Max aka Mosheh - AI with Apex
@aiwithapex.bsky.social
Hey everyone, I'm Max aka Mosheh 👋 Started ‘AAA’ work in June ’24 (AIwithApex.com). Father of 5, still working a day job. Ex-CTO, now Board Pres at RedeemTheOppressed.org. Built ventures, love Voice AI, 20+ yrs in tech. Here to learn, share, and build.
AI isn’t a black box.
Your words → tokens → vectors → attention → next-word bets.
Once you see it as structured focus at scale, you can redesign workflows, roles, and products around it.
What’s the first process you’d rebuild with this mental model?
Your words → tokens → vectors → attention → next-word bets.
Once you see it as structured focus at scale, you can redesign workflows, roles, and products around it.
What’s the first process you’d rebuild with this mental model?
December 20, 2025 at 5:50 AM
AI isn’t a black box.
Your words → tokens → vectors → attention → next-word bets.
Once you see it as structured focus at scale, you can redesign workflows, roles, and products around it.
What’s the first process you’d rebuild with this mental model?
Your words → tokens → vectors → attention → next-word bets.
Once you see it as structured focus at scale, you can redesign workflows, roles, and products around it.
What’s the first process you’d rebuild with this mental model?
AI didn’t fail in 2025.
Old workflows did.
Chatbots on top of bad data, clunky processes, and silos don’t create value.
AI only works when you redesign work end-to-end: clean data, modern infra, new roles, fewer handoffs.
Are you shipping bots—or fixing workflows?
Old workflows did.
Chatbots on top of bad data, clunky processes, and silos don’t create value.
AI only works when you redesign work end-to-end: clean data, modern infra, new roles, fewer handoffs.
Are you shipping bots—or fixing workflows?
December 18, 2025 at 5:50 AM
AI didn’t fail in 2025.
Old workflows did.
Chatbots on top of bad data, clunky processes, and silos don’t create value.
AI only works when you redesign work end-to-end: clean data, modern infra, new roles, fewer handoffs.
Are you shipping bots—or fixing workflows?
Old workflows did.
Chatbots on top of bad data, clunky processes, and silos don’t create value.
AI only works when you redesign work end-to-end: clean data, modern infra, new roles, fewer handoffs.
Are you shipping bots—or fixing workflows?
Old-school “open another app” workflows are dying.
Adobe just moved Photoshop, Express, and Acrobat into ChatGPT.
You describe the edit.
AI does the heavy lifting.
The real advantage isn’t tools—it’s how fast your team moves when tools disappear into the chat.
Adobe just moved Photoshop, Express, and Acrobat into ChatGPT.
You describe the edit.
AI does the heavy lifting.
The real advantage isn’t tools—it’s how fast your team moves when tools disappear into the chat.
December 16, 2025 at 5:50 AM
Old-school “open another app” workflows are dying.
Adobe just moved Photoshop, Express, and Acrobat into ChatGPT.
You describe the edit.
AI does the heavy lifting.
The real advantage isn’t tools—it’s how fast your team moves when tools disappear into the chat.
Adobe just moved Photoshop, Express, and Acrobat into ChatGPT.
You describe the edit.
AI does the heavy lifting.
The real advantage isn’t tools—it’s how fast your team moves when tools disappear into the chat.
You don’t need cloud GPUs to build with AI.
WebLLM, Free LLM Playground, BrowserAI, AgentLLM, and Genspark.ai let you run and compare models right in your browser.
No servers. No installs. No credit card.
What’s stopping you from trying browser-first AI this week?
WebLLM, Free LLM Playground, BrowserAI, AgentLLM, and Genspark.ai let you run and compare models right in your browser.
No servers. No installs. No credit card.
What’s stopping you from trying browser-first AI this week?
December 14, 2025 at 5:57 AM
You don’t need cloud GPUs to build with AI.
WebLLM, Free LLM Playground, BrowserAI, AgentLLM, and Genspark.ai let you run and compare models right in your browser.
No servers. No installs. No credit card.
What’s stopping you from trying browser-first AI this week?
WebLLM, Free LLM Playground, BrowserAI, AgentLLM, and Genspark.ai let you run and compare models right in your browser.
No servers. No installs. No credit card.
What’s stopping you from trying browser-first AI this week?
Vibe coding is fun… until you realize you’re just copy‑pasting code you don’t understand.
AI can write code.
Your edge is thinking.
Shape messy requirements.
Understand data flows.
Debug production.
Design architectures.
Make tradeoffs.
The future belongs to devs who think, not just type.
AI can write code.
Your edge is thinking.
Shape messy requirements.
Understand data flows.
Debug production.
Design architectures.
Make tradeoffs.
The future belongs to devs who think, not just type.
December 12, 2025 at 5:56 AM
Vibe coding is fun… until you realize you’re just copy‑pasting code you don’t understand.
AI can write code.
Your edge is thinking.
Shape messy requirements.
Understand data flows.
Debug production.
Design architectures.
Make tradeoffs.
The future belongs to devs who think, not just type.
AI can write code.
Your edge is thinking.
Shape messy requirements.
Understand data flows.
Debug production.
Design architectures.
Make tradeoffs.
The future belongs to devs who think, not just type.
Old school automation is dying. “Computer-use” AI is next.
Lux doesn’t just read the web. It clicks, types, and scrolls through real desktops at ~1s per action.
Think QA tests, forms, dashboards.
Trust is the bottleneck. Control is the unlock.
What would you hand off first?
Lux doesn’t just read the web. It clicks, types, and scrolls through real desktops at ~1s per action.
Think QA tests, forms, dashboards.
Trust is the bottleneck. Control is the unlock.
What would you hand off first?
December 10, 2025 at 5:57 AM
Old school automation is dying. “Computer-use” AI is next.
Lux doesn’t just read the web. It clicks, types, and scrolls through real desktops at ~1s per action.
Think QA tests, forms, dashboards.
Trust is the bottleneck. Control is the unlock.
What would you hand off first?
Lux doesn’t just read the web. It clicks, types, and scrolls through real desktops at ~1s per action.
Think QA tests, forms, dashboards.
Trust is the bottleneck. Control is the unlock.
What would you hand off first?
Most AI reads more.
The next wave will remember better.
Apple’s CLaRa compresses documents up to 128x into “memory tokens,” then retrieves and reasons in that tiny space—often matching or beating full-text RAG.
The future edge isn’t bigger models.
It’s smarter memory.
The next wave will remember better.
Apple’s CLaRa compresses documents up to 128x into “memory tokens,” then retrieves and reasons in that tiny space—often matching or beating full-text RAG.
The future edge isn’t bigger models.
It’s smarter memory.
December 8, 2025 at 5:57 AM
Most AI reads more.
The next wave will remember better.
Apple’s CLaRa compresses documents up to 128x into “memory tokens,” then retrieves and reasons in that tiny space—often matching or beating full-text RAG.
The future edge isn’t bigger models.
It’s smarter memory.
The next wave will remember better.
Apple’s CLaRa compresses documents up to 128x into “memory tokens,” then retrieves and reasons in that tiny space—often matching or beating full-text RAG.
The future edge isn’t bigger models.
It’s smarter memory.
Prompt engineering isn’t the real AI power move anymore. Context engineering is. It’s not about better sentences, it’s about better environments: profiles, memory, live data, history. The age of prompts is fading. The age of engineered AI worlds is here. Are you ready?
December 6, 2025 at 5:57 AM
Prompt engineering isn’t the real AI power move anymore. Context engineering is. It’s not about better sentences, it’s about better environments: profiles, memory, live data, history. The age of prompts is fading. The age of engineered AI worlds is here. Are you ready?
Prompt hacks are dying.
The real AI unlock isn’t “better prompts” – it’s better context.
Layered user profiles.
Live knowledge.
Selective memory.
Stop chasing magic spells.
Start designing the world your AI thinks inside.
The real AI unlock isn’t “better prompts” – it’s better context.
Layered user profiles.
Live knowledge.
Selective memory.
Stop chasing magic spells.
Start designing the world your AI thinks inside.
December 4, 2025 at 5:57 AM
Prompt hacks are dying.
The real AI unlock isn’t “better prompts” – it’s better context.
Layered user profiles.
Live knowledge.
Selective memory.
Stop chasing magic spells.
Start designing the world your AI thinks inside.
The real AI unlock isn’t “better prompts” – it’s better context.
Layered user profiles.
Live knowledge.
Selective memory.
Stop chasing magic spells.
Start designing the world your AI thinks inside.
An AI just scored 118/120 on the Putnam, but that’s not the real story.
It learned to write proofs, check its own logic, and fix gaps.
The edge won’t be who has AI, but who learns to reason *with* it.
Are you using AI as a calculator—or as a junior thinker with a verifier?
It learned to write proofs, check its own logic, and fix gaps.
The edge won’t be who has AI, but who learns to reason *with* it.
Are you using AI as a calculator—or as a junior thinker with a verifier?
December 2, 2025 at 5:57 AM
An AI just scored 118/120 on the Putnam, but that’s not the real story.
It learned to write proofs, check its own logic, and fix gaps.
The edge won’t be who has AI, but who learns to reason *with* it.
Are you using AI as a calculator—or as a junior thinker with a verifier?
It learned to write proofs, check its own logic, and fix gaps.
The edge won’t be who has AI, but who learns to reason *with* it.
Are you using AI as a calculator—or as a junior thinker with a verifier?
AI agent success isn’t uptime or speed. It’s goal accuracy, workflow adherence, and low hallucinations—tied to cost per outcome. Example: 62%→86% accuracy, 43% fewer hallucinations, $5.40→$3.10 per ticket. Use dashboards, audits, and daily tracking. What’s your take?
November 30, 2025 at 6:05 AM
AI agent success isn’t uptime or speed. It’s goal accuracy, workflow adherence, and low hallucinations—tied to cost per outcome. Example: 62%→86% accuracy, 43% fewer hallucinations, $5.40→$3.10 per ticket. Use dashboards, audits, and daily tracking. What’s your take?
I tested Perplexity’s Comet. For static pages, it’s a rocket: fast summaries, cross-tab extraction, instant tables. On dynamic dashboards, it stalls. Use it as your research copilot, not a one-browser-fits-all. Want my 3-step playbook?
November 28, 2025 at 6:05 AM
I tested Perplexity’s Comet. For static pages, it’s a rocket: fast summaries, cross-tab extraction, instant tables. On dynamic dashboards, it stalls. Use it as your research copilot, not a one-browser-fits-all. Want my 3-step playbook?
Ever wished therapy felt more like an epic adventure? 🏹🎮 Enter Luminari’s Quest: where mental health meets RPG magic. Level up your emotional skills, not just your avatar! Ready to play for your wellbeing? 👉 #TherapeuticGaming #LevelUpLife
June 30, 2025 at 8:23 AM
Ever wished therapy felt more like an epic adventure? 🏹🎮 Enter Luminari’s Quest: where mental health meets RPG magic. Level up your emotional skills, not just your avatar! Ready to play for your wellbeing? 👉 #TherapeuticGaming #LevelUpLife
🚨 Final call: What if you woke up tomorrow knowing you missed $1M+ & a shot at AI stardom? There’s still time (but not much) to submit to the #WorldsLargestHackathon. Build anything, launch fast, impress legends. Deadline: 2pm PT! 👉 worldslargesthackathon.devpost.com #AI
June 29, 2025 at 1:14 PM
🚨 Final call: What if you woke up tomorrow knowing you missed $1M+ & a shot at AI stardom? There’s still time (but not much) to submit to the #WorldsLargestHackathon. Build anything, launch fast, impress legends. Deadline: 2pm PT! 👉 worldslargesthackathon.devpost.com #AI
🚀 Builder universe is on FIRE: Bolt.new’s $1M sprint vs Lovable.dev’s founder-friendly marathon! 🏃♂️💡 Flashy launches or sustainable shipping—which side are you on? Drop your pick! 👇 #BuildInPublic #AIHackathon
June 20, 2025 at 1:36 PM
🚀 Builder universe is on FIRE: Bolt.new’s $1M sprint vs Lovable.dev’s founder-friendly marathon! 🏃♂️💡 Flashy launches or sustainable shipping—which side are you on? Drop your pick! 👇 #BuildInPublic #AIHackathon
What you see is the back end of Motivate-Mate.com
This SaaS, engineered for a 6-week hackathon, has made me over $8k in sales and won me over $3k in prizes.
I sell the boot-camp course for this solution for $800 -- and I'm giving away a copy for FREE.
Enter here: course-giveaway.paperform.co
This SaaS, engineered for a 6-week hackathon, has made me over $8k in sales and won me over $3k in prizes.
I sell the boot-camp course for this solution for $800 -- and I'm giving away a copy for FREE.
Enter here: course-giveaway.paperform.co
May 26, 2025 at 5:28 AM
What you see is the back end of Motivate-Mate.com
This SaaS, engineered for a 6-week hackathon, has made me over $8k in sales and won me over $3k in prizes.
I sell the boot-camp course for this solution for $800 -- and I'm giving away a copy for FREE.
Enter here: course-giveaway.paperform.co
This SaaS, engineered for a 6-week hackathon, has made me over $8k in sales and won me over $3k in prizes.
I sell the boot-camp course for this solution for $800 -- and I'm giving away a copy for FREE.
Enter here: course-giveaway.paperform.co
Forget coding bootcamps—2025 is all about ✨vibe coding✨. Build AI-powered apps just by typing what you want! Are no-code platforms like @adaptive_computer’s ac1 the future of creativity—or the end of traditional dev? Ready to build, or be built over? #NoCode #AIRevolution
April 29, 2025 at 5:33 AM
Forget coding bootcamps—2025 is all about ✨vibe coding✨. Build AI-powered apps just by typing what you want! Are no-code platforms like @adaptive_computer’s ac1 the future of creativity—or the end of traditional dev? Ready to build, or be built over? #NoCode #AIRevolution
Zapier, n8n, Make, Airtable—you're just scratching the surface. These 5 underrated AI workflows can save your team hours daily. From auto-meeting summaries to smart form routing, it's time to level up. 🧠⚡️ #AIautomation #nocode #productivity
April 25, 2025 at 2:16 PM
Zapier, n8n, Make, Airtable—you're just scratching the surface. These 5 underrated AI workflows can save your team hours daily. From auto-meeting summaries to smart form routing, it's time to level up. 🧠⚡️ #AIautomation #nocode #productivity