7 min read
Automating Outbound Sales Emails with AgentHub
A breakdown of how to send personalized emails at scale using AI.
Co-Founder/CEO
If you’re a recruiter or investor, you most likely have a database of potential candidates and their linkedin profiles somewhere. Let’s have AI enrich those leads for you like a hyper efficient employee.
January 17th 2024What if you could focus on finding the best candidates and let AI do the heavy lifting by analyzing their work history, summarizing their background, categorizing them into an industry and scoring them with custom criteria. That’s exactly what this automation does and it takes less than 5 minutes to build completely from scratch with AgentHub.
A LinkedIn Profile URL (must be a public profile. We do not spoof any LinkedIn credentials) either one at a time or in bulk through an Excel or Google Sheets file!
Ex: https://www.linkedin.com/in/max-brodeur-urbas-1a4b25172/
A complete summary of the candidate ready for inserting into an Excel or Google Sheets file!
Full name: Max Brodeur-Urbas Professional Background: Founder of AgentHub (YC W24), experienced in software engineering with a history of working at Microsoft and competitive programming leadership at McGill University. Industry: AI Score: 80
In this post I’m going to explain exactly how each step of this automation works so you can customize it to meet your exact needs. If you want to skip the technical side of things, scroll to the end of this post to start using the automation.
As a general overview of the automation, we scrape the persons profile, ask the AI a series of very specific questions using their profile as context and output the answers wherever you normally track your leads.
We use our web scraping node to read the profiles content in preparation for AI processing. This node takes in a link to any website and outputs text as if you copy pasted all the visible text on the screen.
Note: This does not work with private Linkedin profiles. We don’t use any fake credentials or bots to navigate the website. Just good old web scraping.
Each of these steps utilizes a different AI node on AgentHub. In case you’re not familiar, AgentHub is a no-code AI automation builder. Drag and drop your way to extremely powerful solutions. If you’re curious about what AgentHub is or why it exists check out this blog post.
Let’s start off with summarizing their professional background as this is the most straightforward step. We pass the contents of their profile into an “Ask AI” node on AgentHub and request that the AI generate a short one sentence summary of their background.
You'll notice the prompt being used is quite generic, feel free to add extra instructions or even an example summary if you want to provide the AI with extra guidance for formatting purposes.
Our scoring node forces the AI to output a number between 0 and 100 that it chooses based on custom criteria you define.
Here is our scoring description. Keep in mind it’s extremely generic as this demo automation was made for anyone to use. Add specific criteria that you’d want in a good score vs what would be missing to cause a bad one.
Categorization on AgentHub is a super popular feature. It allows you to define an arbitrary number of category labels + descriptions and have the AI select one for a given piece of content.
In this demo automation, I set the categories to common startup fields like AI, Biotech, Fintech etc. You can modify these to reflect the sorts of candidates you're expecting for your use case.
I wrote more about how categorization it works here
Last but not least, we’ll use the ‘Extract Key Info’ node to have the AI return values from the input text. In this case the value we’re having it return is the candidate’s full name. This is useful for bookkeeping purposes and for when you want to reach out to the candidate.
You could modify this node to extract any sort of info from the profile like their current job title, their current company name etc.
We have now finished extracting useful info from the Linkedin profile but we need to store it somewhere for future reference. For the sake of this demo we’ll store the info in Google Sheet as that’s the most popular flow. You could do any number of things with it however (email it, store it in airtable, store it in an excel file, post it to twitter...)
We’ll take the contents of all of these AI steps and pass them into the “Write to Google Sheets Node”. This demo is generating a new Google Sheet on each run (for the sake of letting people get started more quickly) but in practice you’ll want to specify an existing google sheet so it can keep adding to the same ongoing database.
If you have any questions or want a custom automation for your business email me at max@agenthub.dev, follow me on Twitter or join the AgentHub Discord
Here is some more content related to Gumloop, LLMs and automation. Enjoy!
7 min read
A breakdown of how to send personalized emails at scale using AI.
Co-Founder/CEO
5 min read
Tips for how to approach cost-effect data labeling with LLMs
Co-Founder/CEO
3 min read
The journey of building AgentHub, from discovering AutoGPT to creating a platform for reliable and cost-effective AI automation.
Co-Founder/CEO
Automate your complex business processes without writing a single line of code. No credit card required.