Every engagement you run in ELLA lives inside its own Client Workspace: a secure, shared space where you and the people you invite work through one client's exit planning together. This article explains why Workspaces exist, how we keep them isolated from one another, and how they fit into the rest of your account.
If you only read one thing: a Workspace holds everything for a single client, and nothing crosses from one client's Workspace into another's.
Most advisors carry a dozen or more active relationships at once, and each one comes with its own documents, its own history, and its own confidential details. When advisors reach for a general-purpose AI tool to move faster, all of that context ends up sharing a single thread, which is one careless prompt away from mixing information between clients.
We built the Workspace to solve that at the structural level. Each client gets a dedicated space, so the financials you uploaded for one business, the questions you have asked, and the deliverables you have drafted all stay attached to that specific engagement. When you open a Workspace, ELLA is working from that client's context and only that client's context. When you close it and open another, you are starting from a clean, separate foundation.
This is what lets you move quickly without carrying risk. You get the speed of working with AI across your whole book of business, with the confidence that each client's information stays where it belongs.
Sandboxing means each Workspace is walled off from every other Workspace in your account.
A few things follow from that design:
Documents and data stay put. Anything uploaded into a Workspace, whether by you, the business owner, or an invited collaborator, belongs to that Workspace. It does not appear, get referenced, or get suggested inside any other Workspace.
The AI's memory is per-Workspace. As you work, ELLA builds up context about the engagement so each interaction is smarter than the last. That memory is scoped to the single Workspace it was built in. It informs future responses for that client, and it never carries over when you switch to a different one.
No cross-contamination between clients. Because context is contained, there is no bleed from one engagement into the next. Two clients in the same industry will not have their details blur together.
We also do not use the content you put into a Workspace to train foundational AI models, except in anonymized and aggregated form for performance, security, and quality purposes, and you can opt out of the quality-evaluation use in your account settings. The full detail lives in our Terms of Service and Privacy Policy.
Your account is organized in three layers, from the top down:
Layer | What it is | Who creates it |
|---|---|---|
Organization | The top-level container for your practice. It holds all of your Workspaces and the people you work with. | The Advisor |
Workspace | One contained space per client engagement, holding that client's documents, insights, and deliverables. | The Advisor, inside the Organization |
People | The Advisor, Business Owner, and any Exit Team Members invited into a given Workspace. | Invited by the Advisor, or by the Business Owner where allowed |
In practice, you create your Organization once, then spin up a new Workspace for each client. Inside each Workspace, you invite the people who need to be part of that specific engagement, and you set what each of them can see and do.
Advisor. You quarterback the engagement. You create the Workspace, invite everyone else, and control access.
Business Owner. The client. Owners are active participants, not just recipients. They contribute data during fact finding, upload financials, answer intake questions, and review deliverables as the work progresses.
Exit Team Members. The collaborators you pull in for their piece of the work, such as an attorney reviewing an agreement or a CPA running the tax analysis. They contribute inside the shared Workspace with the access you grant them.
Each Workspace is where the actual work happens, across three connected capabilities:
Fact finding. Structured intake and document ingestion. Upload the paperwork and answer the intake questions, and ELLA organizes it into a coherent picture of the business.
Sensemaking. Ask questions against the full context of the engagement and get analysis grounded in that client's specifics, not generic output.
Deliverables. Pull insights directly into client-ready documents, refine them, and share them, so the deliverable reflects the owner and the business.
Alongside these, the Workspace holds the uploaded documents and the AI memory described above, all of it scoped to that one client.
You control access to each Workspace. When you invite a Business Owner or an Exit Team Member, you decide what they can see and do, and you can adjust those permissions as the engagement changes.
Because access sits in your hands, a couple of responsibilities come with it. ELLA does not manage individual permissions or access roles on your behalf, so it is worth reviewing who has access to a Workspace before you invite someone new or share sensitive material. Keeping login credentials confidential matters too, since activity in a Workspace is tied to the account that performs it.
Whoever uploads a document or inputs data into a Workspace retains ownership of it, including inside shared Workspaces. ELLA holds your content so you can work with it, and you keep the rights to it.
When an engagement ends and you close out or cancel, you retain ownership of your data, access to the Service is turned off, and we delete User Data in line with our data retention policy. For the complete picture of ownership, confidentiality, and how we handle data, see the Terms of Service and Privacy Policy.
Can context from one client ever show up in another client's Workspace? No. Documents, data, and the AI's memory are scoped to the single Workspace they live in, and none of it carries across.
Who can see a Workspace? Only the people invited into it: you, the Business Owner, and any Exit Team Members you add. You set what each person can access.
Does ELLA use my clients' data to train AI models? We do not use your AI Inputs or Outputs to train models, except in anonymized and aggregated form for performance, security, and quality evaluation. You can opt out of the quality-evaluation use through the analytics controls in your account settings.
