How to Set Up a Collaborative AI Workspace for Your Team in Geekflare Chat

I was talking to a tech startup founder and his team, and that’s when I understood how this is the case with so many of these companies right now. One team member starts using ChatGPT Plus. Another one of them signs up for Claude Pro. And within a month, you have 4-5 people paying for different AI subscriptions, writing the same prompts from scratch, and none of them have any idea about what the other team members have figured out.

If you are running AI models for personal experiments, then it’s fine. But if this is how your team operates AI, then you need to think again. It makes a lot more sense to have a shared workspace if:

• Multiple people on your team use AI daily
• You are paying for more than 1 subscription
• You want consistent outputs across your team members (same brand voice, data sources, or prompt patterns)
• You want to keep a track of how your team members are using AI

Geekflare Chat‘s Business plan consolidates all of this into a single account. 5 seats with access to every premium model and 15,000 pooled credits. Plus, the team has access to shared prompt libraries, a shared knowledge base to feed company documents, and there are role-based access controls. The cost is only $29/month for the whole team. That’s only a few bucks over a single ChatGPT Plus or Claude Pro subscription costs.

Let’s test this in real-time!

Creating a Workspace and Inviting Your Team

It literally takes less than 5 minutes to set this up.

Step 1: Create your workspace

Once you sign up, get to the dashboard, click “Create Workspace” and give your new workspace a name and a URL slug. Each workspace runs independently with its own permissions, prompts, and conversations. Please note that only Admins can create a new workspace.

You can create unlimited workspaces on the Business plan. If your team runs distinct functions where you don’t want different departments to see each other’s work, you can set that up. For instance, there are client-facing and internal ops teams, so you can have separate workspaces to keep everything clean. By default, the users on the Pro plan get one workspace.

Step 2: Invite your team members

Click ‘Members’ in the left menu and ‘Invite Members’ from the top right to invite new members to your workspace. You can add people by email and assign them to the relevant workspace during the invite.

Step 3: Set roles

Each member can be assigned one of the three roles: Owner, Admin, or Member. Owners and Admins can create and delete workspaces, view all workspaces in the organization, and manage permissions. On the contrary, members can only see workspaces that they’ve been added to.

To update someone’s role later, go to Members, select the person, click the three-dot menu under their workspace, and change their role or remove them.

There’s one thing here to note for team leads. Chats inside a workspace are not automatically visible to everyone. A conversation is visible only to the person who created it and anyone it’s been explicitly shared with. Admins can see all workspaces, but they can’t read individual chats unless they are shared with them. This is privacy by default, not surveillance by default.

Building a Shared Prompt Library

The easiest and fastest way to get consistent output across your team is to stop letting everyone write their own prompts from scratch.

Geekflare Chat’s prompt library serves two purposes over here. First, it ships 310+ readymade prompts across categories such as Marketing, Customer Support, Project Management, Sales, HR, Finance, and more. These are all pre-built by experts with variable-enabled templates for your team to start working with immediately. The second thing is that you can create and share your custom prompts among your team members to take things to the next level.

How to access it: Click the ‘Prompts’ icon in the chat input box to see the full library of prompts organized by category.

Using a readymade prompt: You can use any prompt from the category you want. If the prompt contains variables (such as [Topic], [Tone], or [Industry]), the interface prompts you to fill them in before inserting the final text into the chat.

Creating a custom prompt: Geekflare Chat allows you to build a tailored prompt for your team or clients’ needs. You can add variables for the parts that change each time. Once you write the prompt, you can choose the sharing scope: share it within your workspace, keep it private, or publish it organization-wide so everyone in your workplace can access it.

This is where you get access to the full power of this setup. A senior marketer spends 20 minutes building 5 solid prompts. Now, every junior marketer on the team can produce the same quality of outputs with these prompts, without needing to learn prompt engineering from scratch. The library eliminates the gap between your best prompt writer and everyone else.

Setting Up a Shared Knowledge Base

The Knowledge Base standardizes everything that the AI knows about your company. You can upload all your company documents, like SOPs, brand guidelines, competitive intelligence, onboarding materials, product specs, and more. Once you’ve uploaded everything, you can reference it in any conversation in that workspace. This is how AI is going to answer the questions based on your actual documents rather than relying on its general training data.

How to set it up: Navigate to the Knowledge Base section in your workspace and upload all your files. You will get up to 250 MB storage on the Business plan and 500 MB on the Scale plan.

What to upload first: First, begin with the documents that your team is going to reference most often. It could be anything, like brand voice guidelines and product messaging docs for the marketing teams. Engineering teams can start with architecture decisions and internal documentation. At the same time, your sales teams can begin with objection-handling docs and battlecards.

You should note one thing: every message that queries the Knowledge Base is going to cost an extra 5 credits on top of the model’s base cost. If your team runs 50 Knowledge Base queries a day on Advanced models, that’s 1,000 credits per day. So, you need to plan accordingly for your team. You can get the full tier-by-tier breakdown in the credits explainer guide.

Getting Your Team Productive in an Afternoon

When you sign your team up for Geekflare Chat, you won’t really need a week-long rollout. The setup is only three steps, each building on the last.

Hour 1: Create your workspace and invite the team

Follow the walkthrough we talked about above to create your workspace and get your entire team logged in. They will know exactly how to switch workspaces and access everything.

Hour 2: Seed the Prompt Library and Knowledge Base

Get your team lead or admin to spend an hour uploading the 5 most-referenced company documents to the Knowledge Base. This will make it available to the entire team. Plus, curate 10 prompts that would match your team’s most common tasks and add them to the Prompt Library and share access across the workspace. This is how you start feeding, and the library can grow from there.

Hour 3: Run a live session

Rather than doing this in isolation, it’s better to get your team on a call where one person shares the screen. Now, take a real task and walk through it together with your team. I’d suggest picking a concrete task like summarizing your client brief using the brand voice prompt you built or extracting the key numbers from last quarter’s reports. Sending out irrelevant training documents won’t do a thing. Instead, it’s better to go with a successful real-world example like this.

After that, people are self-sufficient. They know where the prompts are, that the knowledge base exists, and that they’ve seen a real output. Everything else they learn by doing.

For individual onboarding habits that help AI adoption stick, our first week guide gives each team member a day-by-day plan.

The Pricing Math

I know you would have run this math through your head so many times as a team leader. Let me simplify it for you.

Geekflare Chat Business Plan gives your team 15,000 pooled credits per month, and it’s not 3,000 per person. So, if your team has one heavy user and another is light, the credits can be compensated accordingly. If your team is using Advanced-tier models (15 credits per message) the most, then 15,000 credits translate to 1,000 Advanced model queries per month across the team. So, that’s almost 200 queries per person. This is more than enough for the majority of teams, but you always have the option to top up those credits if you are running out of them at any point.

What Admins Can and Can’t See?

I know every team member would ask this question before signing up for a shared AI tool. They want to know: Can my boss read my chats?

No, your boss can’t read your chats.

Let me tell you what visibility looks like on Geekflare Chat. Owners and Admins have the privilege to view all workspaces in the organization and to create or delete them. On the contrary, members can only see workspaces to which they’ve been invited. But workspace visibility is completely different than chat visibility. A conversation is visible only to the person who created it and to anyone it has been explicitly shared with. Workspace membership does not automatically grant access to other members’ conversations.

I know this matters a lot because people won’t opt for something that they can’t trust. Nobody is going to use a shared AI workspace for drafting sensitive emails or brainstorming ideas if they think their manager is reading the logs.

You don’t need to worry about the privacy of your chats with this, by any chance.

If you are ready to onboard your team onto Geekflare Chat, then you only need to sign up here!