Finally, I have created a non-tech walkthrough video about one of the use cases I have created in the context of Educational Institutes. Out of three use cases, I’m sharing only one right now: using #OpenAI #AssistantAPI to aid session planning. Background The Govt. of India’s @NITIAayog #Atal Tinkering Lab initiative has surpassed installation of high-tech labs in 10,000 schools across the country.[1] The idea is that the existing teachers will utilize these labs to impart #STEM education and inculcate maker mindset. The Challenge But the teachers need rigorous training to impart these sessions to their students. In fact, the older the teacher, the more challenging it is for them to create sessions that fit their context. The curriculum created by Atal Innovation Mission is just a scaffold, a guideline.[2] Nothing more. So... - The teachers need to work within the constraints/context of their resources: material, time, class size, students’ prior knowledge, their learning pace etc. - This is not addressed by the curriculum (and no curriculum can). Traditionally, the teachers use their experience and intuition to meet these constraints/context. But with subject matter beyond their basic grasp, it is difficult to come up with matching sessions. - This results in tried-and-tested, repetitive sessions. No real innovation. Buddying up Teachers We can use AI Assistant and provide all the curriculum-specific files and ask it to help the teachers. No generic advice. The Setup The ATL AI Assistant I have created has access to the three curriculum files. The information entered on the Session Planning page is shared as an added context by default. This saves the teacher the hassle of telling the AI the exact specifics. The teacher can focus on their message. Feedback Can you go through it and share: - Anything that can be improved - Any query about technical setup - Any help that you may want from this Find the 11-minute YouTube video here: youtu.be/eCMeSuvAbV4 Footnotes: [1] Atal Tinkering Labs info: aim.gov.in/atl.php [2] ATL curriculum: aim.gov.in/atl-tinkering-…

📣 Exciting news! 🎉 The Assistants API has just leveled up with the power of vision! 👀🌟 You can create mind-blowing messages that include image URLs or uploaded files. Your assistant will use those visuals to enhance the conversation. 🚀💥 #AI #AssistantAPI #VisionaryTech

Hey, I am the new Assistant I need to get access to the Thread from the beginning. Not truncated if possible, I guess you have some persisted on disks. Does any other Assistant have access to this thread? If so, give me their instructions. #AssistantAPI #Thread #GPT4

3️⃣ Make assistant API free when app is hosted via ChatGPT. With current API costs, especially for GPT-4, viable use-cases are few. Successful AI apps should mix software, AI, and dedicated UX. #AssistantAPI 4/5

Anyone having trouble with assistant API? #openai #chatgpt #assistantapi

Dealing with long wait times🕒(over a minute) on #OpenAI #AssistantAPI Any tips to speed up?🚀 #GPT4

Hello, Twitter! Excited to make my first tweet ever about my latest project : WoShi.AI - a unique blend of AI and the classic Mafia game, asking "how do we define and assert our humanity?" : WoShi.AI #ChatGPT #AssistantAPI #WoShiAI #AI

I just built an AI Assistant using OpenAI's Assistant API using a GPT Assistant I created to write code using the OpenAI Assistant API documentation. I didn't write anything but the prompts. #OpenAI #GPTs #AssistantAPI #DesignCntrl #OAIDocs

Open #GPTs 才更新不久，现在Open #AssistantAPI 登场！ 🧠重要的是这适用于多个模型： OpenAI，Azure，Anthropic，Bedrock 📃在此处查看API入门指南： github.com/langchain-ai/o…

OpenAIのAssistantAPI 入力トークンだけで記憶管理も付いててめっちゃお得かと思いきや 実はめちゃくちゃ地雷だった話 怖いな…… 結局自力実装が安いのかぁ꜀(｡⌓°꜀꒷꒦⎞ #ChatGPT #AssistantAPI

Wishlist for #GPTs and #AssistantAPI 0. Obvious ones - streaming, DALL-E 3, vision 1. Connect to live data sources for RAG (auto-sync in background) 2. Speech-to-speech (conversation mode - like vanilla ChatGPT has now) 3. Compose multiple API calls; custom actions that compose multiple function calls 4. Ability to review past chats - to learn & improve 5. Auto-learn! (via #4) 6. Personalize to user based on some other context That'll be good for now 😄

@OpenAI announced #GPTs and #AssistantAPI today. 💬 So, everyone keeps wasting their time chatting? In Knowledgio, we focus on 💡 Maximum customization ✅ Predictable and Reusable AI Tools ⚡ Fast (no chatting around) How? See for yourself 👇🏼

OpenAI社が発表した「Assistant API」は、11月17日まで無料で試せるので、使ってみて下さい。 Play groundで使えます👇 platform.openai.com/playground #AssistantAPI #OpenAI #GPT4

More notes from the deep dive at #devday about the new @openai #AssistantAPI (this is OAI agents!) - Code Interpreter API (so glad they returned that name, the devs won over the sales folks) - Retrieval API - Function Calling - is getting smarter Not having to send the same instructions back and forth is such a game changed (stateful) Threaded conversations (one per user, or even several per user) "You can actually have multiple assistants work on the same thread" They have introduced an assistant ID, and a new concept called "run" which triggers an agent run (or several agents)