Introducing Projects: Context-Aware Workspaces for Your AI

The more you use AI, the more your work piles up in one place: client conversations next to personal notes, one book's outline tangled with another's, this quarter's strategy sitting beside last year's experiments. It all works — until your AI starts pulling the wrong context into the right task.

Projects fix that. They let you separate the different bodies of your work inside ARMES, so your AI focuses on exactly one workstream at a time — while the knowledge you want everywhere stays available everywhere.

Why one big pile of context isn't enough

Great AI output depends on great context. But "more context" isn't the same as "the right context." When everything lives in a single undivided space, your AI has to guess which of your worlds a request belongs to — and a freelancer's confidential client notes can leak into another client's deliverable, or your novel's plot can bleed into a different book.

The fix isn't separate accounts or separate apps. It's a way to tell your AI which workstream you're in right now — and have it adapt automatically. That's a Project.

Meet Projects

A Project is a workstream. Switch to it, and ARMES — including your AI — focuses on just that body of work, while your shared, everyday material stays available everywhere.

Out of the box, everything lives in a default view called All Projects. When you're ready to compartmentalize, you create Projects like "Acme Client," "Personal," or "Book Two," and start scoping your chats, notes, and files to them. Switching is a single click from your profile menu, and a small colored dot on your avatar reminds you which Project you're "in."

How it works: a lens you switch

There are two simple ideas, and understanding the difference unlocks the whole feature.

1. The active Project — your global lens. At any moment, one Project is active. It does two things at once:

• Filters what you see — your chats, notes, and files narrow to that Project (plus your shared items).
• Scopes what you create — anything new you make while a Project is active is automatically filed into it.

2. Per-item visibility — where each thing lives. Separately, every chat, note, file, or template has its own "Visible In" setting. You can move a single item into a Project — or back to "All Projects" — without changing your active lens.

The mental model in one sentence: the active Project is where you're standing; an item's "Visible In" is where that item lives.

Because new items inherit the active Project automatically, most of the time you never file anything by hand. You just switch your lens first, then work.

What's scoped, and what's shared

A Project can hold your chats, notes, files, folders, prompt templates, and even its own custom instructions. Your AI agents themselves are never divided up — the same roster is available everywhere; what changes per Project is the context they draw from and the instructions they follow.

Item	Belongs to a Project?	Notes
Chats	Yes	New chats inherit the active Project
Notes	Yes	Created in the active Project; movable anytime
Files	Yes	Scoped just like notes
Folders	Yes	A folder can be scoped on its own
Prompt templates	Yes	New templates inherit the active Project
Custom instructions	Yes — per Project	Each Project carries its own
Your AI agents	No	Always available; they act within the active Project

The All Projects layer is the secret to keeping this from becoming busywork. Anything left there — your standard bio, a reusable template, a master reference note — shows up no matter which Project you're in. So inside any specific Project you actually see two things at once: that Project's items, plus your shared material. You never see items from other Projects. Each workstream stays walled off, while your evergreen knowledge stays universal.

Your AI follows the active Project — automatically

Here's the part that matters most: when a Project is active and your AI looks through your context to answer you, it sees the active Project's material plus your shared "All Projects" material — and nothing from other Projects. When "All Projects" is your lens, it searches everything.

So a Project search is inclusive, not exclusive: you always get the current workstream layered with your shared knowledge, exactly mirroring what you see on screen.

A quick worked example. Say you have three notes:

• "Acme brand guidelines" → in the Acme Project
• "My writing style" → in All Projects (shared)
• "Personal goals" → in the Personal Project

Active Project	What your AI can pull in
Acme	Acme brand guidelines + My writing style
Personal	Personal goals + My writing style
All Projects	All three

Your shared writing style shows up everywhere, while the Acme and Personal material never leak into each other.

A best-practice idea, built in: the right context at the right time

There's a reason this design feels so clean. The hardest problem in getting consistently great results from AI isn't the model — it's context engineering: making sure the assistant sees what's relevant and not what isn't. Dumping everything into one space buries the signal in noise and risks confidential cross-talk. Walling work into rigid silos goes too far the other way and forces you to duplicate the things you actually want everywhere.

Projects strike the balance: isolation where it protects you, sharing where it helps you. Confidential, context-specific material stays scoped; evergreen knowledge stays universal. Your AI gets a precise, relevant slice of your world for every task — which means sharper answers, fewer mistakes, and zero cross-contamination. That's not just an organizing trick; it's good AI design you don't have to think about.

To go a step further, each Project carries its own custom instructions — global ones for every agent, plus per-agent fine-tuning. The same assistant can be your formal B2B copywriter in one Project and your playful personal-brand voice in another. Context-appropriate behavior, without juggling multiple assistants.

Real ways people use Projects

• Freelancers & agencies — a Project per client. Switch to "Acme" and every chat, brief, and file is theirs; your reusable intake checklist lives in All Projects so it's always on hand. Clean separation, zero cross-contamination.
• Work vs. personal — one buttoned-up Project, one casual one. Your bio and writing style stay shared so both benefit. One app, two completely different modes.
• Authors — a Project per manuscript, each with its own world bible and voice. Your AI stays "in character" for the right book and never bleeds plot between them.
• Students — a Project per course, so a tidy semester replaces one giant pile of notes.
• Founders & operators — Projects for Product, Marketing, and Ops, each with the right context and the right tone. Company-wide facts stay shared.

It's safe to reorganize

Worried about cleanup later? Don't be. Deleting a Project never deletes your work — its chats, notes, files, and templates simply fall back to "All Projects" and stay fully accessible. The only thing removed is that Project's custom instructions. Reorganize freely; move any item between Projects anytime with the "Visible In" control.

On every plan — including Free

Projects aren't a premium add-on. They work on the Free tier, on Pro, and on Ultra — because focused, private, context-aware AI shouldn't be locked behind a paywall.

If you're new to ARMES, the fastest way to feel the difference is to create a Project for one slice of your life and watch your AI lock onto it.

Start your 3-day Pro trial →

Or start free — no credit card required. Same Projects, same clean separation, on the house.

One workspace. Every workstream. Perfectly separated.