prizt

Started coding years ago. Lost momentum. Took detours into Graphic design, UI/UX and picked up other skills as well Now I’m back, focused on software engineering. Building in public.🚀

Tech in 2026: Every company is either building AI, Selling AI, Integrating AI, or explaining why they don’t need AI.

I’ll refractor it later

Founders: It’s Friday. Network day. Startups grow faster with the right people around you. Introduce yourself and network 🤝

@MrGreencoin @tobiflamme 😔😔

5M impressions isn't magic. 👀 Reply consistently. Show up daily. Engage in the right conversations. Do that for 30 days and you'll be surprised how far your content travels

@tobiflamme Let’s get it

Nigeria is one of the strongest communities on 𝕏 🇳🇬 Let’s keep showing up, supporting each other, getting monetized, and creating opportunities that change lives ❤️

Programming is actually 70% of thinking and 30% of writing code.

I still do, sometimes

It's pretty annoying that Codex uses agents.md and Claude Code uses Claude.md. There should be some industry standards to this stuff?

The speedup isn’t just in volume. On open-ended coding problems where answers are unclear, Claude’s success rate is now 76%—a 50 point jump in just 6 months. Many engineers also say Claude’s code quality is now on par with human code; we expect it to be better within the year.

Microsoft launched their own coding AI and it's FREE inside GitHub Copilot starting today 😳 not rented from OpenAI. built by Microsoft. it's called MAI-Code-1-Flash and it's rolling out to ALL copilot tiers right now part of microsoft's new MAI model family: - MAI-Code-1-Flash (coding) - MAI-Thinking-1 (reasoning, 35B params) - MAI-Image-2.5 (image gen) - MAI-Transcribe-1.5 (speech, 43 languages) who gets it: - free plan ✅ - pro plan ✅ - pro+ plan ✅ - max plan ✅ what makes it different: - trained inside copilot's production environment, not externally tested then shipped - 60% fewer tokens on complex tasks vs comparable models - 85.8% on microsoft adversarial coding benchmark - 256K context window full setup guide (github copilot + MAI-Code-1-Flash - free): step 1: create a free github account > go to github.com/signup > sign up for free step 2: enable github copilot > go to github.com/github-copilot… > select the free plan - no credit card needed step 3: install the VS Code extension > open VS Code > press ctrl+shift+x (windows) or cmd+shift+x (mac) > search "GitHub Copilot" > install the extension by GitHub step 4: sign in > VS Code will show a "Sign in to GitHub" prompt > click it, approve in the browser, return to VS Code step 5: open copilot chat > press ctrl+shift+i (windows) or cmd+shift+i (mac) > the chat panel opens on the right side step 6: select MAI-Code-1-Flash > click the model dropdown at the bottom of the chat panel > choose MAI-Code-1-Flash from the list > done already have copilot? skip straight to step 5 microsoft built this without openai data, free coding model wars are getting serious. bookmark this and start using this free model 🔖

Slyyyy annoying stuff

😭

claude code is fucking insane i know literally NOTHING about coding. ZERO. and i just built a fully functioning web app in minutes. http://localhost:3000/ check it out

@nestymee @acooldora i feel like the ui on claude code is worse than codex. super unorganized, cant do remote control via app unless on terminal etc.

How to give Claude perfect memory and stop context loss. One of the biggest issues with Claude is that it's a "black box." You feed it context, it forgets 80% of it, and the 20% it actually remembers is not relevant at all. I struggled with this for months, and only recently have I found the cure. Here's how to fix it: There are three layers- pick the one that fits you best. LAYER ONE: Basic fixes (takes 5 minutes) • Go to Settings → Memory and clean it out. Delete anything outdated or irrelevant, and keep only the context you actually want. • Fill in your Project Instructions for every project you use. You can voice-prompt your context into a Google Doc and upload it as a PDF. •Tell Claude directly mid-conversation: "Remember that I never want [x]" "Update your memory with [x]" This alone fixes 90% of memory problems. LAYER TWO: Context file system (60 mins to set up) Create a "Claude Master Folder" on your desktop with four markdown files: • Instructions.MD: your rules and preferences • Memory.MD: Claude's evolving brain (add this line: "Update Memory.MD with my preferences over time") • Context.MD: project-specific context • Archive: weekly backup so nothing gets lost Attach this folder to any Claude Code or Cowork session, and Claude reads it automatically. LAYER THREE: AI second brain (the advanced setup) Download Obsidian → create a vault → point Claude at the folder. Inject Andrej Karpathy's LLM Knowledge Base system prompt (source on GitHub). Feed it your notes, exports, and data. Claude builds a self-evolving wiki of everything you think, write, and research. Be sure to bookmark this and fix your Claude memory issues now.

SaaS isn't dying. 🤖 SaaS founders charging $99/month for ChatGPT wrappers are. 🤦💀

スマホでClaude codeをリモート操作する設定めちゃくちゃ面倒じゃない？

I'm hiring a PM for Claude Code, focused on model performance. If you have experience writing agentic evals and want to integrate research ideas into our core products, I'd love to hear from you here: job-boards.greenhouse.io/anthropic/jobs…

Whenever I see someone using anything but VS Code, I just assume they know more about programming than me.

Are there people who learn to code just to build their own idea? Not to freelance or work for any company .