> ## Documentation Index > Fetch the complete documentation index at: https://docs.lusha.com/llms.txt > Use this file to discover all available pages before exploring further. # How to use Intent as a signal in your AI chats > Ask your AI assistant which accounts are actively researching your category, right inside the chat. > [!NOTE] > > **Audience:** End users who already have the Lusha MCP connected and want to surface Intent signals in natural language, inside Claude, ChatGPT, or any other MCP client. If you haven't connected yet, see [Lusha MCP Server](https://docs.lusha.com/user-guide/mcp/lusha-model-context-protocol-mcp-server.md). Intent shows you which companies are researching topics relevant to what you sell, sourced from Bombora's intent data and refreshed weekly. As a signal, Intent tells you what changed - when a company starts researching a topic, when it picks up, and when it fades - so you can prioritize hot leads as they emerge. This guide shows you how to ask for it through the Lusha MCP in Claude, ChatGPT, or any other MCP client. ## Before you start Connect Lusha to your AI assistant: * **Claude:** go to **Settings → Connectors**, add `https://mcp.lusha.com`, and sign in. * **ChatGPT:** find Lusha in the marketplace and start a chat. * **Other MCP clients:** use the MCP URL and your API key from API Hub. > [!TIP] > > For full setup, see the [Lusha MCP Server](https://docs.lusha.com/user-guide/mcp/lusha-model-context-protocol-mcp-server.md) guide. ## What is Intent as a signal? * **New intent topic:** the company started substantive research on a topic it had no activity on before (score reached 60 or higher). * **Intent score increase:** research on a topic increased versus the previous week. * **Intent score decrease:** research on a topic decreased versus the previous week, but is still substantive (score stays above 60). * **Intent topic removed:** research dropped below the activity threshold (60), so the topic no longer has a score. ## How to ask for it **1. Find your topics** Ask the assistant to search the topic catalog, for example: "What intent topics match 'sales engagement'?" Topics aren't a predefined list you set up in advance - search the full catalog of about 18,000 topics and pick any that fit, changing your selection on every request. This step is free. **2. Choose your companies** Name the companies or domains you want to check, or ask the assistant to find companies by filters such as industry, size, and location. **3. Ask for intent signals on your topics** Name the signal types and topics. Example: "For these 10 companies, show new intent topics and score increases on 'CRM Software' since last week." **4. Read the results** Each result shows the signal type, the topic, and the score - see [Read the results](#read-the-results) below. **5. Act on them** Save the results to a Workspace table, rank them, find the decision makers, or draft outreach. You'll know it worked when the assistant returns a list of companies with a signal type, topic, and score for each one - that's your confirmation the request matched real activity. ### Example prompts * "Which companies on my target list started researching 'Cloud Security' this week?" * "Show me intent score increases on 'Salesforce' and 'CRM software' for these domains since last week." * "Find companies in Germany with 50 to 200 employees, then check which show new intent on 'Data Integration'. Save them to a table." * "Which of my customers started researching a competitor's topic? Show the topic and the score change." ## Read the results * **Score:** from 60 to 100. Signals only appear once a company's research is substantive (60+); within that range, a higher score means more research activity than usual. * **Previous score:** the score for the same topic the week before. * **Change rate:** the percent change versus the previous week. Score increases and decreases carry it. * **Topic removed:** the topic fell below the activity threshold (60), so it has no score. > [!NOTE] > > Intent refreshes weekly, so check once a week. A rising score on a new topic is a stronger cue than a single high score. Intent is company-level - it shows what an account researches, not which person. ## Combine it with other signals * **Signal score:** one 0 to 1 score per company or contact from its active signals, to rank a table. Try asking: "Score these companies by signal activity and show the top 10." * **ICP fit score:** whether the account matches your Target Profile, so you act on fit and intent together. Try asking: "Score these accounts for ICP fit, then keep the ones with new intent on 'CRM Software'." * **Hiring surge and headcount growth:** budget and expansion behind the research. Try asking: "Which of these companies show intent and are also hiring in sales?" * **Funding and company news:** a timing trigger to open the conversation. Try asking: "Show intent increases and any funding news in the last 90 days." * **Job changes:** re-engage past buyers who joined an account that is now researching your topic. Try asking: "Which accounts with rising intent also have a past contact who changed jobs?" * **Lookalikes:** more accounts that resemble the ones showing intent. Try asking: "Find companies similar to these 5 accounts showing intent." * **Decision makers:** the buying group at each in-market account. Try asking: "Find the decision makers at the top 5 accounts by intent." * **Website visits:** first-party interest from companies visiting your site, next to third-party research. Try asking: "Which companies with new intent also visited our website?" ## Tips * Pick topics that describe what you sell and what your buyers research. Start with one to three - you can always search and add more later. * Look at the trend, not just the score. A topic climbing week over week matters more than one that is flat. * Test on a small batch first, such as 25 companies, before running a full list. * Ask for intent signals by name - they aren't included when you ask the assistant for all signals. ## Good to know * Filtering works by topic, not by category. * Intent signals are weekly, so results change week to week. * Signals only fire once research reaches a 60+ score; activity below that isn't surfaced. > [!NOTE] > > 💡 Billing is success-based - credits apply only to intent signals actually found, never to a topic with no match. Each topic costs 1 credit per company signal found; name the topics you care about instead of pulling everything, since every signal returned is charged. Ask the assistant to check your remaining credits and rate limits any time. You're all set - next time you open the chat, ask which accounts are heating up this week. ## Related articles * [What is Lusha Intent](https://docs.lusha.com/user-guide/intent/what-is-lusha-intent.md) * [Lusha MCP Server](https://docs.lusha.com/user-guide/mcp/lusha-model-context-protocol-mcp-server.md) * [Signal Score and ICP Fit Score](https://docs.lusha.com/user-guide/ai-recommendations/signal-score-and-icp-fit-score.md) ## Related topics - [What is Lusha Intent](https://docs.lusha.com/user-guide/intent/what-is-lusha-intent.md) - [How to use the AI chat to build a table](https://docs.lusha.com/user-guide/building-your-workspace-tables/how-to-use-the-ai-chat-to-build-a-table.md) - [How AI Recommendations work](https://docs.lusha.com/user-guide/ai-recommendations/how-ai-recommendations-work.md) This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.