Curious how many of you use low/no code automation tools like n8n or Zapier and how do you use them? Thinking of setting one up, but not sure which one to use to automate a few things.

Bonus points if there is a way to set up webhooks on free Teams or Signal.

I don't use that, but we're in the process of automating Ms. KRSW's job via Python, using Claude AI to write the code. Their current systems were processing about 3-5 projects per hour. The little Python script Claude did for us in about 90 minutes of messing around with it can do 1 project every 10-30 seconds, and so far does a better job.

Neither one of us knows Python. My last programming gig was 30 years ago and that was in C, and I've not touched it since. We're learning Python so we can fix this thing if it breaks.

Hrm. My use case is actually for personal use. Starting with RSS feeds, but building out some automation tasks for my home lab to do a few things so I don't have to worry about it (eg, forwarding logs, kicking off scans, etc.). I'm also learning python, but that's for company resources so whole different ballgame.

Check out Momen.app and Bubble.io � these are full stack no-code dev tools that let you do anything a full program could do. A little more learning curve and a LOT more functionality.

I use Power Platform a bit at work, but that's about it.

I have only used Power Automate at work. Co-workers talk about using Replit, StackBlitz, etc in their personal life, but I haven't tried any of them. I don't want to pay some platform massive amounts of money every month to host my creation.

Momen does not do that, and Bubble charges only a small fee.

Thanks. Probably more than what I need. I doubt what I would need (for a homelab) would be anywhere near what any of these platforms would provide. Even now it's more convenience than need for me to get this stuff going.

Platforms like n8n and zapier and make have free tiers of service... if it doesn't have high requirements (eg, taking RSS feeds and manipulating them) then you shouldn't even have to worry. I was actually considering hosting an instance of n8n on a low power mini PC at home. Just need to install linux and docker. Now if you're talking a full blown interactive app managing your home, that's a different story.

I am currently using Comet (perplexity.ai browser) to automate some financial research and general invoice reconciliation from email/gdrive. I know there is Dai and GenSpark too but Comet is my thing at the moment. They are all invite only at the moment though.

ChatGPT / Claude recently added MCP support so it can be linked to a variety of apps. Have a look at them too.

Claude has had MCP support for a while, but I didn't know about ChatGPT. I'll have to give that a try.

Hadn't heard of Comet. No wonder why Perplexity made an offer for Chrome if the courts were going to force Google to divest the browser.

Ah didn�t know about Claude.

Yeah it�s their invite only browser. If you have mac, look at Dia and genspark. I believe all are invite only but you get them within few days, whereas comet you gotta wait a while or get an invitation from existing member.

yeah Google gets to keep chrome for now, but comet trying to buy them for like 1.5-2x their worth was interesting.

I use lots of coding agents. I geek out on this stuff. Even run some LLMs on a server at home.

I'm not a developer, but I was a computer science minor. I'm pretty good with Python and C#. I use AI coding agents to build quick PoC (proof of concept) apps for my work. It's really insane what some of these things can do. They do make mistakes, and it's important to review the code (or even have AI review it, lint it, test it, etc.)

My preferred coding agent is Claude Code. I have a Pro subscription through my employer (it's $20/mo). I also use Gemini's free CLI coder from time to time. I hear good things about ChatGPTs recent codex agent, but I don't want to pay for another sub to try it out. In general coding agents are really well versed in the most common languages like Python, Javascript, C#, etc. However, I've found they're not so good on lesser used languages like C++. They're still great, but I suspect they get much less usage so there's less to train them on. My company built an internal coding agent that's trained on millions of lines of code in our codebases. It uses Claude as its foundational model.

I use CC in VS Code, but there are tons of other coding agents. Cline and Roo Code are really great for trying out different models and creating an agentic workflow. If you want to test some out, I'd recommend getting an OpenRouter account, put $5-10 in there, connect it to Cline/Roo and test some models (many are free). If you have a Gemini, Claude Pro, or ChatGPT paid subscription, you can get a lot of "free" use out of their CLI tools. They plugin really well to IDEs like VS Code, Cursor, etc.

It's fascinating (to me) to watch how fast things are evolving. The big three (OpenAI, Google, Anthropic) are constantly trying to one up each other.

This is what's so cool about coding agents. You really don't have to know the language syntax, you just need to know how to manage the agents. The future of coding will be architecting, prompting and context engineering. Having said that, if you want to make production grade code, it is important to put the code in a proper review/test cycle (with a dev who can read/interpret it).

For stuff I actually have to code, ChatGPT or Copilot is good enough. I just get it to kick out a basic set of code and then modify it to my needs. Unfortunately it can be a challenge, but with client privacy, it's better to err on the side of caution.

That said, I often try to validate any code by running it through a different AI to make sure it actually makes sense. AI hallucinations just makes things difficult.

I also don't put client data into a public LLM. While many of them (especially paid versions) do have privacy/training policies in place, I still wouldn't chance it. Instead, I usually have it write code "around" the data. i.e. instead of providing the data, I just provide the structure of it. The other option is that I keep that stuff on a local LLM that's not even connected to the internet.

The more I use AI, the more I've learned that it's fairly easy to get rid of hallucinations by having better context and prompting. PRD files, <agent>.MD files, relevant tools, TODO lists and setting "memory"s within the agent can almost eliminate hallucinations. The biggest mistake I see our new coders making is feeding in too much context to solve a problem.... i.e. sending in a 200 page PDF, when the relevant section is all on two pages. It's a bit like lawyers burying the other side's lawyers with irrelevant information during discovery.

Unfortunately that hasn't been my experience. I've asked some straightforward questions and the AIs I have access too often get it wrong. I mean "Provide me with the URL from Microsoft describing what event ID 4825 is" (as an example) often provides me with links to reddit and other sites rather than learn.microsoft.com. If I were asking for the AI to explain to me what event ID 4825 means, sure, that might make sense, but I'm just asking for the source material so I can figure it out how it fits into my investigation (BTW, no clue what event ID 4825 is, just used it as an example)

If the AI can't answer questions on publicly available information, I'm not going to trust it with private, confidential information.