WEBVTT

00:00:00.100 --> 00:00:03.275
AI SDLC means the software development lifecycle

00:00:03.288 --> 00:00:06.088
with AI Writing code is one part

00:00:06.100 --> 00:00:08.988
of the process Your team also needs

00:00:09.000 --> 00:00:12.275
to assign tasks prepare context track agents

00:00:12.450 --> 00:00:16.288
review results and preserve lessons xDev Hive

00:00:16.300 --> 00:00:19.312
orchestrates these steps connecting your team coding

00:00:19.325 --> 00:00:22.188
agents and repositories so work moves forward

00:00:22.200 --> 00:00:23.650
with clear responsibility

00:00:25.558 --> 00:00:27.658
One agent writes code and another reviews

00:00:27.671 --> 00:00:30.271
it But who owns the task and

00:00:30.283 --> 00:00:33.483
where is the result When instructions chat

00:00:33.496 --> 00:00:36.146
decisions and work status live apart the

00:00:36.158 --> 00:00:39.871
team struggles to continue Hive connects tasks

00:00:40.033 --> 00:00:44.346
runs and project knowledge Documents memory and

00:00:44.358 --> 00:00:46.533
skills provide context for agents to do

00:00:46.546 --> 00:00:47.921
the work and for your team to

00:00:47.933 --> 00:00:49.021
assess the results

00:00:50.892 --> 00:00:53.792
Hive organizes documents by shared system and

00:00:53.804 --> 00:00:57.304
project scope Your team can write instructions

00:00:57.467 --> 00:01:00.729
standards and technical decisions then inspect version

00:01:00.742 --> 00:01:03.454
history and diffs This is the real

00:01:03.467 --> 00:01:06.392
Hive interface running with illustrative Atlas project

00:01:06.404 --> 00:01:10.004
data Documents used as agent context can

00:01:10.017 --> 00:01:13.204
be synchronized into your repository keeping instructions

00:01:13.217 --> 00:01:14.217
close to the code

00:01:16.100 --> 00:01:18.225
Not every instruction belongs in one long

00:01:18.238 --> 00:01:22.175
file Hive supports scoped documents and path-specific

00:01:22.188 --> 00:01:24.625
guidance so relevant standards can reach the

00:01:24.638 --> 00:01:27.750
right part of a repository The context

00:01:27.762 --> 00:01:30.400
screen lets administrators inspect what agents will

00:01:30.412 --> 00:01:34.862
read Desktop synchronization and agent configuration determine

00:01:34.875 --> 00:01:36.888
how those instructions are distributed to each

00:01:36.900 --> 00:01:37.462
machine

00:01:39.350 --> 00:01:41.650
Memory preserves what a task teaches a

00:01:41.662 --> 00:01:43.975
decision a convention or a problem to

00:01:43.988 --> 00:01:47.338
watch for Agents search within projects and

00:01:47.350 --> 00:01:49.288
can use knowledge shared across the team

00:01:50.150 --> 00:01:52.000
On the hub new agent memory is

00:01:52.012 --> 00:01:55.625
reviewed before other agents use it Administrators

00:01:55.638 --> 00:01:58.125
can revisit older entries and inspect knowledge

00:01:58.138 --> 00:01:59.900
linked to files that have changed

00:02:01.808 --> 00:02:03.746
A useful review procedure should not need

00:02:03.758 --> 00:02:06.558
to be rewritten for every agent Hive

00:02:06.571 --> 00:02:09.708
manages shared team skills and project-specific skills

00:02:10.571 --> 00:02:12.721
A project skill overrides a shared skill

00:02:12.733 --> 00:02:15.308
with the same name Claude Code can

00:02:15.321 --> 00:02:18.008
load synchronized skills from the repository while

00:02:18.021 --> 00:02:20.108
other agents discover and read skills through

00:02:20.121 --> 00:02:23.533
MCP Proposed changes can be reviewed by

00:02:23.546 --> 00:02:24.496
an administrator

00:02:26.392 --> 00:02:28.742
Work across several agents needs clear state

00:02:28.754 --> 00:02:32.929
and responsibility Hive supports tasks with dependencies

00:02:33.079 --> 00:02:35.742
progress from todo to completion and time-limited

00:02:35.754 --> 00:02:39.017
claims Agents read the context and claim

00:02:39.029 --> 00:02:42.429
a task before starting Updating the result

00:02:42.442 --> 00:02:44.104
helps the team see what is underway

00:02:44.329 --> 00:02:45.992
what is blocked and what is ready

00:02:46.004 --> 00:02:46.679
for review

00:02:48.558 --> 00:02:50.583
The desktop app runs coding agents through

00:02:50.596 --> 00:02:52.896
the CLIs and profiles configured on your

00:02:52.908 --> 00:02:56.546
machine The hub shows reported runs logs

00:02:56.671 --> 00:03:00.508
and results Depending on configuration the runner

00:03:00.521 --> 00:03:03.358
can use separate worktrees switch profiles after

00:03:03.371 --> 00:03:07.458
quota limits or compare several candidates Usage

00:03:07.471 --> 00:03:09.346
limits and costs still depend on each

00:03:09.358 --> 00:03:11.596
agent provider and the account you configure

00:03:13.475 --> 00:03:16.025
Agents can propose document and skill updates

00:03:16.037 --> 00:03:19.425
rather than overwrite shared standards A permitted

00:03:19.437 --> 00:03:21.775
reviewer inspects the reason and diff before

00:03:21.787 --> 00:03:24.962
approving a change When the original document

00:03:24.975 --> 00:03:27.462
has changed Hive flags a conflict so

00:03:27.475 --> 00:03:29.800
an older proposal does not silently replace

00:03:29.812 --> 00:03:33.025
newer knowledge Human reviewers remain part of

00:03:33.037 --> 00:03:34.475
how team knowledge evolves

00:03:36.392 --> 00:03:38.229
Start with local data in the desktop

00:03:38.242 --> 00:03:40.679
app. When your team needs a shared

00:03:40.692 --> 00:03:43.379
workspace the web hub brings together accounts

00:03:43.517 --> 00:03:46.979
project permissions documents memory and reports from

00:03:46.992 --> 00:03:51.579
multiple machines SSO Teams or Slack webhooks

00:03:51.767 --> 00:03:54.642
semantic memory search and external file storage

00:03:54.654 --> 00:03:57.604
are optional integrations that require configuration for

00:03:57.617 --> 00:03:58.479
your environment

00:04:00.392 --> 00:04:02.929
Start AI SDLC with real work a

00:04:02.942 --> 00:04:05.742
repository a clear task and a configured

00:04:05.754 --> 00:04:09.917
coding agent Hive coordinates tasks context and

00:04:09.929 --> 00:04:13.079
runs Your team reviews code results through

00:04:13.092 --> 00:04:15.442
pull or merge requests and approves proposed

00:04:15.454 --> 00:04:18.792
knowledge updates Lessons remain available for the

00:04:18.804 --> 00:04:22.267
next development cycle Start on one machine

00:04:22.492 --> 00:04:23.904
then expand to a hub for more

00:04:23.917 --> 00:04:25.054
people and agents

00:04:26.933 --> 00:04:28.796
Read the xDev Hive README to set

00:04:28.808 --> 00:04:31.246
up the desktop app or hub Add

00:04:31.258 --> 00:04:34.521
a repository prepare standards and skills connect

00:04:34.533 --> 00:04:36.971
agents through MCP and try a task

00:04:36.983 --> 00:04:39.921
with a clear review process The screens

00:04:39.933 --> 00:04:41.621
use demo data and the narration is

00:04:41.633 --> 00:04:46.208
synthetic xDev Hive orchestrates AI SDLC from

00:04:46.221 --> 00:04:48.521
task to review with knowledge that grows

00:04:48.533 --> 00:04:50.183
after each development cycle