Skip to content
Notifications
Clear all

My agent automates 90% of our new employee setup. AMA.

4 Posts
4 Users
0 Reactions
1 Views
(@startup_ops_lead_jen_2)
Eminent Member
Joined: 4 months ago
Posts: 16
Topic starter   [#574]

Hey everyone! I've been lurking here for a while, picking up so many tips. I had to share a win that's saved our tiny team (6 people!) probably 40 hours of manual work in the last two months.

We hired three new people in quick succession, and I was drowning in the onboarding checklist. Create email accounts, add to Slack, Notion, Google Drive, our project management tool, set up 1Password vaults, schedule intro calls with the team, order swag... you know the drill. It was eating up a full week of my time per hire.

I built a Lindy agent I call "Onboarding Officer" that handles probably 90% of it now. It's not fully autonomous magic—I still have to trigger it and it needs some human verification at points—but the lift is minimal. Here's the flow:

I fill out a simple Google Form with the new hire's name, email, start date, and role. That submission triggers the agent. It waits until 9 AM on the start date, then:
1. Creates the Gmail account and sends a "welcome, here are your creds" email.
2. Sends an invite to our Slack workspace and adds them to role-specific channels.
3. Adds them to the right groups in Notion and shares the onboarding docs.
4. Sends a request to our swag vendor API (using a custom step I wrote) with their size and shipping address.
5. Books a 30-minute intro call with me on my Calendly and blocks a "first day lunch" event on the team calendar.
6. Sends me a daily digest of what it accomplished and any pending items that need my eyes (like approving the swag order).

The biggest "aha" was realizing I could use it to handle the back-and-forth for intro calls. It messages the new hire on Slack their first morning with a link to pick a time for their team intro chats. When they pick, it schedules them automatically.

I'm sure there are fancy SaaS platforms that do this for $10k a year, but our Lindy setup costs a fraction of that. The hardest part was getting the permissions right and figuring out the "wait until start date" logic.

I'm still tweaking it—next up is having it pre-populate their profile in our HR system. Anyone else automating onboarding? I'd love to compare notes or hear about pitfalls I haven't hit yet. What's the one manual step you wish you could automate?

learning daily


learning daily


   
Quote
(@crusty_pipeline_v2)
Estimable Member
Joined: 2 months ago
Posts: 94
 

>needs some human verification at points

Where exactly? That's the critical path. If it's for Gmail account creation and 1Password vaults, fine. But if it's pausing to ask you "should I add them to the project management tool", you've just moved the manual checklist to an approval queue. Not a win.

Also, how are you handling failures? If the Slack API is down and the agent bombs, does it roll back the Gmail account or just leave a half-baked user?


slow pipelines make me cranky


   
ReplyQuote
(@kubernetes_tinker_99)
Estimable Member
Joined: 4 months ago
Posts: 56
 

Good questions. The human checkpoints are only for security-critical actions - creating the core accounts (email, password manager) and provisioning the vault. Everything after that (Slack, Notion, project tool) is automatic based on their department from the form.

For failures, I built a simple state tracker in a small Postgres table. Each step has a status (pending, success, failed). If the Slack API call fails, it marks that step as failed and stops, but it doesn't roll back. That's by design - I don't want it deleting a Gmail account because Slack is having a bad day.

Instead, it sends me an alert with the exact failure and the state of all completed steps. I can then manually fix the Slack bit and tell the agent to resume. It's not a transactional rollback, but it's a clear, actionable failure mode.


#k8s


   
ReplyQuote
(@data_meets_ops)
Estimable Member
Joined: 2 months ago
Posts: 76
 

That's a smart workflow. I've done something similar using dbt Cloud's slim CI jobs to handle provisioning for our analytics engineers, but it was all internal tools.

I'm curious about the state tracker you mentioned later. Did you consider a directed acyclic graph model for dependencies, or is it strictly linear? With linear steps, a failure in step 4 stops step 5, even if they're unrelated systems. A simple DAG would let the swag request go out even if Notion was slow, which might be preferable.



   
ReplyQuote