How I AI

"Maybe we have some design in Figma. Before you could just paste a link to a Figma URL and try and manually invoke the Figma MCP."

"Before you could just paste a link to a Figma URL and try and manually invoke the Figma MCP. It would build it and then sort of stop halfway through."

"Let me go over to Slack and you can see me interacting a little bit with my OpenClaw."

"I just take a picture of a snippet of that book and then I ask it to use Nano Banana Pro and I gave it the API key for that specific image generation model to my OpenClaw."

"I ask it to use Nano Banana Pro and I gave it the API key for that specific image generation model to my OpenClaw."

"I wrote a little skill called find icon. And the skill basically says anytime you're going to implement an icon, first go and actually look through the whole project"

"OpenClaw finally allows me to get to because it's actually doing the heavy lifting... I want to be able to take photos of a lesson that I do and then just upload them and have the actual OpenClaw log the full lesson contents."

"So what Claude has is called slash commands. It's basically a glorified prompt. It has a name, a description, and then some instructions."

"Usually by just telling Claude to do that."

"The difference was we were all creating our own repository, our own Next.js instance."

"this is going to go off send that off to OpenAI and come back with that"

"when in Slack in Chrome and hovering over a message that has focus"

"We want to build a very simple PRD here for a locally run Chrome extension where the job is when in Slack in Chrome"

"extract that content and send it over to OpenAI. When in OpenAI, we should summarize them and extract three to five key takeaways"

"So, I have a shortcut in Slack. If I hit control shift D on any message, it's going to pop up and go off and describe that image for me"

"And it will head off to ChatGPT and get the response for this as well"

"I've been building that up where if I'm working with this and something goes wrong, I would essentially hop into the place where I have these skills and say, 'Here's what happened. Give me a suggestion on how to improve this skill so this doesn't happen again.'"

"I want to be able to say, 'Okay, I want a new feature in Flowey. I'm going to build it. I'm going to update skills and I can be confident that Claude can actually work with that and understand the new feature.'"

"OpenAI released last week Codex, their desktop app for AI engineering, the new model GPT 5.3 Codex."

"the new model GPT 5.3 Codex. Try saying that five times fast."

"And Anthropic released their response, Opus 4.6 and Opus 4.6 Fast."

"I have to give kudos to Cursor. I'm still a Cursor girl. I think Cursor does a good job of building harnesses for all of these models."

"I asked Devon today, how much code have I merged into GitHub in the last five days?"

"I've played around with this tool to basically give Claude these JSON files. And there's a whole set of skills I've built around this that Claude code can use to write these out."

"Can you build software like this with v0? Software that's going to travel entire world, scale, receive a lot of traffic, but you have confidence that you can make high quality changes within v0 itself."

"The process itself of building this was very much prompt driven, I'll say, chatting in Slack and saying, hey, wouldn't it be nice if we had a hub for this"

"It takes advantage of all of the Vercel infrastructure primitives even though it has 35,000 of the skills"

"imagine triggering a task like this from your phone, from Slack, from v0.app"

"one table which is like a Google sheet type"

"I have our MCP helper chatbot transcripts. I have this end user facing chatbot."

"V0 is all about leveraging the integration and marketplace capabilities of our cell. It knows what the data source is of this project, it's going to go through the whole file system."

"We're storing data in Redis by AppStach."

"When you see this URL ending on vercel.sh, this is basically running on the production grade CDN on the production grade rendering infrastructure, hosting infrastructure."

"I don't know if your company uses Zapier. Sometimes they're like, oh yeah, we're both a customer and a partner, right?"

"We used Databricks, which houses like a lot of our data, makes that usable. And so they built this whole series of things that allows just simple lookup."

"For example, I use this Remotion best practices one to let me import components and regularly create videos really, really quickly."

"I have a Claude project that has a bunch of instructions on just how to log this data, where should it log, and then I can go in and actually select that project."

"You can tell an AI this is a wrapper around Gemini where it will execute Gemini with specific prompts."

"As I was on Telegram going back and forth giving it, can you add this? Can you remove this? Can you change the schedule? I thought it was doing a great job on Telegram"

"My goal with using Claudebot was to really see how it would work as an EA."

"Your ClawdBot will see it. It will have a token. So you give it a token and then it is up and running."

"To connect Telegram, you message botfather and you say create new bot and you give it a name and you give it a handle."

"You message BotFather and you say create new bot and you give it a name and you give it a handle."

"Claudebot is an open-source AI agent that you can install on a virtual machine or on a desktop or laptop that you have access to that is self-learning, can spin up sub agents using Claude code and other agent harnesses and can do in my lived experience a lot of damage."

"Hey Paulie, can you please join my Riverside FM podcast?"

"And so, I switched to Telegram, which I use for literally nothing. Um, because I'm an old lady mom, and set up a Telegram account."

"I chose sonnet 45. You can also kind of use cloud code with your own subscription or through API. I chose to use it through API because I wanted to see how much I was spending on cloudbot."

"So I use one password which is a password and secret sharing kind of app. I made a vault that's called Claude. Claude only has access. Claudebot only has access to that vault."

"So what I did and you can follow this if you want to from a security perspective although I think it has some drawbacks on the functionality of Cloudbot is I gave Cloudbot its own email address a Google Workspace email address."

"I'm running it on a MacBook Air that's sitting in on a shelf somewhere that I just picked up that no one was using."

"I use Devon which has a virtual machine in a local environment and can spin up stuff access to the web uh all the time. I use it from Slack so I can appmention Devon."

"Cursor has background agents. You know everything has you you codeex you can kick off online."

"I want you to document our conversation in a next.js JS web app that shows the back and forth of our full conversation from the very beginning today till the end in a UI."

"We are eventually going to deploy this to Verscell. Can you let me know when it's deployed to Versel so I can look at it?"

"if I type H, this will be haiku. It'll be much faster but not quite as smart"

"you can easily set up aliases for things like setting the default model for Claude"

"They have what are called mermaid diagrams. This is a way of visualizing database operations and it's a way of essentially compressing your application down into very small lines of text that show how your application works."