A few weeks ago I wrote about why AI spits out generic phrases when you give it vague prompts. The fix was simple: write longer, more detailed prompts with real context. Tell the AI who you are, who you're talking to, what tone you want.

That advice works but retyping all that context every single time gets old fast.

There's a better way, and thagt way is Projects.

Every major AI platform now has some version of persistent workspaces. Claude calls them Projects. ChatGPT calls them Projects too. Perplexity went with Spaces. Grok has Workspaces. Gemini took a slightly different approach with something called Gems. Mistral's Le Chat has Projects as well.

The names are different but the idea is the same. You create a container, upload your files, set your instructions once, and every conversation inside that container starts with all that context already loaded. No more explaining yourself from scratch.

Here's how I actually use this. I have a Claude Project specifically for writing this newsletter. It has my previous issues uploaded. It has instructions about my voice, my audience, my pet peeves about corporate jargon. When I sit down to draft, I don't spend the first ten minutes getting Claude up to speed. I just start working.

You could do the same thing for client work. Upload the requirements doc, the brand guidelines, the previous deliverables. Set instructions like "always write for small business owners in the Gulf Coast who are new to this topic." Now every conversation in that project knows the context without you repeating it.

But this isn't just for work. I've been thinking about setting one up for my house. Think about all the random details you forget until you need them. What's labeled on your breaker panel. When your house was built, the model numbers for your various systems. Your plumber's name and number, your electrician, your pest control company. The paint colors you used in each room. When you last replaced the roof.

You could dump all of that into a single project and suddenly have an AI that actually knows your house. "Which breaker controls the outdoor outlets?" "When did we last have the HVAC serviced?" "What was the name of that guy who fixed the garbage disposal?" All answered instantly, without digging through filing cabinets or scrolling through old emails.

The details vary a bit across platforms. Claude gives you a massive context window that can expand even further when you upload a lot of files. ChatGPT lets free users upload five files per project, while Pro users get forty. Perplexity does something clever where you can search your uploaded documents and the web at the same time. Grok's version includes live code execution if you're into that. Gemini's Gems are more like custom personas you can share with coworkers.

Where do you find these features? In Claude, there's a Projects section in the left sidebar. ChatGPT has Projects in the same spot. Perplexity puts Spaces in the left panel too. None of them are hidden, but they're easy to miss if nobody points them out.

Most of these features work on free tiers with some limits. Paid plans give you more files, longer context, and extra tools. But you can absolutely experiment today without spending anything.

The real shift here isn't about the AI getting smarter. It's about the AI finally remembering who you are and what you're working on. That's what turns a chatbot into something that actually feels like a collaborator.

If you use any of these tools regularly, try setting up one project this week. Pick something you do repeatedly, or something you always forget. Give it the context once. See how it changes the conversation.

// community highlights

Know someone who’d enjoy this?

Don’t miss the next issue.