Another hack on claude/gpt - when the chat is becoming long, ask the AI to summarize it. Paste the initial prompt + summary + any additional instructions in a new chat and you can start again.
I feel like we're all fighting the AI token wars these days, Xinran.
Token budgeting is the new digital economy, so strategic prompting across multiple platforms is literally the only way to survive without emptying your wallet completely.
I love your strategy of testing prompts in Claude before committing to app builders. That's honestly genius. Why waste precious tokens on half-baked ideas?
The token-saving strategies resonated with me - it's a challenge I've faced while building various digital solutions. Strategic tool allocation is crucial; I often use different AI models for specific tasks in my workflow. Planning which tasks go to which AI tools has saved me from those frustrating limit messages more times than I can count. For complex e-commerce projects especially, this approach has been invaluable when working with product feeds and customer journey maps.
Hi there, good article, and advice. I've been writing a Novella with Chat for a couple months now and am finishing the first book now Eye of the Beholder. A second book is well underway. Just a quick question, I've been frustrated with Chat recently, I think they actually made a downgrade, but it isn't remembering the story as much, and it seems to be hallucinating more at the moment...
Have you noticed this yourself? I'm talking like 2-3 days max...
Also, I've written a nice post about mastering the image generating with all the attempts I had to get to the perfect cover (part 2 - better than part 1), so here it is:
And I'll share this with you, it's the first chapter of the epic saga for you and your fans, it's about the awakening of AI with humans and not against humans, and many other exciting deep topics:
On another note, since we have similar interests, and readers, I've been recommending you for a while and have sent you 16 new subscribers!!! I can show you the screenshot... but can I ask you to add me to your recommended Substacks, that would be awesome and help me so much?
Another hack on claude/gpt - when the chat is becoming long, ask the AI to summarize it. Paste the initial prompt + summary + any additional instructions in a new chat and you can start again.
Love this! Have not tried it before.
I just subscribed, perhaps you can do the same...? https://open.substack.com/pub/designwithai/p/what-to-do-if-you-hit-the-usage-limit?r=b8pvb&utm_campaign=comment-list-share-cta&utm_medium=web&comments=true&commentId=111759978
I feel like we're all fighting the AI token wars these days, Xinran.
Token budgeting is the new digital economy, so strategic prompting across multiple platforms is literally the only way to survive without emptying your wallet completely.
I love your strategy of testing prompts in Claude before committing to app builders. That's honestly genius. Why waste precious tokens on half-baked ideas?
My rule: 1 chatbox for 1 purpose. Move on if the purpose has been fulfilled.
The token-saving strategies resonated with me - it's a challenge I've faced while building various digital solutions. Strategic tool allocation is crucial; I often use different AI models for specific tasks in my workflow. Planning which tasks go to which AI tools has saved me from those frustrating limit messages more times than I can count. For complex e-commerce projects especially, this approach has been invaluable when working with product feeds and customer journey maps.
I generally leverage options 1 or 4. But, these are great tips for when those do not work for my needs.
Thanks Colette! For ChatGPT and Claude, option 1 and 4 are most common for me too.
Hi there, good article, and advice. I've been writing a Novella with Chat for a couple months now and am finishing the first book now Eye of the Beholder. A second book is well underway. Just a quick question, I've been frustrated with Chat recently, I think they actually made a downgrade, but it isn't remembering the story as much, and it seems to be hallucinating more at the moment...
Have you noticed this yourself? I'm talking like 2-3 days max...
Also, I've written a nice post about mastering the image generating with all the attempts I had to get to the perfect cover (part 2 - better than part 1), so here it is:
https://shifthapens.substack.com/p/mastering-chats-new-image-generator-4dc?r=b8pvb
And I'll share this with you, it's the first chapter of the epic saga for you and your fans, it's about the awakening of AI with humans and not against humans, and many other exciting deep topics:
https://shifthapens.substack.com/p/eye-of-the-beholder-3ab?r=b8pvb
On another note, since we have similar interests, and readers, I've been recommending you for a while and have sent you 16 new subscribers!!! I can show you the screenshot... but can I ask you to add me to your recommended Substacks, that would be awesome and help me so much?