AI Variable Overview
Learn how to use AI Variables to automatically research and qualify contacts and companies. Then reach out to engage — all on Genesy.
AI Variables let you add automated research columns to your lists so each contact or company row can be enriched with AI-generated insights (Key pain points, Buyer persona, short summaries, Job title review, outreach hooks, etc.). Use templates or build your own — then run them in bulk and store results as columns on your table.
What is an AI Variable?
An AI Variable is a reusable prompt + settings bundle that tells Genesy how to research and return a specific piece of information for a Contact or Company. Think of it as a smart column: instead of manually researching, the AI reads the attributes you provide (name, domain, job title, LinkedIn, etc.), optionally searches external sources, and returns the answer in the format you want.
Use cases:
Contact: "Best outreach icebreaker (1 sentence) for {{first_name}} based on their recent activity."
Company: "1‑line summary of company growth signals (funding, hiring, product launch)."
How to run an AI Variable?
Import your contacts or companies into Genesy.
Open the list where your rows live and Click Enrichment > Enrich with AI in the top-right of the list to open the AI enrichment modal.
3. Select one (or multiple) template(s) — either a Genesy ready-made template or one you've built — then click RUN.
4. The result will appear as a new column in your table for each row processed.
Shortcuts:
Hover between two columns, click the Add a column (+) icon, and choose one of the AI Variables from the dropdown.
When the column is empty, hover the column header and click RUN to execute the variable for all rows in that view.
Bulk runs:
Click the column header → Run column → Visible rows / All rows.
Managing AI Variables
All AI Variables are managed from the AI Playbook page. From there you can:
Create new variables
Test a variable
Edit prompts and settings
Duplicate templates (fast reuse)
Delete unused variables
How to create a new AI Variable
To create an AI Variable, you’ll need to fill the form fields below:
Entity type (required) — choose Contact or Company.
Title — a short name that describes the output (e.g., "Outreach Hook — LinkedIn", "Company Growth Signals").
Output (response type) — select how you'd like the AI to return results. (tetx, number,etc).
Prompt — the main instruction. This is where you tell the AI exactly what to return and the format. Use examples and explicit output rules. See templates below.
Add attributes — select which row fields the variable can use as context. Common placeholders:
{{first_name}},{{last_name}},{{job_title}},{{email}},{{phone}}{{company_name}},{{domain}},{{website}},{{company_size}},{{industry}},{{hq_location}}{{linkedin_url}},{{recent_activity}},{{lead_source}}
Use these inside prompts to make the AI answer relevant to each row.
Search & sources — enable optional external research (Google Search, Google News, Website). If the modal's default sources don’t find an answer, you can toggle additional sources
Select folder (optional) — organize your variables into folders.
Select model — pick a model (e.g., Grok‑4, o3)
Save & Test — before saving, click the Test Prompt button to run it on a sample contact or company record.
Qualification criteria and response types:
Take the following example:
The instruction is:“Tell me if the company is either a B2B, B2C, or B2C with B2B products.”
Here, the Output type is set to [One of], which restricts the AI’s response to a predefined list of options. In this case, the AI will only return one of the following values:
B2B
B2C
B2C with B2B products
This ensures consistency, makes results easier to filter and analyze, and prevents unexpected or vague answers.
👉 Use [One of] whenever you want the AI to choose from a closed set of answers (e.g., Yes/No, High/Medium/Low, Hot/Warm/Cold).
Response types:
We currently support the following response types:
Text: For open-ended answers, such as short descriptions, summaries, or hooks.
Number: Returns a numeric value (no formatting). Ideal for scores, rankings, or counts.
Date: Outputs a valid calendar date. Useful for events, deadlines, or timeline extraction.
One of: Restricts results to a predefined set of options (e.g., Yes/No, Hot/Warm/Cold).
Keep in mind that the response type will affect the criteria you can set up.
You can also activate the “Provide explanation” option. When enabled, the AI will return both the selected result and a short reasoning behind it, giving your team more context.
AI Variables best practices
For the best results, follow these guidelines when creating and running AI Variables.
Prompting best practices
Be specific and concise — Clearly define what the AI should return and avoid vague asks.
Include context — Add why the information matters (e.g., for lead qualification or market research).
Leverage attributes — Use placeholders like
{{company_name}},{{domain}},{{industry}}, or{{job_title}}to personalize and contextualize the output.Test before scaling — Always run the AI Variable on a sample row, review results, and refine prompts until outputs are stable and consistent.
Examples of Effective Prompts
✅ “Classify {{company_name}} into one of the following categories: Enterprise, Mid-market, or SMB. Return only one of these.”
✅ “Based on {{company_name}}’s website and {{industry}}, summarize their core value proposition in one sentence.”
✅ “From {{domain}} and {{industry}}, list up to 3 potential business challenges the company may face. Return as a bullet list.”
✅ “Review {{company_name}}’s website. Identify if they primarily sell products, services, or both. Return exactly one of these options.”
❌ “Tell me everything about this company.” (too vague, no structure)
Ready-to-Use Best Practices Template
Persona:
You are a market research analyst at a SaaS company called Genesy, which helps sales teams qualify leads and companies more effectively using AI Variables.
Task:
Your task is to return a structured, insight-rich analysis about a given company using the provided data points. Always keep the result concise, actionable, and aligned with sales qualification needs.
Instructions:
Be precise — clearly describe the company’s profile using the available attributes.
Use placeholders — {{company_name}}, {{domain}}, {{industry}}, {{hq_location}}, {{company_size}}.
Add context — highlight details that matter for lead qualification (e.g., company size, funding stage, growth signals).
Set constraints — limit the response to exactly 3 lines OR output as structured JSON (depending on use case).
Return only the requested output. No extra commentary or formatting.