I build AI products at scale. I share what it takes to be an AI PM. Only practical, real-world advice. No-fluff, no theory.

Joined November 2014
30,217 Photos and videos
Pinned Tweet
You could doom scroll Youtube today. Or you could build your own AI Assistant with Claude Code (in less than 3 hours, without writing a single line of code) I made a free course that teaches you Claude Code inside Claude Code. The course includes 6 core lessons, and takes only 3 hours. By the end of it, you will have built your own Second Brain from scratch which runs on 4 independent agents. Here is what you will learn ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿญ - ๐Ÿฎ๐Ÿฑ ๐—บ๐—ถ๐—ป: ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป ๐—ต๐—ผ๐˜„ ๐˜๐—ผ ๐˜„๐—ฟ๐—ถ๐˜๐—ฒ ๐—ฝ๐—ฟ๐—ผ๐—บ๐—ฝ๐˜๐˜€ ๐˜๐—ต๐—ฎ๐˜ ๐—ฎ๐—ฐ๐˜๐˜‚๐—ฎ๐—น๐—น๐˜† ๐˜„๐—ผ๐—ฟ๐—ธ. You learn the 5 components of a prompt that gets you a good answer the first time, every time. This is the skill every other lesson is built on. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฎ - ๐Ÿฏ๐Ÿฌ ๐—บ๐—ถ๐—ป: ๐—•๐˜‚๐—ถ๐—น๐—ฑ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ณ๐—ถ๐—ฟ๐˜€๐˜ ๐—ฆ๐—ธ๐—ถ๐—น๐—น. Once a prompt works, you want to use it again without typing it out. You learn how to turn it into a Skill that Claude picks up on its own when you ask in plain English. Build multiple skills and then see how Claude routes your query to the right one automatically. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฏ - ๐Ÿฏ๐Ÿฑ ๐—บ๐—ถ๐—ป: ๐—•๐˜‚๐—ถ๐—น๐—ฑ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ณ๐—ถ๐—ฟ๐˜€๐˜ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜. A Skill runs when you ask. An agent runs on its own to hit a ๐˜จ๐˜ฐ๐˜ข๐˜ญ you set. You learn the four parts every agent is made of. Brain, goal, tools, memory. And then you build your very first AI agent end to end. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฐ - ๐Ÿฐ๐Ÿฑ ๐—บ๐—ถ๐—ป: ๐—–๐—ผ๐—ป๐—ป๐—ฒ๐—ฐ๐˜ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—”๐—œ ๐—”๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ผ ๐˜๐—ต๐—ฒ ๐—ฟ๐—ฒ๐—ฎ๐—น ๐˜„๐—ผ๐—ฟ๐—น๐—ฑ. An AI agent that is limited to a prompt can only do a few things. In this module, you learn how to give it real tools, so it can take real actions in real apps. You will connect your agent to Google Calendar, so it can create meeting invites, reminders, events for you when needed. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฑ - ๐Ÿฑ๐Ÿฌ ๐—บ๐—ถ๐—ป: ๐—•๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ ๐˜๐—ฒ๐—ฎ๐—บ ๐—ผ๐—ณ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜๐˜€. One agent doing everything gets slow and messy fast. You learn how to create AI agent teams that run your entier system. One coordinator/CEO gives orders. Specialists do the work in parallel. You watch four of them work at once on a live dashboard. (Exactly what you see in the visual below.) ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฒ - ๐Ÿฐ๐Ÿฌ ๐—บ๐—ถ๐—ป: ๐— ๐—ฎ๐—ธ๐—ฒ ๐—ถ๐˜ ๐—ณ๐—ฎ๐˜€๐˜๐—ฒ๐—ฟ ๐—ฎ๐—ป๐—ฑ/๐—ผ๐—ฟ ๐—ฐ๐—ต๐—ฒ๐—ฎ๐—ฝ๐—ฒ๐—ฟ. Every AI system has the same three constraints: quality, latency, cost. You learn which one to turn for which problem, and watch the speed and cost change on screen as you do. The course is 100% FREE. But not for long. It includes everything I know about Claude Code and you can learn it in 3 hours for Free. Start today. Link in comments
3
1
15
2,261
JustAnotherPM | Sid retweeted
This is a $1,000 course. I'm giving it away free. I'll tell you why (because it's important.) Keep reading. A year ago, I needed an AI executive assistant for myself. Not a chatbot. Something that ran on my machine, prepped my mornings, and managed my calendar without me micromanaging it. I couldn't find one. So I built it. I called it a Daily Brain. The first version took me 4 weeks. Most of those 4 weeks went to one thing: my first prompt. Not getting it to work in a sandbox. Getting it to work on MY machine. Once I got past it, every agent I built came in days, not weeks. That's when I realised something. The steps I took to get my first prompt working are universal. Every builder needs them. So I turned the build into a 6-module course. 3.5 hours. No coding knowledge required. What you build: (Quick note: the course is built inside Claude Code. The only way you take the course is by building inside Claude Code. 100% focused on BUILDING) ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿญ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ ๐—บ๐—ฒ๐—ฒ๐˜๐—ถ๐—ป๐—ด-๐—ป๐—ผ๐˜๐—ฒ๐˜€ ๐˜€๐˜‚๐—บ๐—บ๐—ฎ๐—ฟ๐—ถ๐˜€๐—ฒ๐—ฟ You build a tool that reads your meeting notes and writes the summary. You learn the 5-part prompt shape. 99% of prompts fail because they skip 3 of the 5 parts. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฎ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐˜๐˜„๐—ผ "๐—ฆ๐—ธ๐—ถ๐—น๐—น๐˜€" A summariser and one of your choice. You learn how Claude routes between Skills based on what you say. No commands. No menus. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฏ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ ๐—ณ๐˜‚๐—น๐—น๐˜† ๐—ถ๐—ป๐—ฑ๐—ฒ๐—ฝ๐—ฒ๐—ป๐—ฑ๐—ฒ๐—ป๐˜ ๐—ฑ๐—ฎ๐—ถ๐—น๐˜†-๐—ฏ๐—ฟ๐—ถ๐—ฒ๐—ณ๐—ถ๐—ป๐—ด ๐—ฎ๐—ด๐—ฒ๐—ป๐˜ The agent that reads your work and tells you what to do next. You learn the four parts every working agent has: Brain, Goal, Tools, Memory. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฐ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ๐—ป ๐—ฎ๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ต๐—ฎ๐˜ ๐—บ๐—ฎ๐—ป๐—ฎ๐—ด๐—ฒ๐˜€ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ฐ๐—ฎ๐—น๐—ฒ๐—ป๐—ฑ๐—ฎ๐—ฟ You build an agent that connects to Google Calendar and books meetings for you based on the details from your emails and tasks. You learn MCP โ€” the protocol that connects AI to any app. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฑ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ๐—ป ๐—ผ๐—ฟ๐—ฐ๐—ต๐—ฒ๐˜€๐˜๐—ฟ๐—ฎ๐˜๐—ผ๐—ฟ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ต๐—ฎ๐˜ ๐—ฐ๐—ผ๐—ป๐˜๐—ฟ๐—ผ๐—น๐˜€ ๐˜„๐—ถ๐˜๐—ต ๐˜๐—ต๐—ฟ๐—ฒ๐—ฒ ๐˜€๐—ฝ๐—ฒ๐—ฐ๐—ถ๐—ฎ๐—น๐—ถ๐˜€๐˜๐˜€ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜๐˜€ You build a system where one coordinator runs three specialists in parallel โ€” notes, calendar, follow-ups. You learn the orchestration pattern every production AI system runs on. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฒ: ๐—ฌ๐—ผ๐˜‚ ๐—น๐—ฒ๐—ฎ๐—ฟ๐—ป ๐—ต๐—ผ๐˜„ ๐˜๐—ผ ๐˜๐˜‚๐—ป๐—ฒ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—”๐—œ ๐˜€๐˜†๐˜€๐˜๐—ฒ๐—บ You learn how to monitor and control quality, latency, and cost. You learn how to alter different parts of your product to control the three levers. What you keep after the course: The 5-part prompt shape. Agent anatomy. The chief-and-workers pattern. The three knobs. ๐—ช๐—ต๐˜† ๐—ถ๐˜€ ๐—ถ๐˜ ๐—ณ๐—ฟ๐—ฒ๐—ฒ? The hard part of learning to build is the first hour. After that the curve gets easy. I wanted to remove the friction. (Link in comments) 90 days from now, the PMs who took it are going to look very different from the PMs who saved it for later.
5
8
20
1,977
This is a $1,000 course. I'm giving it away free. I'll tell you why (because it's important.) Keep reading. A year ago, I needed an AI executive assistant for myself. Not a chatbot. Something that ran on my machine, prepped my mornings, and managed my calendar without me micromanaging it. I couldn't find one. So I built it. I called it a Daily Brain. The first version took me 4 weeks. Most of those 4 weeks went to one thing: my first prompt. Not getting it to work in a sandbox. Getting it to work on MY machine. Once I got past it, every agent I built came in days, not weeks. That's when I realised something. The steps I took to get my first prompt working are universal. Every builder needs them. So I turned the build into a 6-module course. 3.5 hours. No coding knowledge required. What you build: (Quick note: the course is built inside Claude Code. The only way you take the course is by building inside Claude Code. 100% focused on BUILDING) ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿญ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ ๐—บ๐—ฒ๐—ฒ๐˜๐—ถ๐—ป๐—ด-๐—ป๐—ผ๐˜๐—ฒ๐˜€ ๐˜€๐˜‚๐—บ๐—บ๐—ฎ๐—ฟ๐—ถ๐˜€๐—ฒ๐—ฟ You build a tool that reads your meeting notes and writes the summary. You learn the 5-part prompt shape. 99% of prompts fail because they skip 3 of the 5 parts. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฎ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐˜๐˜„๐—ผ "๐—ฆ๐—ธ๐—ถ๐—น๐—น๐˜€" A summariser and one of your choice. You learn how Claude routes between Skills based on what you say. No commands. No menus. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฏ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ ๐—ณ๐˜‚๐—น๐—น๐˜† ๐—ถ๐—ป๐—ฑ๐—ฒ๐—ฝ๐—ฒ๐—ป๐—ฑ๐—ฒ๐—ป๐˜ ๐—ฑ๐—ฎ๐—ถ๐—น๐˜†-๐—ฏ๐—ฟ๐—ถ๐—ฒ๐—ณ๐—ถ๐—ป๐—ด ๐—ฎ๐—ด๐—ฒ๐—ป๐˜ The agent that reads your work and tells you what to do next. You learn the four parts every working agent has: Brain, Goal, Tools, Memory. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฐ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ๐—ป ๐—ฎ๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ต๐—ฎ๐˜ ๐—บ๐—ฎ๐—ป๐—ฎ๐—ด๐—ฒ๐˜€ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ฐ๐—ฎ๐—น๐—ฒ๐—ป๐—ฑ๐—ฎ๐—ฟ You build an agent that connects to Google Calendar and books meetings for you based on the details from your emails and tasks. You learn MCP โ€” the protocol that connects AI to any app. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฑ: ๐—ฌ๐—ผ๐˜‚ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ๐—ป ๐—ผ๐—ฟ๐—ฐ๐—ต๐—ฒ๐˜€๐˜๐—ฟ๐—ฎ๐˜๐—ผ๐—ฟ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ต๐—ฎ๐˜ ๐—ฐ๐—ผ๐—ป๐˜๐—ฟ๐—ผ๐—น๐˜€ ๐˜„๐—ถ๐˜๐—ต ๐˜๐—ต๐—ฟ๐—ฒ๐—ฒ ๐˜€๐—ฝ๐—ฒ๐—ฐ๐—ถ๐—ฎ๐—น๐—ถ๐˜€๐˜๐˜€ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜๐˜€ You build a system where one coordinator runs three specialists in parallel โ€” notes, calendar, follow-ups. You learn the orchestration pattern every production AI system runs on. ๐— ๐—ผ๐—ฑ๐˜‚๐—น๐—ฒ ๐Ÿฒ: ๐—ฌ๐—ผ๐˜‚ ๐—น๐—ฒ๐—ฎ๐—ฟ๐—ป ๐—ต๐—ผ๐˜„ ๐˜๐—ผ ๐˜๐˜‚๐—ป๐—ฒ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—”๐—œ ๐˜€๐˜†๐˜€๐˜๐—ฒ๐—บ You learn how to monitor and control quality, latency, and cost. You learn how to alter different parts of your product to control the three levers. What you keep after the course: The 5-part prompt shape. Agent anatomy. The chief-and-workers pattern. The three knobs. ๐—ช๐—ต๐˜† ๐—ถ๐˜€ ๐—ถ๐˜ ๐—ณ๐—ฟ๐—ฒ๐—ฒ? The hard part of learning to build is the first hour. After that the curve gets easy. I wanted to remove the friction. (Link in comments) 90 days from now, the PMs who took it are going to look very different from the PMs who saved it for later.
5
8
20
1,977
Most PMs spend their careers describing software for someone else to build. A few stopped. They learned Claude Code, and now they build it themselves. One shipped a working AI second brain in a weekend. No engineer. No coding knowledge You can build yours too. In 30 mins. Learn Claude Code inside Claude Code.
2
1
14
1,849
Every few years, one tool separates the PMs who get promoted from the ones who get stuck. 2014 it was SQL. 2019 it was Figma. 2026 it's Claude Code. The ones who learn it early move up early. You can learn it for FREE. 6 lessons. You build your own AI second brain by the end. (link in comments)
1
3
10
1,520
The slowest part of being a PM was never the thinking. It was the waiting. Wait for the sprint. Wait for priority. Wait two quarters to learn the idea was wrong. The PMs moving fastest stopped waiting. They build it themselves with Claude Code. Learn Claude Code for FREE. Start building on your own today ๐Ÿ‘‡
2
1
7
1,343
Include these 5 components in your next Claude prompt and see how your world transforms. 1. ๐—ฅ๐—ผ๐—น๐—ฒ: who is the AI acting as? โ€‡ย ย This helps set perspective. Enable the LLM know what type of language to use -- a senior PM talks differently compared to a senior sales executive 2. ๐—–๐—ผ๐—ป๐˜๐—ฒ๐˜…๐˜: what extra information does the model have? โ€‡ย ย The more context the model has the higher the chances it will create responses that are relevant and accurate. 3. ๐—ง๐—ฎ๐˜€๐—ธ: what should the model do? โ€‡ย ย Give clear instructions, so the model knows exactly what it needs to do. "Write an email" is not specific. "Write a 200 word launch announcement, short sentences, bullet points, clear, 5th grade english...:" is specific. 4. ๐—–๐—ผ๐—ป๐˜€๐˜๐—ฟ๐—ฎ๐—ถ๐—ป๐˜๐˜€: what boundsaring should the model know about? โ€‡ย ย This will help the model focus on the right things (and stay away from the wrong things.) Without constraints, the model will make assumptions. Not all of them will be right. 5. ๐—˜๐˜…๐—ฎ๐—บ๐—ฝ๐—น๐—ฒ๐˜€: what does good output look like? โ€‡ย ย Tell the model what good, bad, ugly looks like, so it has a benchmark. Need help getting started?Sign up for my free course that teaches Claude Code inside Claude Code. Module 1 shows you exactly how to create en effective prompt with all 5 components. (Link in comments)
3
4
8
2,060
Are there any online / IRL builder or vibe coding communities in London?
2
420
You could doom scroll Youtube today. Or you could build your own AI Assistant with Claude Code (in less than 3 hours, without writing a single line of code) I made a free course that teaches you Claude Code inside Claude Code. The course includes 6 core lessons, and takes only 3 hours. By the end of it, you will have built your own Second Brain from scratch which runs on 4 independent agents. Here is what you will learn ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿญ - ๐Ÿฎ๐Ÿฑ ๐—บ๐—ถ๐—ป: ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป ๐—ต๐—ผ๐˜„ ๐˜๐—ผ ๐˜„๐—ฟ๐—ถ๐˜๐—ฒ ๐—ฝ๐—ฟ๐—ผ๐—บ๐—ฝ๐˜๐˜€ ๐˜๐—ต๐—ฎ๐˜ ๐—ฎ๐—ฐ๐˜๐˜‚๐—ฎ๐—น๐—น๐˜† ๐˜„๐—ผ๐—ฟ๐—ธ. You learn the 5 components of a prompt that gets you a good answer the first time, every time. This is the skill every other lesson is built on. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฎ - ๐Ÿฏ๐Ÿฌ ๐—บ๐—ถ๐—ป: ๐—•๐˜‚๐—ถ๐—น๐—ฑ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ณ๐—ถ๐—ฟ๐˜€๐˜ ๐—ฆ๐—ธ๐—ถ๐—น๐—น. Once a prompt works, you want to use it again without typing it out. You learn how to turn it into a Skill that Claude picks up on its own when you ask in plain English. Build multiple skills and then see how Claude routes your query to the right one automatically. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฏ - ๐Ÿฏ๐Ÿฑ ๐—บ๐—ถ๐—ป: ๐—•๐˜‚๐—ถ๐—น๐—ฑ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ณ๐—ถ๐—ฟ๐˜€๐˜ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜. A Skill runs when you ask. An agent runs on its own to hit a ๐˜จ๐˜ฐ๐˜ข๐˜ญ you set. You learn the four parts every agent is made of. Brain, goal, tools, memory. And then you build your very first AI agent end to end. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฐ - ๐Ÿฐ๐Ÿฑ ๐—บ๐—ถ๐—ป: ๐—–๐—ผ๐—ป๐—ป๐—ฒ๐—ฐ๐˜ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—”๐—œ ๐—”๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ผ ๐˜๐—ต๐—ฒ ๐—ฟ๐—ฒ๐—ฎ๐—น ๐˜„๐—ผ๐—ฟ๐—น๐—ฑ. An AI agent that is limited to a prompt can only do a few things. In this module, you learn how to give it real tools, so it can take real actions in real apps. You will connect your agent to Google Calendar, so it can create meeting invites, reminders, events for you when needed. ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฑ - ๐Ÿฑ๐Ÿฌ ๐—บ๐—ถ๐—ป: ๐—•๐˜‚๐—ถ๐—น๐—ฑ ๐—ฎ ๐˜๐—ฒ๐—ฎ๐—บ ๐—ผ๐—ณ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜๐˜€. One agent doing everything gets slow and messy fast. You learn how to create AI agent teams that run your entier system. One coordinator/CEO gives orders. Specialists do the work in parallel. You watch four of them work at once on a live dashboard. (Exactly what you see in the visual below.) ๐—Ÿ๐—ฒ๐˜€๐˜€๐—ผ๐—ป ๐Ÿฒ - ๐Ÿฐ๐Ÿฌ ๐—บ๐—ถ๐—ป: ๐— ๐—ฎ๐—ธ๐—ฒ ๐—ถ๐˜ ๐—ณ๐—ฎ๐˜€๐˜๐—ฒ๐—ฟ ๐—ฎ๐—ป๐—ฑ/๐—ผ๐—ฟ ๐—ฐ๐—ต๐—ฒ๐—ฎ๐—ฝ๐—ฒ๐—ฟ. Every AI system has the same three constraints: quality, latency, cost. You learn which one to turn for which problem, and watch the speed and cost change on screen as you do. The course is 100% FREE. But not for long. It includes everything I know about Claude Code and you can learn it in 3 hours for Free. Start today. Link in comments
3
1
15
2,261
"I built this over the weekend" is only half the story. The real story is "I built this over the weekend, and abandoned it on Monday (because it was shit.)" Building is 10x easier. Launching is 20x tougher. Getting adoption is 100x tougher. Here's why that last one is brutal. Your customer already opens the same eight or nine apps every day. That list almost never changes. For yours to get in, one of theirs has to get pushed out. So building more products doesn't create more room to use them. The day is still 24 hours. The home screen is still full. Most people build anyway, without ever checking whether anyone had the problem in the first place.
2
5
1,094
JustAnotherPM | Sid retweeted
Jun 3
You basically need to be unemployed to keep up with all this AI stuff.
637
1,044
12,942
455,593
Don't pay $100/month for the Claude Max plan. Use these 7 hacks to protect your tokens (and your money): 1. When you keep ๐—บ๐—ฎ๐—ป๐˜† ๐˜๐—ผ๐—ฝ๐—ถ๐—ฐ๐˜€ ๐—ถ๐—ป ๐—ผ๐—ป๐—ฒ ๐—ฐ๐—ต๐—ฎ๐˜, Claude rereads all of them every time you send a message, even the ones you are done with. ๐—™๐—ถ๐˜…: Start a new chat for each topic. When one gets long, ask for a short summary, copy it, and paste it into a fresh chat. ๐Ÿฎ. ๐—ช๐—ต๐—ฒ๐—ป ๐˜†๐—ผ๐˜‚ ๐˜‚๐—ฝ๐—น๐—ผ๐—ฎ๐—ฑ ๐—ฎ ๐—ฃ๐——๐—™, Claude pays for it twice. Once to read the words, and again because it turns every page into an image and reads that too. Fix: Copy the text out, paste it into a plain document, and upload that. Same words, a fraction of the cost. 3. Claude has three models.: A fast cheap one, a balanced one, and a heavy expensive one. Most people leave it on the heavy one for everything, even fixing typos. Fix: Use the cheap one for small tasks and the heavy one for hard problems. 4. When Claude gets something wrong, you probably say "no, I meant this." That reply stacks on top of everything above it, and Claude rereads the whole pile to answer. Fix: scroll up, edit your original message, and run it again. 5. If you upload the ๐˜€๐—ฎ๐—บ๐—ฒ ๐—ณ๐—ถ๐—น๐—ฒ ๐˜๐—ผ ๐—ณ๐—ถ๐˜ƒ๐—ฒ ๐—ฑ๐—ถ๐—ณ๐—ณ๐—ฒ๐—ฟ๐—ฒ๐—ป๐˜ ๐—ฐ๐—ต๐—ฎ๐˜๐˜€, you pay to read it five times. Fix: Put it in a Project once instead. Every chat inside that Project can use the file without uploading it again. 6. You explain everything about you and your work repeatedly. Fix: Open Settings, then Capabilities, and tell Claude who you are, what you work on, and how you like answers. It carries that into every new chat, so you stop burning the first few messages re-explaining yourself. 7. Asking Claude to make a doc, sheet, or slide costs more than a normal answer. You build one, scrap it, and build again. And pay the cost for every iteration. Fix: Get the plan right in a plain chat first, then build it once. I changed those seven things and never hit the limit again. Same usage. Saving tokens is the easy part. The real shift is when you ๐˜€๐˜๐—ฎ๐—ฟ๐˜ ๐—ฏ๐˜‚๐—ถ๐—น๐—ฑ๐—ถ๐—ป๐—ด ๐˜„๐—ถ๐˜๐—ต ๐—–๐—น๐—ฎ๐˜‚๐—ฑ๐—ฒ. I made a free course that does exactly that. You learn Claude Code inside Claude Code. 6 Modules: 1. How to write effective prompts 2. How to build a skill 3. How to build an Agent 4. How to connect apps via MCP 5. How to build a second brain 6. How to optimize cost, latency, quality It takes 5 hours to build all of this. Completely FREE. (link in comments)
5
1
17
1,890