Parsa AI can write emails tailored to the recipient using data from your CRM. Rather than starting from a blank screen, you describe what you want to say and the AI drafts a complete, personalized email using: * The contact or lead's name, company, title, and recent activity * The linked opportunity or deal context * Your previous email history with that person * Any custom instructions or tone guidance you provide You review the draft before anything is sent.

***

Chat Workflow

## Generating Emails from Chat

The most flexible way to generate emails is through the AI chat panel.

**Basic prompt:**

``` Write a follow-up email to John Smith at Acme Corp about the proposal we sent last week. ```

**With context and tone guidance:**

``` Write a warm re-engagement email to Sarah Kim. She's a lead who went quiet 3 weeks ago after an initial discovery call. Keep it brief and low-pressure. ```

**For a specific goal:**

``` Draft an email to Marcus Rodriguez introducing our new enterprise tier. He's in the Negotiation stage and his main concern was onboarding time. ```

After the AI drafts the email, you'll see: * **Subject line** * **Email body** * **Suggested send time** (optional) Click **Send** to deliver it immediately, **Edit** to revise it, or **Discard** to start over.

***

Automation

## AI Email Mode in Automations

When building an automation, the **Send Email** action offers three modes: | Mode | Description | | ---------------- | ------------------------------------------------------------------------------------- | | **Template** | Uses a pre-built email template with variable substitution | | **Custom** | Write a static email body directly in the automation builder | | **AI-Generated** | Provides a prompt to the AI; each email is generated fresh for the specific recipient | The **AI-Generated** mode is powerful for sequences where each message should feel personal rather than templated. For example: * A nurture follow-up 7 days after a lead goes quiet, where the AI references the lead's industry and role. * A proposal follow-up that references the proposal title and the account name. To configure an AI email action: 1. Add a **Send Email** action to your automation. 2. Select **AI-Generated** as the mode. 3. Write a prompt template. You can use CRM variables inside the prompt.

***

Prompt Variables

## Customizing AI Prompts

Prompts in AI email actions support CRM template variables, which are filled in at send time:

| Variable | Resolved to | | ----------------------- | ----------------------------- | | `{{contact.firstName}}` | Contact's first name | | `{{contact.lastName}}` | Contact's last name | | `{{contact.company}}` | Contact's linked account name | | `{{contact.title}}` | Contact's job title | | `{{lead.firstName}}` | Lead's first name | | `{{opportunity.name}}` | Opportunity name | | `{{opportunity.value}}` | Opportunity value | | `{{user.firstName}}` | Assigned rep's first name | | `{{workspace.name}}` | Your workspace name |

**Example automation prompt:**

``` Write a short, friendly follow-up email to {{contact.firstName}} at {{contact.company}}. They are in the {{opportunity.stage}} stage and the rep is {{user.firstName}}. Mention that we're happy to answer any questions and suggest a quick call. Keep the email under 100 words. ```

***

Templates

## Template Variables in Email Templates

Static email templates (not AI-generated) use the same `{{variable}}` syntax. Any variable that isn't found in the CRM for a specific recipient will be left blank unless you provide a fallback:

``` Hi {{contact.firstName | there}}, ```

The `| there` fallback means the email reads "Hi there" if the first name isn't available.

***

Quality Control

## Reviewing AI Email Quality

A few best practices:

Always include a tone directive in your prompt — "professional," "casual," "brief," or "empathetic" significantly shapes the output quality. AI emails may occasionally hallucinate facts not in your CRM. Never rely on AI to fill in pricing, dates, or contractual commitments without reviewing the output first.

* Keep prompts focused — one clear goal per email generates better results than multi-part instructions. * Reference CRM fields explicitly — the AI performs better when it has concrete details (name, company, stage) rather than abstract concepts. * Test AI automations with your own email address before going live to review output quality.