> For the complete documentation index, see [llms.txt](https://twentytwo.documentationai.com/llms.txt). > Use it to discover all available pages before exploring further. # Create an Assistant Go to **My Assistants** and click **Create Assistant**. The setup is a guided 5-step flow: pick your industry and use case, choose a voice, define the goal, and Twenty2 builds the full conversation for you to refine. --- ## Step 1: Select your industry Choose the industry that best describes your business like Retail & E-commerce, Healthcare & Medical, Finance & Banking, Real Estate, and several others. This determines the use-case templates you'll see in the next step. --- ## Step 2: Choose a use case Pick what you want the assistant to help with. The options shown here adapt to the industry you picked, for example, a Healthcare business sees options like Customer Support, Outbound Sales, Lead Qualification, and Prescription Reminders, while for other industries, you will see a different mix relevant to them. --- ## Step 3: Pick a language and voice Select the primary language your customers speak, then choose an AI voice. Each voice has a name and a recommended use case to help you pick the right one, use the play button to hear a sample before deciding. You also set the **Caller Name** here, what the assistant introduces itself as on calls (e.g. "Hi, this is Varsha calling from Shiprocket."). --- ## Step 4: Give your assistant an identity Fill in your Assistant Name (used internally), Business Name, a Business URL, and the **Goal**. The Goal is a plain-language description of what the assistant should achieve on every call. Write it roughly and click **Refine with AI** to have it rewritten into a complete, structured prompt covering tone, objection handling, and the outcomes the assistant should drive toward. Once you click Next, Twenty2 generates the assistant's full conversation workflow. --- ## Step 5: Review the generated flow The Flow Editor opens automatically with a complete conversation already built — opening greeting, objection handling, callback scheduling, and a clear path to a final outcome, all based on your goal. This is your starting point, not the finished product: review it, adjust anything that doesn't match how you'd actually want the call to go, and add a [Knowledge Base](/guide/assistants/knowledge-base) if your assistant needs to reference specific business information. When you're happy with it, [test and publish](/guide/assistants/test-publish) your assistant. --- Learn how to review and edit your assistant's conversation flow.