Semantic Search and Content Moderation for Slack

Imagine you're part of a fast-growing team, and like many modern workplaces, you rely heavily on Slack for day-to-day communication. As your team scales, two challenges become increasingly apparent. First, you find yourself spending more and more time scrolling through Slack to find that one crucial message or file. It's like looking for a needle in a haystack. 

Second, as conversations multiply, it becomes harder to maintain a positive and respectful communication environment. Inappropriate or toxic messages can slip through the cracks, affecting team morale and potentially causing larger issues down the line.

Sound familiar? You're not alone. 

In this blog post, I'll walk you through how to use Graft's Modern AI Platform to add both semantic search and content moderation to Slack. By the time you finish reading, I'll demonstrate how Graft makes it easy for any team to enhance their Slack experience.

Be sure to check out our Deep Dive into AI-Powered Semantic Search.

How to Add Semantic Search and Content Moderation to Slack with Graft

Let's explore how one team used Graft to unlock hidden insights and safeguard their Slack community. By the end, you'll understand how Graft puts advanced AI to work for your business.

With just a few clicks, you'll be able to find meaning in message data and promote respectful communication norms.

Step 1. Connect Slack

First things first, you'll want to connect your Slack workspace to Graft. This is a straightforward process, and it's the foundation for everything else we'll do. Here's how to do it:

1. Go to Create Data Source
2. Select the Slack Connector
3. Follow the prompts to connect Graft to your Slack workspace.
4. Choose the channels you want to focus on. You can select all of them or just a few, depending on your needs.

And just like that, you've set up the initial connection!

Graft will pull in both structured and unstructured data from Slack—everything from text messages to photos, videos, and audio files.

Step 2. Link Data Together by Creating an Entity

Now that your Slack is connected, you'll want to create an entity in Graft. This entity keys in on the client message ID and picks the best foundation model and embeds all of the resulting data.

Entities contain all of the data related to your use cases. This includes the ingested raw data from one or more data sources, embeddings from one or more fields, saved sql queries, visualizations, custom Enrichments (classifiers) and their predictions.

1. Go to the 'Entity' section in Graft.
2. Follow the pre-built template to select the field

Step 3. Add Toxicity Enrichment for Content Moderation

Now, let's add some enrichments to your Slack data. This is where we'll implement the content moderation feature to flag or filter out inappropriate or toxic messages.

1. In your entity, look for the 'Add Enrichments' option.
2. Select 'Toxicity Scoring Enrichment' from the list.
3. Confirm that the enrichment has been applied to your Slack messages.

Step 4. Schedule Your Workflow

You've got your data embedded and stored in a vector database and a toxicity score on all of those messages. 

Now you'll want to schedule this workflow to run at intervals that make sense for your team. Choose how often you want the workflow to run—be it once a day, every hour, or even every minute.

5. Access Results

Finally, let's get to the part you've been waiting for—semantic search. Graft offers a variety of ways to access this feature like APIs, but for this guide, we'll use the SQL interface.

1. Go to the 'SQL Interface' in Graft.
2. Use the query builder to select the 'Slack Message Entity' and perform a 'Similarity Search.' You can also write your own SQL queries.
3. Run your query to find the Slack messages most closely related to your search term.
4. You can allow API access with a click of a button.

And there you have it! By following these steps, you've not only made your Slack experience more efficient but also more respectful and aligned with your company's values.

Graft didn't just call some embedding model or create a toy implementation. It built a fully functional production implementation that you can interact with directly via the UI or use the APIs to wire up with a production use case.
~ Adam Oliner, CEO & Founder of Graft

And the best part? You didn't have to build or maintain any infrastructure, write any code, or worry about all the ugly details of foundation models or embeddings.

Real-World Use Cases

1. Meet Katie, the Community Manager

Katie is the community manager at a fast-growing blockchain startup. She oversees a Slack community with over 5,000 highly engaged members. Katie needs to closely monitor conversations to foster a constructive and positive environment.

One day, a member named Sam posted feedback about a new crypto wallet feature. Katie wanted to reference Sam's comments in an upcoming team meeting but couldn't find the message in the crowded #product-feedback channel. Using Graft's semantic search, Katie quickly located Sam's message by searching for "wallet feedback."

Later that week, Graft's content moderation flagged an inappropriate meme that a new member posted. Because the toxic content was automatically detected, Katie could promptly remove the meme and send the new member a reminder about community guidelines.

2. Get to Know Michael, the HR Director

Michael leads HR for a retail company with 700 employees across 4 states. He relies on Slack to connect with individual locations. Recently, an insensitive comment from a regional manager spurred several employee complaints.

Thanks to Graft, Michael was notified of the toxic message as soon as it was posted. He addressed the situation directly with the manager and avoided a larger controversy. Michael also uses Graft's semantic search whenever he needs to reference past conversations with employees or managers.

3. Meet Aisha, the Sales Representative

Aisha manages enterprise sales for a software startup. She frequently collaborates with clients in Slack channels. When an important partner mentioned budget constraints, Aisha wanted to revisit their previous pricing discussions. Rather than combing through thousands of old messages and threads, she quickly found the relevant history with a semantic search for "budget."

The next week, one patron posted an inappropriate GIF. Aisha appreciated Graft's immediate content moderation alert. She removed the GIF and gave the client a polite reminder about professional conduct in partner channels.

AI that's Customizable and Accessible to Your Team

At Graft, our mission is to make powerful, sophisticated AI accessible for organizations of all sizes. We let users focus on solving real problems rather than getting bogged down in data and infrastructure complexity.

In this blog, I demonstrated how easy it is for you to enrich their Slack experience with semantic search and content moderation using our platform. With just a few steps, you can unlock hidden insights, improve information discovery, and promote healthy communication norms.

The benefits extend far beyond Slack. Graft offers a gateway to harnessing the potential of all your unstructured enterprise data. Our Modern AI Platform seamlessly integrates across data sources and business workflows.

You don't need an army of data scientists or machine learning engineers to start seeing value. We handle the heavy lifting behind the scenes so you can focus on your use cases. Our intuitive interface makes it simple to tailor robust AI to your unique business needs.

I'm proud to be part of a team that turns advanced technology into accessible solutions for our customers. If you're ready to stop leaving your data's potential untapped, contact us for a free trial today. Our experts will partner with you to maximize the impact of AI in your organization.

The future of business is AI-driven. With Graft, you’re ready for AI.