Internal Comms & Meetings

How to draft a slack announcement as Small IT and ITOps Teams

Internal Comms & MeetingsFor Small IT and ITOps Teams2 apps12 steps~24 min to set up

You're a 2-person IT team and Slack announcements are somehow still a manual fire drill. A planned maintenance window comes up, you need to tell 300 employees what's going down, when, and what to do — but first you're cross-referencing the Jira ticket for the exact scope, checking the AWS Cost Explorer anomaly that triggered the change, pulling the incident timeline from PagerDuty, and trying to remember what you told people last time you ran a similar window. You draft something in Notion, paste it into Slack, realize you forgot to mention the VPN workaround, edit it, and ping your teammate to proofread. Twenty minutes gone for a 4-sentence announcement. Multiply that by every onboarding wave, SSO migration, license audit result, and unplanned outage.

Build this on Starch → See the apps used

Internal Comms & MeetingsFor Small IT and ITOps Teams2 apps12 steps~24 min to set up

Outcome

What you'll set up

A Starch app that drafts Slack announcements from your existing Jira tickets, PagerDuty incidents, or AWS alerts — so you describe the situation once and get a ready-to-post message back

A repeatable workflow that pulls context from the systems you already run (Jira, AWS, Notion runbooks) and formats it into clear, employee-ready language without you writing from scratch

A searchable archive of every announcement you've sent, so when someone asks 'when did we last take down VPN?' you have the answer in seconds

The Starch recipe

Apps, data, and prompts

The combination of Starch apps, the data sources they pull from, and the prompts you use to drive them.

Apps used

Knowledge Management Email Agent

Data sources & config

Jira and Notion are connected from Starch's integration catalog; the agent queries them live when a draft is requested. AWS is connected directly to Starch — Cost Explorer and CloudWatch are queried on demand when an alert fires. PagerDuty is connected from Starch's integration catalog and queried live to pull incident metadata into announcement drafts. Slack is the delivery target; Starch drafts the message and you post it (or set up an automation to post directly to a channel on your approval).

Prompts to copy

Connect Jira from Starch's integration catalog. When I give you a Jira ticket number, pull the summary, affected systems, and planned resolution window, then draft a Slack announcement for a general employee audience. Keep it under 100 words, plain English, no IT jargon. Include what's affected, when, and what employees should do in the meantime.

Connect Notion from Starch's integration catalog. Every time I publish a Slack announcement draft, save a copy to our IT Announcements database in Notion with the date, incident type, and ticket reference so we have a searchable history.

Connect AWS from Starch's integration catalog — Starch syncs your AWS Cost Explorer and CloudWatch data on demand. When an anomaly crosses $500 in unexpected spend, draft a Slack message for the #it-ops channel explaining what spiked, the affected service, and that we're investigating.

Run these in Starch → or paste them into your favorite agent

Walkthrough

Step-by-step

1 Connect Jira from Starch's integration catalog. Starch will query it live — paste in a ticket number or describe the incident and Starch pulls the relevant fields (summary, components, priority, resolution ETA) automatically.

2 Connect Notion from Starch's integration catalog so Starch can read your existing runbooks and write finished announcements back to an IT Announcements log database you designate.

3 Connect AWS directly to Starch — Starch queries your Cost Explorer and CloudWatch on demand, so cost anomalies or service health events can trigger announcement drafts without manual input.

4 Connect PagerDuty from Starch's integration catalog. When an incident fires, Starch pulls the incident title, affected service, and severity to pre-populate the announcement draft.

5 Tell Starch what you want: 'Build me an app that takes a Jira ticket or PagerDuty incident, drafts a Slack announcement for a non-technical audience, and saves a copy to Notion.' Starch builds the app from that description.

6 Set your tone and format rules once — for example: 'Always lead with what's affected, then when, then what employees should do. Never use acronyms without spelling them out. Keep it under 120 words.' Starch applies these on every draft.

7 For planned maintenance windows, trigger the app manually: paste the Jira ticket number, review the draft (usually takes under 60 seconds to generate), make any edits, and post to the relevant Slack channel.

8 For unplanned outages or AWS cost spikes, set an automation: 'When a PagerDuty incident is opened with severity P1 or P2, draft a Slack announcement and post it to #it-incidents for my review within 2 minutes.'

9 For recurring communications — monthly license audit results, quarterly security reminders, onboarding-week system access notices — schedule the automation: 'Every first Monday of the month, pull our Jira IT-board summary and draft a Slack update for #general about upcoming IT changes.'

10 All published announcements sync back to your Notion IT Announcements database with date, incident type, and Jira ticket reference — so next time someone asks when VPN was last down or when SSO migration happened, you search Notion and find the exact message you sent.

11 If you want to go further, tell Starch: 'Build me a dashboard that shows the last 30 IT announcements, grouped by type (maintenance, outage, policy change), with links to source tickets.' Starch builds the view from the Notion data it's already writing to.

12 Post-incident, use the Meeting Notes app to capture the retrospective call, then tell Starch: 'Take the action items from today's retro and draft a follow-up Slack message to #it-ops summarizing what we're fixing and by when.'

See this running on Starch

Connect your tools, describe what you want, and the agent builds it. Closed beta is free.

Try it on Starch →

Worked example

March 2026 Okta SSO Migration Announcement

Sample numbers from a real run

Time to draft announcement (before Starch)	22
Time to draft announcement (with Starch)	4
Employees affected by SSO cutover	300
Jira tickets referenced in draft	3
Notion pages Starch pulled for runbook context	2

In March 2026, your team is migrating 300 employees to a new Okta SSO configuration — every app they touch daily is getting a new login flow. You open Starch, paste in the three Jira tickets covering the migration scope (identity provider cutover, affected apps list, rollback window), and type: 'Draft a Slack announcement for all-company. Explain that SSO is changing on March 14 at 6pm PT, which apps are affected, what employees need to do before Friday, and who to contact if their access breaks.' Starch queries Jira live, pulls the affected apps list from the ticket description, cross-references the runbook you have in Notion, and returns a 90-word draft in under 30 seconds. You trim one line, post it to #general and #it-announcements, and Starch writes the final version back to your Notion IT Announcements log with the date and ticket references. Two days later when the VP of Sales asks 'did IT send something about this?' you find the exact message in 10 seconds instead of scrolling through Slack history.

Measurement

How you'll know it's working

Time-to-post for unplanned outage announcements (target: under 5 minutes from incident open to Slack message sent)

Percentage of IT announcements with a corresponding Notion archive entry (target: 100% — no more lost comms history)

Number of follow-up 'what's happening?' Slack DMs to IT after a major announcement goes out (a proxy for announcement clarity)

Reduction in time IT spends per announcement draft (before vs. after — track weekly for first 30 days)

PagerDuty P1/P2 incidents with an employee-facing Slack notification sent within 3 minutes of incident open

Comparison

What this replaces

The other ways teams handle this today, and how the Starch version compares.

Writing the announcement manually in Slack or Notion

You control every word but you're writing from scratch every time — no context pulled from Jira or PagerDuty, no archive built automatically, and every announcement is only as good as what you remember to include at 11pm during an outage.

Slack Workflow Builder

Good for simple templated notifications (e.g., 'maintenance at 6pm'), but it can't read a Jira ticket, pull an AWS anomaly, or generate plain-English prose — you're still filling in the template by hand.

ChatGPT or Claude directly

You can paste in ticket details and get a draft, but there's no connection to Jira, PagerDuty, or Notion — you're copying and pasting context every time, and nothing gets archived automatically.

Zapier or Make (for automated notifications)

You can trigger a Slack message when a Jira ticket hits a certain status, but the message is a template with field substitutions — not a drafted, human-readable announcement. And building the Zap still takes longer than describing what you want to Starch.

On Starch RECOMMENDED

One platform — knowledge management, email agent all running on connected data. Setup in plain English; numbers stay current via scheduled syncs and live agent queries.

Try it on Starch →

FAQ

Frequently asked questions

Can Starch post directly to Slack, or does it just draft the message?

Both. Starch connects to Slack from its integration catalog and can query or post to channels. You can set it up so Starch drafts for your review first (the safer default for all-company messages) or posts automatically to a specific channel like #it-incidents when a P1 PagerDuty alert fires. You choose the approval step based on the announcement type.

We use PagerDuty for on-call. Can Starch actually read our incidents?

Yes. PagerDuty is reachable from Starch's integration catalog — the agent queries it live when your app runs. Connect it once from the integration browser and Starch can pull incident title, severity, affected service, and timeline to pre-populate any announcement draft.

Our runbooks live in Notion and nobody reads them. Can Starch actually use them to write better announcements?

Yes. Starch connects to Notion from its integration catalog and can query your pages and databases live. If your Notion runbook for a VPN outage includes a 'what employees should do' section, Starch can pull that text and include it in the draft. It won't invent workarounds — it uses what's actually in the doc.

Is Starch SOC 2 certified? We have a security team that will ask.

Not yet — Starch is not SOC 2 Type II certified as of today. That's worth knowing before you route sensitive incident data through it. For most IT announcement workflows (maintenance windows, SSO changes, license audits), the data involved is operational rather than personal, but check with your security team on your specific data classification.

We use Okta and Jamf. Can Starch connect to those?

Okta and Jamf are reachable from Starch's integration catalog — the agent queries them live. Starch doesn't replace them and won't push policy changes back into Jamf, but it can read device data, user provisioning status, or enrollment state to pull context into an announcement or dashboard. If a specific Okta or Jamf endpoint you need isn't in the catalog, Starch can also automate the web UI through browser automation — no API required.

What if I want to build a full IT onboarding orchestrator, not just announcements?

Describe it to Starch and it builds it. For example: 'Build me an app that takes a new hire's name and start date from our Jira onboarding ticket, checks their Okta provisioning status, drafts a welcome Slack message to #general, and adds a checklist task to follow up on day 3 if their laptop hasn't been enrolled in Jamf.' That's a single natural-language prompt — Starch assembles the connections and the workflow from there. The Slack announcement capability described on this page is one component of a larger onboarding orchestrator you can build the same way.