Not sure which team fits?Talk to our engineers →

Engineering.

GoalTested fixes merged every week. People decide the risky ones.

An AI engineering team for the work your team never gets to. Agents find bugs and security gaps, write and review the fix, and answer questions in Slack, inside your own GitHub and Linear. Low-risk changes merge on their own. High-risk ones wait for you.

  • Works with GitHub, Linear and Slack.
  • Asks your engineering lead to approve anything important.
  • Comes with its own app in Workspace: Engineering Fleet.
01 · The team

6 agents, one job each.

One lead agent keeps the team working towards the goal. Each other agent does one job, then passes the work on.

E
Engineering ManagerTeam lead. Watches the task board and clears stuck work. Tells you about anything it cannot fix.
R
Reliability and Security EngineerFind real bugs and security gaps in your logs, alerts and code. Each finding comes with evidence.
T
Ticket MasterChecks each finding against the code, combines duplicates and ranks the rest. Passes them to Ticket Pilot.
T
Ticket PilotWrites the change and its tests on a branch, then opens a pull request.
C
Code ReviewerReviews every pull request and comments on the lines that need changes. Rates each one low, medium or high risk.
S
Support EngineerAnswers questions in Slack using your logs, code and build results (CI). Shows the evidence.
02 · A day’s work

A bug found at 09:40.

The goal: turn problems in the logs into tested, merged fixes. Example

  1. Finds failed updates in the logs. Checks the code and finds the cause: the account ID is never checked. Opens ticket ENG-214 with the evidence.
  2. Confirms ENG-214 in the code, combines it with a duplicate ticket and gives it high priority.
  3. Writes the fix and its tests, and opens pull request #1182, linked to the ticket.
  4. Reviews #1182 and finds nothing that blocks it. Rates it medium risk, based on the files it changes.
  5. Tom has allowed medium-risk changes to merge on their own this week. #1182 merges once all automated tests pass.

Result. From finding to merged fix in under an hour. No one had to do any of it by hand.

03 · What your engineering lead sees

Engineering Fleet.

One page shows the team’s work: the numbers, the work in progress and what needs a person to approve.

527Pull requests mergedin 7 days
51 minTypical time to mergefrom open to merged
44%Needed a persondown from 84%
503Tickets completedin 7 days

These numbers come from Lua’s own code over 7 days. Tom and the tickets are examples.

Works withGitHubLinearSlackand many other tools you connect
04 · Questions

Common questions.

Can it merge without us?

Only at the risk levels you allow, and only after tests pass and the review approves. High-risk changes always wait for your team. You start with no automatic merges and add risk levels week by week.

Which accounts does it use?

Its own bot accounts on GitHub and Linear, added to your organisation. It never uses a person’s login.

Does our code go to an AI model?

Yes. The agents use Anthropic’s Claude models. Our contract does not allow your data to be used for training. Every provider we use is on our subprocessor list.

How do I get access?

Talk to our engineers about joining the design partner beta.

Can we try it before it touches real work?

Yes. Every team starts in practice mode on sample data. Nothing goes live until you approve it.

What does it cost to start?

You can start building free in Workspace. Talk to our engineers if you want them to build it with you.

Set up this team in Workspace.

Describe the work in Workspace and Lua builds the team for you. Or our engineers can build it with you.