Projects as a Layer of Context Within a Space
Think of a Project as an additional, more specific layer of context nested inside its parent space.
Layer | Example |
Company context | "We are a technology company. Our product is elvex." |
Personal Space (My Space) | "I'm Allison, Head of AI Success." |
Shared Space | "This is the Customer Success team's hub — agents, playbooks, customer data." |
Project | "This is the Q3 renewal push for Acme Corp — deal notes, contacts, renewal history." |
When you open a thread inside a Project, elvex injects all active context layers automatically — personal, company, and the project's own context — so the AI already understands the specific initiative without any setup.
The key design principle: a Project's context should describe what makes this initiative distinct from the broader team. It doesn't replace the parent space, it deepens it. You explicitly choose which resources from the parent space (or anywhere else) to pin into the Project, so the Project has exactly the focused context it needs.
Example: Your "Customer Success" group space might have a datasource with all customer data and a renewal-tracking agent. Your "Acme Corp Renewal" project pins that same renewal agent and just the Acme-specific data, plus the deal thread history — giving you a tightly scoped workspace that inherits the team's tooling but speaks specifically to this account.
When Should You Use Projects?
Use a Project when you have a discrete piece of work that:
Has its own goal, context, and resources — separate from the broader team hub
Benefits from context that accumulates automatically over time (auto-context-extraction is on by default for Projects)
Involves multi-step or multi-agent workflows where a shared, evolving context foundation matters
Great for: a customer account, a product launch, a research initiative, an ongoing campaign — anything initiative-specific.
How to Open a Project
You can start a thread inside a Project from several places:
Spaces sidebar — find the Project in your sidebar, click it, and start a new thread from the Project's start screen
Parent Space → Projects tab — open the parent group space, go to the Projects tab, find the project card, and use the "New thread" action from the kebab menu (⋯)
Quick Create — use the Create menu in the icon rail and select Space → Project to create a new project; on the start screen you can immediately begin chatting
My Space — if the Project is pinned to your personal space, it appears in your My Space sidebar and can be opened from there
When you open the Project's start screen, you'll see:
A parent space eyebrow above the project title (e.g., "Customer Success Team") linking back to the parent space
The project's pinned resources — agents, datasources, and threads
An Auto-context status pill showing whether context is being learned from conversations
Should You Chat Directly With a Project?
Yes — that's the primary way to use a project. Opening a thread directly within the Project is the intended workflow. The Project's start screen is your launching point: it shows you the Project's context, resources, and prior threads, then you just type your message and go.
What you get when you chat inside a Project:
All context layers injected automatically — personal, company, and the project's own context
Pinned resources immediately available — no need to manually attach agents or datasources
Auto-context growth — with auto-extraction on, each conversation makes the project smarter over time, so future threads start from a richer foundation
Your own thread, shared foundation — each team member gets their own focused conversation, but everyone shares the same Project context, pinned resources, and accumulated knowledge
You can also pin a completed thread back into the Project so teammates can reference it — turning individual work into shared, reusable context for the whole team.
Can a Project Appear in Multiple Spaces?
Yes. A Project has one canonical parent (the "Hosted by" space set at creation), but like any other resource in elvex, it can be pinned to additional spaces. The Project itself lives in one place — it just surfaces wherever it's been pinned.
Are Projects Multiplayer?
Yes, with an important nuance. When a Project is shared, all members share the same Space context, pinned resources, and auto-extracted knowledge — so everyone benefits from the same accumulating foundation.
However, each user opens their own individual thread when chatting within the Project. Conversations are not shared live between users. What is shared is the foundation: the context, agents, datasources, and any threads that have been pinned back to the Project. Members can pin notable threads into the Project so others can reference them — turning individual work into institutional knowledge.