📌 This article is for users of elvex 2.0. Find out which version you're using.
What a Space Is (and Isn’t)
A Space is a curated AI workspace — not just a folder. Everything you pin to a Space (agents, datasources, threads) is understood by elvex in relation to each other, so the AI always has an accurate picture of what’s there and how to use it.
Think of a Space like a place that you enter and based on what you ask elvex to do, it will use the items in the room to get the job done. It has context on what the work is, what tools are available, and what’s been done before.
Spaces are not:
A shared inbox or Slack channel
A way to let everyone work in the same thread (each user has their own)
A permission shortcut (Space membership ≠ resource access unless you want it to)
The Two Types of Spaces
My Space
Every elvex user has a My Space — a personal space that belongs only to them. It’s the home base for individual work and plays a special role in how elvex understands you:
Personal context — a plain-language description of your role, priorities, and working style that flows into every conversation you have, automatically
Pinned resources — agents, datasources, or threads you pin here are made available to the elvex home agent on your behalf without needing to attach them each time
My Space is how you teach elvex who you are. You don’t create it — it’s always there.
Shared Spaces
Shared Spaces are what creators build. When you create a Space and share it with others, you’re giving a team a persistent, shared AI context to work from together. Each member opens their own threads, but the Space’s resources, context, and accumulated learning flow into every conversation automatically.
Creators typically build Shared Spaces for two reasons: teams and projects.
Reason | Best For | Key Behavior |
Team | Ongoing functions (sales, marketing, support) | Persistent context and shared agents/datasources that stay relevant long-term |
Project | Bounded work with a clear goal | Context auto-updates as new threads accumulate; grows smarter over time |
Before You Create a Space: Ask These Questions
Who is this for? — Individual users, a team, a function, or the whole org?
What's the core use case? — What will people chat about inside this Space?
What resources does that use case need? — Agents, datasources, past threads?
How should elvex behave here? — Any specific norms, priorities, or instructions?
Who should manage membership? — Individual users or a synced User Group?
Setting Up Your Space: Step by Step
1. Name it clearly
Use a name that describes the purpose, not just the team.
✅ Sales Deal Prep or Q3 Product Launch
❌ Marketing or Misc
2. Write the Space context
This is the most important thing you'll do. Space context is a plain-language description that tells elvex what the Space is for, how to use the resources in it, and any norms to follow. Think of it as your onboarding doc for the AI.
A good Space context includes:
What this Space is for (1–2 sentences)
What each pinned resource is used for (e.g., "The Battlecard Agent is for live deal prep; the Competitor Datasource is for background research")
Any team-specific norms or priorities (e.g., "Always cite sources," "Prioritize brevity in responses")
Who the Space is for and their context (e.g., "This is used by Account Executives preparing for prospect calls")
Tip for Project Spaces: Enable auto-updating context. elvex will learn from each new thread in the Space and keep the context current — no manual maintenance required.
3. Pin the right resources
Only pin what's directly relevant to the Space's purpose. More isn't better — every resource you pin shapes how elvex reasons inside the Space.
Resource Type | When to Pin |
Agents | When there's a specialized task the team will repeat (e.g., a Battlecard Agent for sales calls) |
Datasources | When the team needs to ground AI answers in specific internal knowledge (e.g., a product docs datasource, a CRM export) |
Threads | When a prior conversation contains important context, decisions, or outputs worth preserving |
4. Manage permissions deliberately
Space membership and resource access are separate. Adding someone to a Space does not give them access to the agents or datasources inside it — unless you choose to.
Best practice: Use User Groups to manage access at scale.
Create or sync a Group that mirrors your team roster in your IdP
Add the Group to the Space (gives members access to the Space)
Grant the same Group access to each pinned resource (gives members access to the content)
When team membership changes, updating the Group cascades everywhere automatically
elvex will prompt you to use Groups when it detects multiple individual members with the same role — follow that nudge.
How People Work Inside Your Space
Each person can have their own threads so conversations stay focused and personal. What’s shared is the Space’s foundation: pinned resources, context, and accumulated learning, which flow into every conversation automatically.
Members have three ways to engage:
1. Chat with the entire Space Open a conversation directly in the Space and elvex draws on all pinned resources and context automatically. No setup, no manual attaching — the Space is the implicit context for everything.
2. Chat with an individual agent inside the Space If a specific agent is pinned (e.g., a Battlecard Agent or a Data Analyst), members can open a thread directly with that agent for a more focused interaction. The agent still operates within the Space’s context, but the conversation is scoped to what that agent does best.
3. Start a Space thread and drag in additional resources Members can open a Space conversation and augment it mid-thread by dragging in extra files, datasources, or threads. This is useful when a task needs something beyond the Space’s standard setup — a one-off document, a specific prior thread, or external context that doesn’t need to live in the Space permanently.
This means the quality of every conversation in your Space depends heavily on the quality of your Space context. Invest time in writing it well.
Common Patterns by Use Case
Sales Team Hub
Pin: Battlecard Agent, Objection Handler Agent, prospect research datasource
Context should include: ICP definition, key competitors, how to handle pricing questions, tone guidelines
What to avoid: Pinning every thread from every deal — be selective
Project Workspace
Pin: Relevant spec docs datasource, working threads with key decisions, a research agent
Context should include: Project goal, timeline, key stakeholders, what “done” looks like
Enable: Auto-updating context so the Space learns as work progresses
Onboarding Space
Pin: HR Policy Agent, company knowledge datasource, onboarding checklist thread
Context should include: Who this is for (new hires), what they can ask, where to go for escalations
Share with: All-company or a “New Hires” Group
Executive Briefing
Pin: Reporting agent, KPI datasource, prior briefing threads
Context should include: Which metrics matter, preferred level of detail, delivery format (bullet summary vs. narrative)
Creator Checklist
Before sharing your Space with the team:
Name clearly reflects the purpose
Space context is written and explains each pinned resource
Only relevant resources are pinned (not everything you might ever need)
Permissions are set — Space membership AND resource access
User Groups are used (or planned) for any team with 3+ members
Auto-updating context is enabled for project Spaces
You've chatted with the Space yourself to test the experience
What Makes a Great Space vs. a Mediocre One
Great Space | Mediocre Space |
Context explains why each resource is there | Resources pinned without explanation |
Focused on one purpose or team | Dumping ground for everything |
Uses User Groups for permissions | Individual permissions per user, not maintained |
Agents are purpose-built for the use case | Generic or unrelated agents pinned |
Context updated as work evolves | Context written once and forgotten |
Members know what to expect when they open it | Members start from scratch every time |