Ask AI Action

The Ask AI action lets you leverage AI models directly in your HubSpot workflows. Use it to analyze data, generate content, classify records, and automate decisions.

How It Works

1. The action sends your prompt to an AI model via OpenRouter
2. The AI processes the request with any CRM data you include
3. The response is returned in your chosen output format
4. You can use the output in subsequent workflow actions

Configuration Options

Model Filter

Choose which AI models to display:

• Recommended Models - Top picks from major providers (GPT-4, Claude, Gemini, etc.)
• All Models - Complete list of available models

Model Selection

Select from a wide range of AI models including:

• OpenAI GPT-4 and GPT-4o
• Anthropic Claude 3 and Claude 3.5
• Google Gemini
• Meta Llama
• And many more

Output Type

Choose how the AI should format its response:

Output Type	Use Case
Text	Free-form responses, content generation, summaries
Select/Multiselect	Classification, choosing from predefined options
Number	Scoring, ratings, numerical analysis
Boolean	Yes/no decisions, true/false classifications

Target Property (Optional)

When using Select/Multiselect output, you can specify a CRM property. The AI will choose from that property’s valid values, ensuring the output can be directly saved to the record.

Prompt

Write your prompt to the AI. You can include:

• Personalization tokens from the enrolled record
• Instructions for the AI’s behavior
• Context about the task

Web Search (Optional)

Enable web search to allow the AI to find up-to-date information. This uses additional credits but is useful for:

• Current events or news
• Company research
• Market information
• Technical documentation

Output Fields

The action provides multiple output fields based on your selected output type:

• Text/HTML Output - For text responses
• Select Output - For select/multiselect responses
• Number Output - For numerical responses
• Boolean Output - For true/false responses
• Citations - Web sources used (when web search is enabled)

Example Use Cases

Lead Scoring

Analyze lead data and assign a score:

Analyze this lead and assign a score from 1-10:
- Company: {{company.name}}
- Industry: {{company.industry}}
- Employee count: {{company.numberofemployees}}
- Recent activity: {{contact.recent_activity}}

Consider factors like company size, industry fit, and engagement level.

Output type: Number

Content Classification

Categorize support tickets:

Classify this support ticket into one of the available categories:
Subject: {{ticket.subject}}
Description: {{ticket.content}}

Output type: Select, Target property: ticket_category

Email Summarization

Summarize recent communications:

Summarize the key points from recent communications with this contact:
{{contact.recent_email_content}}

Keep the summary to 2-3 bullet points.

Output type: Text

Competitor Analysis

Research competitor mentions:

Search for recent news about {{company.name}}'s main competitors in the {{company.industry}} space. Summarize any relevant developments.

Output type: Text, Web search: Enabled

Best Practices

1. Be specific in prompts - Clear instructions produce better results
2. Include relevant context - Provide the AI with necessary CRM data
3. Choose the right output type - Match the output to your workflow needs
4. Test with sample records - Verify results before activating workflows
5. Monitor credit usage - Check your usage in app settings

Troubleshooting

Empty or unexpected responses

• Check that personalization tokens have values
• Simplify your prompt
• Try a different AI model

Credits running low

• Review your plan in Settings
• Consider upgrading for more credits
• Optimize prompts to reduce token usage

Need Help?

• Contact support@daeda.tech