Learning Lindy

Image Recognition

Learn how to use Lindy's image recognition capabilities by building a Networking Follow-up Assistant.

In this guide, you're going to learn how to leverage image recognition to automate networking follow-ups in less than 5 minutes. The Lindy you're going to build will act as a Networking Assistant, streamlining the whole networking follow-up process: from extracting contact information from business cards to performing contact research to updating your CRM and sending personalized follow-ups.

Whether you're at conferences, networking events, or business meetings, Lindy handles the administrative work so you can focus on building meaningful professional relationships.

Here's what this Lindy will do:

Extract contact information from business cards using image recognition
Research new contacts across the web to gather professional insights
Create and update CRM records automatically
Draft personalized follow-up emails incorporating discovered information
Provide a seamless system for managing new professional connections

Let's get started!

Create a New Lindy

Click New Lindy below Home in the top left of your workspace to begin building your Lindy.

Click "Start from scratch" to create a new Lindy.

Add a Trigger

Click "Select Trigger", then click the Chat trigger and select Message Received.

Triggers are events that "wake up" a Lindy. Examples include time-based triggers (e.g., every weekday at 9 am) or event-based triggers (e.g., when you receive an email).

We’re selecting the Chat trigger because to initiate this workflow, we will send Lindy an image or screenshot in Lindy’s task view. Lindy will then automatically use image recognition to analyze the content and take proceeding actions.

Add a Browser Search Action

Now that Lindy can receive business card images, we'll set up research capabilities so Lindy can gather more information about your new contact.

Select "Perform an Action".

Click the Browser action, then select Search the Internet.

Set the "Search Query" field to “Prompt AI” mode and add this prompt:

Search for professional information about [PERSON'S NAME] at [COMPANY].
Look for:
- Recent company news or achievements

You can customize the browser search to look for specific things you want to research.

Some examples are:

- LinkedIn profile information
- Current role and responsibilities
- Professional background

In addition, you can select a specific search engine for your query; however, for this Lindy, we’ll leave the “Search Engine” to “Auto” mode – enabling Lindy to select the best search engine for the job via her intelligence.

After this action, we are going to have Lindy extract the content from the most relevant URL found from this initial search step.

Click “Add Step” and select “Perform an action”.

Click the Browser action, then select Extract URL Text.

Set the "URL" field to “Prompt AI” mode and add this prompt:

The most relevant URL found from your search

Create a CRM Contact

Now we're going to make Lindy do the most tedious part - creating detailed CRM records from business cards!

Click "Add Step" and select "Perform an Action".

Search for HubSpot (or your preferred CRM - Salesforce, Airtable, etc.) and select Create Contact.

Authenticate your CRM account.

You'll need to configure several fields to ensure Lindy creates detailed contact records:

Required Fields

The Email field is required - Lindy needs this to create the contact in your CRM. Switch this field to "Prompt AI" mode and use a prompt that refers to the business card:

Email address extracted from the message received

Contact Properties

For each contact property, you have three options:

Prompt AI:
- Best for extracting specific information from the business card or web research
- Examples:
  - First Name: Extract the first name from the business card‍
  - Last Name: Extract the last name from the business card‍
  - Company: Use the company name found on the business card‍
  - Job Title: Extract the job title from the business card
  - Summarized web research you found about the contact

Auto:
- Lets Lindy intelligently fill fields based on all available information
- Useful for data that might be mentioned in a contact’s online presence
- Example: Industry, Company Size, Social Media Links

Set Manually:
- Best for fields that should have the same value for every new contact
- Example: Lead Source could be set to: Networking Event

Draft a Follow-Up Email

Finally, we'll have Lindy send a personalized follow-up email to the contact using the gathered information.

Click "Add Step" and select "Perform an Action".

Choose Gmail (or your preferred email service), then select Send Email.

Within the Send Email action, you can set Lindy to “Ask for confirmation” or have her draft the email and “Save as draft”, offering you the chance to review and approve before sending.

Configure the Send Email action to suit your preferences, example below:

For the “To” field, switch to “Prompt AI” mode and use the prompt:

email extracted from the message received

For the “Subject” field, switch to “Set Manually” mode and use the prompt:

Great connecting just now!

For the “Body” field, switch to “Prompt AI” mode and use the prompt:

Draft a warm follow-up email that:

- References where/how we met today (if unknown, just say: “It was awesome to connect!”
- Mentions 1-2 relevant details from my research about them or their company
- Keeps the tone professional yet friendly
- Is concise, easy to read and skimmable
- Ends with "Let me know if there is any way I can be of value to you,"

Run The Workflow

Click the "Tasks" button to access your Lindy's task view.

Drag and drop a business card image onto your Lindy to test the workflow. (hit enter to send)

You will see Lindy:

Extract information from the business card
Perform web research about the contact
Create a new HubSpot contact
Draft and send a personalized follow-up email

Watch the execution in real-time in the Tasks view and review each step's output.

Congratulations!

You've just built a Lindy that turns a business card photo into an automated networking follow-up workflow.

What used to take 15-20 minutes of manual work now happens in seconds. Try using these same image recognition principles to process receipts, analyze documents, or handle any other image-based tasks!

Guides

Medical Scribe

Building a Chatbot: 101