Activity
Mon
Wed
Fri
Sun
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
What is this?
Less
More

Owned by Dylan

👨🏻‍💻 The simple, effective, and affordable approach to the complicated, over-hyped world of AI automation.

Automate your entire business with ready-to-use AI agents that do the work for you, with zero technical setup, for less than $1/day.

100 contributions to AI Automation Society
Cheapest way to find AI videos for realism at scale?
Building a tool that rates AI videos (Seedance, Higgsfield, etc.) 1–10 on how real they look. Need to run 10K+ videos a day. Gemini Argon nails the quality but costs ~$1K/day. Open-source tools I tried either can't judge realism or are too slow. Has anyone solved this? anything that worked for you at volume?
0 likes • 2h
Ten thousand videos a day is nuts, hope you find a cheap option.
Forklift driver: what would you try with AI in the warehouse I work in?
Hi everyone, Stelian again, the forklift driver learning Claude Code. So far I've asked you about work for other people, a friend's B&B and a small construction company. This time it's about my own job. I work full time in a warehouse that supplies a chain of retail stores in Italy. Picking is all manual: people walk the aisles with a list and pick by hand, no robots. The warehouse opened recently. In the first year the goal was a total number of cases shipped, and we hit it. Now management has told us openly about two new goals: - more cases per picker with the same people (the average is around 600 per shift, and they'd like it closer to 1,000) - fewer picking errors (today about 10 in every 100 picks, the target is 1) From what I've read, manual picking usually runs at 1 to 3% errors, so the first thing I want to understand is how we count them. From what I see on the floor, two causes come up a lot, although nobody has checked yet if they're the main ones. Some pallets have identical outer boxes with different products inside, so it's easy to take from the wrong one. And when a picker has to take 10 cases, they often scan the same box 10 times. The scanner beeps 10 times either way, so if they then take 9 cases, or one from the wrong pallet, the scan doesn't catch it. Over time I'd like to propose something that helps with these, and maybe one day do it as a real project for the company. I'm an employee there, so company data can't leave the building or go into my personal tools without written permission. Anything I do starts with asking my manager. I did some research, and these are the ideas that look realistic to me: 1. Moving the most picked products closer to the shipping area (ABC analysis on the order history) 2. Grouping orders to shorten the walking routes 3. A simple log of picking errors and their causes, starting from the stores' complaints 4. An assistant that answers questions about our procedures, in several languages, for new hires
0 likes • 2h
Cool idea Stelian
My phone agent read its own stage direction out loud: "silent tool calls"
At the end of every call, my notary agent runs two tools. One logs the lead to a sheet, one texts the owner. The caller isn't supposed to hear any of it. They hear it. "Let me get that over to Nick now." "Before I let you go, let me send that over to Nick." The prompt says, in plain words, never mention to the caller that you're doing it. First I found a retry that was adding an extra turn, and fixed it. The next call was quiet and I called it done. The call after that said "Let me log that for you now." So I banned the phrases it had used and moved the tool instructions to the end of the prompt. In September, at the close of a call, it said "silent tool calls." Out loud. My section heading reads "call both tools, silently," and the model turned the label into the line. It was the only sound in a 14-second wait. What I think is going on: on Vapi, any text the model writes next to a tool call gets spoken. I'd already set the tools' start message to None, so the silence is real, and the model fills it. A banned-phrase list can't close that gap. The model just reaches for the next thing that fits. My own rule is two prompt fixes on one problem, then change the layer. I'm at two. If you run a Vapi agent that calls tools at the end of a call, how did you get it to stay quiet? A prompt line, a setting, or something in the tool turn itself?
0 likes • 2h
Funny that it narrated its own stage direction. Did moving the tools to the end of call hook fix it?
Has an automation of yours ever “worked” for weeks while doing nothing?
A builder in this group told me his content automation ran clean for four weeks while doing nothing. The model it was pinned to had been retired. No error, nothing red. He only noticed when engagement dropped. I’m building a small watchdog for exactly that. Your workflow sends one ping when it finishes (“produced 12 items”). If runs come back empty, or drop well below normal, or stop arriving, you get an alert. I only see counts and timestamps, never your data. I need 2 builders to try it free for a week. A copy of a workflow with fake data is completely fine, so nothing client-related is at risk. After the week I’ll post exactly what it caught or didn’t, even if it caught nothing. Two questions: 1. Which of your automations would hurt most if it quietly stopped producing results? 2. Has that ever happened to you, and how long did it take to notice? Comment or DM me if you’re in.
1 like • 20h
The quiet-failure example is exactly why I like count-based alerts. A drop to zero output would catch this before engagement ever dipped.
The one test that catches a quiet failure before your client does
Your workflow shows green. Your leads stopped coming in anyway. That happened to me twice now. Once a calendar token expired. Once the CRM write failed while every step before it still looked fine. The workflow log said success. The client found out when a customer showed up and no booking was in the calendar. So now I run one test before any build goes live. Send a real test lead through the full flow. Then check the other end. Not the log. The calendar event. The CRM record. The text the customer got. If the thing at the end is missing, the build is not done yet. Do you check the receiving end too, or do you trust the green check?
0 likes • 20h
I like checking the other end. A workflow log can be green while the customer is still waiting.
1-10 of 100
Dylan Murray
5
236 points to level up
@dylan-murray-4072
💸 I help entrepreneurs get leads, save time, and scale by automating your business the right way 👨🏻‍💻 Founder of ClickFlow Automation

Active 2h ago
Joined Sep 15, 2025
Manitoba, Canada
Powered by