AI Annotations

Overview

AI-powered annotations for table records in Salesnode allow you to automatically enrich your contact data with insightful, AI-generated information. This feature helps you save time by generating annotations such as personalized notes, company research, or enrichment data directly within your custom tables. By integrating AI annotations into your workflow, you can enhance your outbound campaigns with relevant and dynamic content tailored to each contact.

Setting Up AI Annotations for Your Table

To get started with AI annotations, you first need to select the table and configure the annotation parameters. This setup defines how the AI processes your data and where the results are stored.

1. Navigate to the Sidebar and click on Tables.
2. Select the table you want to annotate.
3. Click the Annotations button located in the top bar.
4. Click Add annotation to open the configuration panel.

Configuring Annotation Details

You will need to specify several key settings to tailor the AI annotation process:

• Name: Enter a descriptive label for this annotation to identify it easily.
• Provider: Choose the AI service provider you want to use, such as OpenAI, GroqCloud, Anthropic, Google AI, or Mistral AI.
• AI Model: Select the specific large language model (LLM) offered by your chosen provider.
• Web Search: Enable this option if you want the model to use web search during generation. Note that only certain models support this feature.
• Target Column: Define the table column where the AI-generated annotation will be stored.
• Include Blank Only: When enabled, the annotation will only process records with an empty target column, preventing overwriting existing data.
• Skip Records with Empty Placeholders: Enable this to skip records missing values for placeholders in your prompt, ensuring only complete records are processed.

Crafting Prompts with Table Data and Placeholders

The prompt you create instructs the AI on what kind of annotation to generate. Using dynamic placeholders, you can tailor the prompt to each record’s specific data, making the output highly personalized and relevant.

Using Placeholders in Prompts

Placeholders are inserted using LiquidJS syntax, referencing table columns with the prefix record. This allows the AI to access the values from each row dynamically.

Examples of placeholders include:

• {{record.first_name}}
• {{record["company name"]}}

To insert placeholders easily:

1. Click inside the prompt editor.
2. Press {{ to open a suggestion menu listing all available table columns.
3. Select the desired column to insert its placeholder.

The prompt editor also supports advanced LiquidJS features and functions, enabling you to create complex and conditional prompts as needed.

Previewing Prompt Output

Before saving your annotation, you can preview how the AI will interpret your prompt:

1. Click the Preview button at the top right of the prompt editor.
2. A modal will appear showing sample records with the generated markdown output.
3. Invalid placeholders (those without matching columns) will be highlighted in red.
4. If “Skip records with empty placeholders” is enabled, records missing placeholder values will be excluded from the preview.

Managing Annotations and Execution

Once your annotation is configured, you can control when and how it runs, as well as review the results directly within your table.

1. Use the toggle in the top bar to activate or deactivate automatic annotations. When active, new records inserted into the table will trigger the annotation automatically.
2. For existing records, use the ▶ play button in the top bar to manually execute the annotation in bulk.
3. Review and update generated annotations directly in the table view to ensure data quality and relevance.

Using Annotations in Campaigns

AI-generated annotations can be powerful inputs for your outbound campaigns. For example, you can:

• Personalize email first lines using enriched data.
• Include company research notes to build rapport.
• Add enrichment data to segment and target contacts more effectively.

Best Practices for Effective AI Annotations

Creating effective AI annotations requires thoughtful prompt design and clean data. Follow these guidelines to maximize the impact of your annotations:

1. Use Soft Messaging: Avoid direct sales pitches in prompts; instead, focus on addressing pain points or relevant insights.
2. Keep Prompts Clear and Concise: Clear instructions help the AI generate accurate and relevant annotations.
3. Validate Placeholders: Ensure all placeholders correspond to existing table columns and contain meaningful data.
4. Test with Previews: Use the preview feature to catch errors and refine prompts before running annotations.
5. Monitor and Update: Regularly review annotation outputs and update prompts or settings as needed to maintain data quality.