Claude's New Project Feature!

[@AI.MattEspo]

Thanks for watching! 🎥 How do you think Claude's new project feature can transform your workflow? Share your experiences or questions below! 💡 For more AI learning resources, join our community: AI Education Vanguard Skool. Let's continue the discussion there! #AIEducation #ContinuousLearning

[@human_shaped]

Really great summary. I hadn't gotten around to playing, and you've saved me a lot of time. Good insights about what actually matters. Now to decide if I really want to share everything with them.

[@AI.MattEspo]

Hi @human_shaped, Hahahaha good point! Only sharing what I don't mind being out there! Thank you for the kind words!

[@romaiorjh9390]

Thanks, get insights from your video

[@AI.MattEspo]

Glad it was helpful!

[@The7MinuteLife]

Great job. I learned a lot!

[@AI.MattEspo]

Thank you! 🙏

[@super7ace]

Great video

[@AI.MattEspo]

Glad you enjoyed it!

[@Pyroplan]

What is the the "34% of knowledge size used" mean? Does it mean you could upload 3 times as much until it hit the limit even though you already uploaded large files? How is it doing that. Is it just using the size of the context window? Or is it somehow indexing it or summarizing it to get more knowledge into it? What will happen if you upload more? Will it just not know everything or are you not able to upload more?

[@AI.MattEspo]

Hey there! Great questions! Let me break it down for you: The "34% of knowledge size used" means you've filled about a third of Claude's available context window with your uploaded content. The full context window is 200,000 tokens (roughly 150,000 words or 500 pages), so you still have plenty of room! Yes, you could indeed upload about three times as much before hitting the limit, even if you've already uploaded some large files. Claude's context window is quite spacious! Claude isn't using any fancy indexing or summarization tricks here. It's simply using the entire content you've uploaded to inform its responses, working within that 200,000 token context window. If you upload more content beyond the 200,000 token limit, don't worry - Claude won't suddenly forget everything or prevent you from adding more. Instead, it'll start a new conversation to process the additional content. This helps maintain accuracy while allowing you to work with large amounts of data. The Projects feature is designed to handle complex, long-term projects with lots of information. So feel free to keep adding content as needed for your project. Claude will work with what you give it, starting new conversations if necessary to accommodate everything. Hope this clears things up!

[@Pyroplan]

@@AI.MattEspo Thank you that is helping. And yes, I am aware it is not for full complex projects, more like what we have with ChatGPT and CustomGPTs. I really liked Claude 3.5 Sonnet from what I have seen during my "free" testing. And currently we are using mainly ChatGPT at work, where I created a customGPT for my products I am responsible to get better tailored responses (Product Marketing). And with Claudes 3.5 performance I could image it would do even better.

[@Blakeapatow]

I needed this 😁😁😁

[@AI.MattEspo]

Don’t we all!

[@ldelossantos]

I couldn't get the difference with assistances other than the artifacts feature. Did I miss something?

[@AI.MattEspo]

Good question! Besides the artifacts feature, Claude’s projects have some cool stuff like: 1. Persistent Context: All your chats and docs are kept together, so no need to switch around. 2. Custom Instructions: You can set specific instructions for each project. 3. Document Upload: Upload docs directly into your chats and reference them easily. 4. Knowledge Management: Claude remembers and uses the info from your uploaded docs across all project chats. Does this clear is up?

[@yaseenal-wesabi5964]

How can you link this to a chatbot that can be used on a personal website through API.

[@AI.MattEspo]

Not sure hoe to do that with projects. Might be better to create assistants with Open AI. You cam make API Calls to assistants that you create.

[@avi7278]

No it does not use cross conversation context in projects, you're mistaken. Do an actual test and you will see you cannot refer to other conversations in the project.

[@AI.MattEspo]

Hi @avi7278, Thanks for pointing that out. You’re correct, the current implementation doesn’t support cross-conversation context within projects. I’ll make sure to test more thoroughly in future videos. Appreciate your keen observation!

[@Luca-gb1og]

But isn’t that what GPTs in chat gpt are?

[@AI.MattEspo]

Hi @Luca-gb10g, Great question! GPTs in ChatGPT use a system prompt for guidance in each conversation, but this prompt doesn’t carry over to new projects or separate chats. Each project can have specific instructions, but they won’t apply outside of that project. Thanks for asking!

[@AI.MattEspo]

Its a small difference, Depends on your use case when I would want to use one or the other. Does this help?