Table of Contents
OpenAI
Models / Pricing
API
Customize with own data
Preparing your dataset https://platform.openai.com/docs/guides/fine-tuning/preparing-your-dataset
https://platform.openai.com/docs/guides/fine-tuning
https://levelup.gitconnected.com/how-to-build-your-own-custom-chatgpt-bot-cf4af959adcc
https://platform.openai.com/docs/api-reference/engines/retrieve#fine-tunes/create-training_file
Quadrants
🤖 𝐖𝐡𝐲 𝐘𝐨𝐮𝐫 𝐀𝐈 𝐂𝐨𝐝𝐢𝐧𝐠 𝐀𝐬𝐬𝐢𝐬𝐭𝐚𝐧𝐭 𝐊𝐞𝐞𝐩𝐬 𝐆𝐞𝐭𝐭𝐢𝐧𝐠 𝐈𝐭 𝐖𝐫𝐨𝐧𝐠. 𝐀𝐧𝐝 𝐇𝐨𝐰 𝐭𝐨 𝐅𝐢𝐱 𝐈𝐭
The AI coding debate is stuck in extremes. One side says, “AI will write 90% of code next year.” The other says it’s just a flashy autocomplete. Both are missing the point.
The real question isn’t can AI write code. 𝐈𝐭’𝐬 𝐰𝐡𝐢𝐜𝐡 𝐭𝐚𝐬𝐤𝐬 𝐢𝐬 𝐢𝐭 𝐚𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐠𝐨𝐨𝐝 𝐚𝐭, 𝐚𝐧𝐝 𝐡𝐨𝐰 𝐝𝐨 𝐰𝐞 𝐫𝐞𝐬𝐡𝐚𝐩𝐞 𝐨𝐮𝐫 𝐰𝐨𝐫𝐤 𝐭𝐨 𝐭𝐚𝐤𝐞 𝐚𝐝𝐯𝐚𝐧𝐭𝐚𝐠𝐞?
𝐘𝐨𝐮𝐫 𝐧𝐞𝐰 𝐚𝐫𝐭𝐢𝐟𝐢𝐜𝐢𝐚𝐥 𝐭𝐞𝐚𝐦-𝐦𝐞𝐦𝐛𝐞𝐫 ✅ 𝐒𝐭𝐫𝐞𝐧𝐠𝐭𝐡𝐬: Codes like a solid senior engineer, knows a ton of tech, and is lightning-fast with well-scoped tasks. ❌ 𝐖𝐞𝐚𝐤𝐧𝐞𝐬𝐬𝐞𝐬: It designs like a junior, knows nothing about your team or codebase, and forgets everything unless you tell it otherwise.
💡 𝐓𝐡𝐞 𝐂𝐨𝐧𝐬𝐭𝐫𝐚𝐢𝐧𝐭-𝐂𝐨𝐧𝐭𝐞𝐱𝐭 𝐌𝐚𝐭𝐫𝐢𝐱. A simple model to figure out where AI shines: ✅𝐇𝐢𝐠𝐡 𝐂𝐨𝐧𝐬𝐭𝐫𝐚𝐢𝐧𝐭 + 𝐇𝐢𝐠𝐡 𝐂𝐨𝐧𝐭𝐞𝐱𝐭: Perfect for AI (e.g., removing a feature flag) ❌𝐋𝐨𝐰 𝐂𝐨𝐧𝐬𝐭𝐫𝐚𝐢𝐧𝐭 + 𝐋𝐨𝐰 𝐂𝐨𝐧𝐭𝐞𝐱𝐭: Poor fit (e.g., designing a photo upload feature from scratch)
𝐓𝐨 𝐰𝐨𝐫𝐤 𝐰𝐢𝐭𝐡 𝐀𝐈 𝐞𝐟𝐟𝐞𝐜𝐭𝐢𝐯𝐞𝐥𝐲: 🔒 𝐂𝐨𝐧𝐬𝐭𝐫𝐚𝐢𝐧 𝐭𝐡𝐞 𝐭𝐚𝐬𝐤: Be more directive and break it down. 🧠 𝐏𝐫𝐨𝐯𝐢𝐝𝐞 𝐦𝐨𝐫𝐞 𝐜𝐨𝐧𝐭𝐞𝐱𝐭: Expand prompts, add memory, and point to examples or docs. 🛠️ 𝐋𝐞𝐯𝐞𝐫𝐚𝐠𝐞 𝐭𝐨𝐨𝐥𝐬: Let the agent search code, read internal docs, or fetch tickets.
🛑 Stop treating your AI assistant like a magic box. Start 𝐭𝐫𝐞𝐚𝐭𝐢𝐧𝐠 it like a 𝐬𝐦𝐚𝐫𝐭 𝐛𝐮𝐭 𝐛𝐫𝐚𝐧𝐝-𝐧𝐞𝐰 𝐭𝐞𝐚𝐦𝐦𝐚𝐭𝐞. Guide it well, and it’ll take tons of grunt work off your plate.
📌 Before handing off your 𝐧𝐞𝐱𝐭 𝐭𝐚𝐬𝐤 to AI, ask: 🧩 Is the solution space clear? 📚 Does it have all the context? 👉 If not: reshape the task.
🔗 Huge thanks to Pete Hodgson for the brilliant breakdown. Link to the full article is in the comments

