Unlock the Potential of GPT-4: A Comprehensive Use Case Framework for SaaS businesses
Think Microsoft copilot is great? Here are 8 ways you can start implementing GPT-4 in your SaaS application.
After witnessing the series of announcements in the past two weeks, you might’ve realized like other technology waves we have seen it’s not just a hobbyist tech but something that can create tremendous value for a businesses around the world.
I’m sure like many SaaS companies, you are also looking to see what super powers ChatGPT (or other LLM framework) can give to your application.
If you are thinking along the lines, you have come to the right place. In this blog, we explore the universe of uses cases of ChatGPT-4 can address in an enterprise application and trim it down to a simple framework which you can take back to implement ChatGPT-4 right away in your enterprise SaaS application.
Personalized discovery
It’s just a fancy word for chatbot. But the intention is to forget the chatbot experience we usually think and instead look at the possibilities where a user can discover information gradually as they interact with the chatbot.
There are several types in which we can include this in an enterprise application.
1. Synthesize information
This is where you feed in all your data to ChatGPT-4 or LLM model, it synthesizes the information and presents the most relevant information to the question asked by the user.
This is exactly what Stripe and Morgan Stanely have done in their implementation. Feed in their data (knowledge base) and let user interact with the bot and respond in a concise manner.
GPT is able to understand the user’s question, read detailed documentation on their behalf, identify the relevant section and summarize the solution in one place.
Sample use cases:
Any SaaS or business app - Support queries for how-tos, examples, step by step instructions, common errors, community answered questions, etc.
Help desk/IT service desk app - Embedded chat bot can read through the customer’s knowledge base and suggest answers to their users.
Employee experience apps (HR,IT) - Embedded chat bot can look at various HR policies, org structure, benefits insurance policies, IT policies and guide the employees through various queries.
2. Co-related information
The only reason we are keeping this as a separate category is because, we are going outside of enterprise data and tap into the wider information set that’s publicly available in the internet. It’s best explained with examples and use cases.
Sample use cases for various kinds of enterprise SaaS application:
CRM application
Know more about a company - When working on a certain client, you want to know more about their company, their headquarters etc. This is a “related” information which can be gathered from publicly available data such as their websites, news coverage, stock coverage, etc.
Know more about the contact - If their name and company are available, we can showcase the summary of the information from their LinkedIn profile, blogs, or mentions in different websites.
IT service management
Increase reach of Knowledge base - User will be able to search common Windows, Apple or Cisco problems and tap into support websites of community of these vendors and provide the required solution.
3. Self-service analytics
Where we can get insights into the application data. Normally we would rely on dashboards or analytics tools to get the information, but with Chatbots this information is reachable to any user.
Sample use cases:
IT Operations Management
What is the uptime of this server? (direct data)
How many servers have CPU usage more than 85%? (Derived data)
Tell me the list of servers for the above with their CPU usage (JSON > Table)
HR tools
Tell me the number of people whose work log is less than 4 hours for more than 10 days a month (Derived data)
Show me the list of people who haven’t completed the security awareness training during their onboarding (JSon > Table)
ERP application
How many vendors haven’t paid yet? (Derived data)
Which vendors contracts are up for renewal in the next month? (JSON > Table)
Marketing tools
How many leads did we get this month? (Direct data)
What % of MQL were we able to convert to SQL this month? (Derived data)
Compare the performance of index page vs products page with the following parameters: visits, Bounce rate, time on site, Leads (JSON > Table)
4. Summarize/Transform
While this is similar to synthesize information, the reason for keeping it as a separate category is that use cases we can develop are slightly different. It provides the ability to summarize a series of long and contextual/specific information. Essentially transforming paragraphs of text into bullets or single paragraph or the other way around.
Sample use cases:
CRM - Provide the summary of the last few conversations with the contact
IT Operations - Provide the timeline of the incident from the conversation which can be used in the retrospective
Word/Docs/Show - Provide a 4 line summary of a particular document which can be appended as a reference. (Already available in Google and Microsoft)
Meeting/Communication - With speech to text, we can summarize meeting discussions and append it to MoM.
Help desk - Summarize this ticket (with 30+ threads) and tell me what is being discussed.
Generate
This is more consumeristic aspect of ChatGPT where we have seen several instances of tech twitter consuming ChatGPT to write code, develop applications, create images, or help them write a cold email to a prospect, etc. Let’s explore each of them.
5. Generate text
Big opportunity here is most of us aren’t great writers. And in a office space, we have to write a lot. Be it writing to your boss, or customer or HR. Official communication or communication for hobbyist club. It’s good to have a AI tool in handy right within the application to help generate necessary text.
Sample use cases:
CRM app - Write a cold email to a prospect highlighting the benefits of the tool and how it can help their business.
Marketing app - Give me 5 topic suggestions for the latest product update as shared below (copy paste product feature)
Help desk - Write a apology mail to the customer for no resolving the issue on time
ERP - Write a proposal letter to the vendor about our intention to partner with while highlighting how both of us can benefit from the deal
Productivity (Docs, Mail, Show) - Generate a template for PRD, Generate template for security notifications, etc.
6. Generate code
The ones who know to code are using ChatGPT to increase their productivity. And the ones who don’t know to code are creating applications in record time. The value we are trying to unlock in enterprise applications is making pro-code “customizations” a lot easy to consume.
Often users rely on their software engineers/developers to build a integration or customization. With ChatGPT, they cannot only generate sample code and get instructions to use it in the application but ChatGPT can also help debug their code.
Unlike other use cases, this one cuts across any type of enterprise applications as the users here are application users who are looking for advanced customizations and it highly depends on how customizations are built in every application. A lot of professional services around the globe are now going to be able to churn out their services in record time.
There’s also another set of tools that serves developers like Github and replit where copilot helps in increasing developer’s productivity. It’s a deep rabbit hole if we get into this. But significance of generating code is far more than no-code revolution.
7. Generate Images
Well ChatGPT-4 powered by DALL-E can generate images from a text prompt. When it comes to an enterprise a lot of money is spent on getting “stock images” or designing unique creatives. All of these cost a lot of money and effort. Canva made everyone a designer, DALL-E is making everyone an artist without brush or tool. There’s a big difference. Here we aren’t talking about the scale of creativity that Midjourney or Adobe Firefly enables, but something that’s more mundane but time sensitive and is used by non-designers.
Sample use cases:
Marketing tools: To create image social media post or blog post with brand colors
Presentation tools: To create an image for a slide (Microsoft goes to an extent of creating a whole deck itself with a single prompt)
Designing tools: To create an image for HR safety week poster
Wireframing tools: To create wireframes from text with the design system
8. Generate API-based actions
This is where we are talking about making ChatGPT more contextual to the application and the action(s) we can perform within an application. ChatGPT just released their plugins and if you see the some of the usecases it enables for their launch partner it’s incredible. For example, Zapier, integration as service platform, will enable you to perform an action in a different product just through few texts.
Sample use cases:
CRM application - Get me a list of companies in Florida which have employee count of < 100 along with the count, contact and create a button which will create a contact in my CRM application. Imagine getting Forbes list and filtering through them.
Help desk application - Get me a list of customers whose sentiment is < 50%, their subscription expiry date and create a button to trigger an email to give them 10% discount (coupon code:WISHES10) in their next purchase.
Productivity - Create a meeting with all the people in this chat group for a discussion for next week. Populate the meeting agenda from this doc.
Source: Zapier
We will get into the depth of use cases for specific app in the next few posts. Till then you can you can start exploring these eight avenues to build unique GPT-4 use cases for your SaaS application.
Found it interesting? Do share this post.