What automation have you built the most in 2026 — and which part still drives you crazy?
by•
This year, I keep seeing the same workflows: lead follow-ups, email sorting, content publishing, CRM updates, reporting, and customer support.
Creating the first version is usually easy. The frustration starts with app connections, permissions, broken steps, duplicate actions, and deciding when the agent should ask for approval.
What automation do you build most often, and where do you struggle most: setup, integrations, reliability, or maintenance? Ask me all your questions about automation, and I assure you I will solve them.
5 views
Replies
Maintenance, and it isn't close. The one I rebuild most is a daily digest pushing signups, churn and failed payments into Slack, and it's broken four times this year, every time because an upstream field got renamed and nothing threw an error. The digest just quietly went to zero and I noticed a week later. Setup is a bad afternoon, silent drift is a bad quarter, so what I want is an alert when the shape of the output changes rather than when a step fails.
@asadmalik901 This is the most painful type of failure because technically nothing is "broken." Unlike other automation applications, in Awish, you can add a simple English rule like, "If the record, customer loss, or failed payment data suddenly changes shape, is unusually low, or a required field is missing, stop the summary and alert me on Slack."
The agent can monitor not only whether each step is completed, but also the output itself. Furthermore, you can edit these thresholds and rules later from the automation dashboard without recreating the entire workflow. This will catch the silent deviation long before you have to wait for the summary to look wrong.
@alpertayfurr Writing the rule isn't the hard part, knowing which rule to write is. I only added a row count floor after the digest had been silently empty for six days, so every guard I have trails the failure that taught me to write it. What would actually earn the switch is Awish proposing thresholds from my first two weeks of runs instead of asking me to invent a number I've never measured.