AI for engineering teams earns its keep in the writing around shipping: the standup, the weekly report, release notes, the first hour of an incident and the issue backlog. In Arclyx, five agents read GitHub, write those up and post one message to Slack. None of them comments, merges or files anything.
Standup · Tuesday · 9 merged Monday, 2 waiting on review
#512 Add retry to webhook sender, #515 Show invoice totals in the account currency, and three more.
Each with its link, so nobody has to say it out loud.
#518 Paginate the audit log export, open 31 hours. #521 Cache dashboard charts per workspace, open 26 hours.
Left out: one approved pull request and one draft.
On the status, notes and first drafts nobody has time to write. Each agent reads GitHub without changing it, starts from one sentence and sends one short message. Three also save a page in Notion for a person to finish: the weekly report, the release notes and the postmortem.
Every weekday, what merged, what's waiting on review over a day and which priority issues stopped moving, grouped by person.
GitHubEvery Friday, the week's merged work in a few themes, releases by component and the risks, as a Notion page for leadership.
GitHub
NotionWhen a release is published, customer-facing notes grouped New, Improved and Fixed, drafted in Notion and traced to pull requests.
GitHubNotionWhen something breaks, every change from the last 48 hours, related open issues and the suspects as hypotheses, in a postmortem draft.
GitHubNotionEvery Monday, the week's issues nobody owns by area, likely duplicates paired and the longest-untouched issues listed.
GitHubEach takes a moment where the writing usually slips: the issue nobody triages, the review that waits, the release nobody announces, the incident nobody wrote up and the week nobody summarised. They don't hand work to each other; each reads the same repository at its own moment.
Issue triage counts it with the week's untriaged issues, suggests an area and checks whether it's already filed.
Async standup lists it once it has waited a day for review, under the person who opened it.
Release writer reads the pull requests in it and drafts notes a customer can read.
Incident brief lists what changed in the last 48 hours and drafts the postmortem around it.
Weekly eng update counts what shipped, names the risks and writes the report for leadership.
Not these five. Each gets GitHub tools that list, search and read, and nothing that comments, approves, merges or pushes. In Notion their only writing tool creates new pages, so they can't edit an existing one. Each run's one message goes to you or your team, and every run is recorded.
Hand each piece of recurring writing to its own agent and let it run on a schedule, a webhook or an @mention. Here is a week of an engineering team on Arclyx: each step is the sentence that built an agent, then what it posted in Slack. Try it runs a new agent once as a dry run first.
“Every Monday at 9, write the issue triage digest for our GitHub repository: open issues from the last 7 days that nobody has triaged or assigned, likely duplicates of existing issues, and open issues untouched for 30 days, with a suggested owner area for each, and post it to #eng-triage.”
#508 Webhook retries stop after the first failure, #506 Export times out on large workspaces, and 12 more.
Owner area from each title; the ones it can't place go under Unclear.
Both ask for exports to respect workspace permissions.
Likely, not certain: two of three searches used.
“Every weekday at 9, write the async standup for our GitHub repository: which PRs merged yesterday, which open PRs have waited on review for over a day, and which open issues are blocked or haven't moved in 7 days, grouped by person, and post it to #eng-standup.”
#512 Add retry to webhook sender, #515 Show invoice totals in the account currency, and three more.
Each with its link, so nobody has to say it out loud.
#518 Paginate the audit log export, open 31 hours. #521 Cache dashboard charts per workspace, open 26 hours.
Left out: one approved pull request and one draft.
“When a release is published in our GitHub repository, write customer-facing release notes from the PRs merged since the previous release of the same component, grouped into New, Improved and Fixed, in plain words. Save them as a draft Notion page under Release notes and post a short summary with the link to #releases.”
You can now export any report as a CSV file from its menu.
From #466.
Search keeps your filters when you go back, and loads faster in large workspaces.
From #462 and #471.
“When I describe an incident, list what changed in our GitHub repository in the last 48 hours (merged PRs, releases, commits to main), find related open issues, and draft a postmortem skeleton as a new Notion page under Incidents: timeline, what changed, likely suspects and open questions. Post the likely suspects and the page link to #incidents.”
Merged at 8:40, twenty minutes before the reported start.
Hypothesis: an extra call on every checkout. Not confirmed as deployed.
Merged 15 hours before the start, so a weaker timing match.
Hypothesis: building the index slowed writes to orders.
“Every Friday at 4, write the weekly engineering update for leadership from this week's merged PRs, releases and closed issues in our GitHub repository: what shipped, what's in progress, and risks. Save it as a new Notion page under Engineering updates and post a short summary with the link to #eng-leads.”
Failed webhook deliveries now retry, and slow report queries moved off the main database.
Examples: #512, #527, #530.
Nightly backups skip the largest workspace, and nobody is assigned to it.
Read in detail, with the evidence on the page.
Software that takes a recurring engineering chore, such as the standup or release notes, reads your repository and writes it up for the team. In Arclyx each one starts from one sentence, reads only the tools it's given and sends one short message.
No. They read what your team already did in GitHub, the pull requests, reviews, releases, commits and issues, and write it up. Your engineers keep writing and reviewing code with whatever they use today.
GitHub and Slack, plus Notion for the three that save a page. Each agent gets GitHub tools that list, search and read, and a Notion tool that only creates new pages.
No. They read pull requests, releases, commits and issues. Incident brief lists what GitHub can't answer, such as what's deployed and what the logs show, as open questions for the person on call.
They count from GitHub's own lists and say where a number has limits: the weekly report marks closed issues as an upper bound, the standup marks a review it couldn't read as unverified, and the triage digest says when a count is incomplete.
One sentence in, a working agent out. Join the beta and we'll set you up.
