remembered decision 7, so I don't have to.

00:01:10.040 --> 00:01:12.839
 As you saw last week, we started with just an MCP.

00:01:12.839 --> 00:01:13.839
 That was decision 1.

00:01:13.839 --> 00:01:18.279
 The user feedback pushed us in developing UI faster, so we shipped one and made a superseding

00:01:18.279 --> 00:01:19.839
 decision, decision 6.

00:01:19.839 --> 00:01:23.760
 Brief helped us track that evolution and stay consistent across 16-plus decisions while

00:01:23.760 --> 00:01:25.080
 building at startup speed.

00:01:25.080 --> 00:01:26.639
 We're onboarding teams all week.

00:01:26.639 --> 00:01:30.319
 Teams who are tired of retraining an AI every conversation and want the right thing built

00:01:30.319 --> 00:01:31.319
 the first time.

00:01:31.319 --> 00:01:34.480
 Your AI doesn't know why that code exists either, but Brief does.

00:01:34.480 --> 00:01:36.720
 Sign up at briefhq.ai.