Your SaaS startup is growing. So is the number of support tickets. You want to save time on customer support, and know that AI tools have the capacity to help you scale here.

But you also care deeply about customer experience, and don’t want to compromise on support quality just to get through tickets faster.

If you’re curious about using Open AI/ChatGPT for customer support, this guide will show you:

• The difference between the ChatGPT API and the ChatGPT app
• How ChatGPT can help your customer support team
• How to prepare for using AI in customer support
• Why Outverse might be a better solution for your use case

ChatGPT API vs ChatGPT app

There are two ways you can use ChatGPT for customer support: via the ChatGPT API, and the ChatGPT app on OpenAI's website.

If you want to integrate ChatGPT into your product, you need to use the API; if you want to perform simple, isolated tasks (e.g. translate an email or get advice on CX reporting frameworks), you can use the ChatGPT app.

When customer service leaders talk about using ChatGPT for customer support, they're typically referring to the API.

ChatGPT API for customer support

The ChatGPT API allows developers to integrate ChatGPT into their own products.

You can use the ChatGPT API to:

• Build a customer support chatbot
• Automate your CX workflows
• Provide personalized recommendations to customers
• Generate reports, summaries, and other content based on your data

This guide won't go into detail on the setup — it's a significant amount of work, much of which is technical — so we'd recommend getting this information directly from OpenAI, starting with this blog post on the ChatGPT API.

A custom ChatGPT integration gives you a lot of flexibility: it can handle account-based tasks like payments, renewals, and cancellations.

That said, if your use-case is simpler and you're happy to handle account-based queries yourself, this configurability can actually be a drawback: there's a lot of setup- and maintenance-related overhead that might not be necessary, depending on what you're looking to achieve.

What CX workflows can you automate with ChatGPT?

ChatGPT can help streamline processes, reduce manual workload, and automate day-to-day processes for customer service teams.

Ticket triage

Managing a large volume of customer inquiries and support tickets can be overwhelming for support teams. By integrating the ChatGPT API into your ticketing system, you can automate the triage process, categorizing and prioritizing incoming tickets based on their content and urgency.

Customer follow-ups and reminders

The ChatGPT API can automate follow-up communications, such as post-purchase emails, appointment reminders, or satisfaction surveys.

Personalization

You can use ChatGPT to deliver a more personalized, useful experience to customers without human interaction. ChatGPT can analyze customer data and interactions to generate tailored recommendations, such as product suggestions, content recommendations, or troubleshooting tips.

In the example above, Shopify's consumer app, Shop, uses personalization to help people find products they might like.

Reports and summaries

With a ChatGPT integration, you can analyze customer data trends to generate predictive insights, such as identifying emerging customer pain points or predicting future support volumes.

The ChatGPT API learns and improves over time through feedback and iteration. By collecting data on customer interactions and refining the model based on real-world user interactions, you can continuously enhance the accuracy and effectiveness of your automated CX workflows. This iterative approach allows you to be more responsive to evolving customer needs.

Benefits of using ChatGPT as your customer service assistant

The benefits of using ChatGPT as your AI customer assistant are similar to using any LLM (large language model) powered artificial intelligence assistant: it allows you to deliver great customer service at scale without increasing workload for your customer service team.

Instant responses, 24/7

With the ChatGPT API powering your customer service chatbot, your customers can get instant responses to their queries round the clock. This increases support availability for other timezones, and — if the responses are helpful — can provide a quick, highly satisfactory experience.

Scalability

ChatGPT's ability to handle a large number of concurrent conversations means that customer queries are addressed promptly, even during peak periods.

Consistent responses

As humans we all have different perspectives, different experience, and different approaches to problems. Even the same human will approach things differently depending on variables like workload, priorities, or even the kind of day they're having.

This means responses from human agents can vary, even within a highly aligned team. You'll get much more consistency from ChatGPT — responses are always based on the provided training data and fine-tuned parameters, and adhere to brand voice and standards.

Cost-efficiency

Integrating ChatGPT can reduce the operational costs associated with customer support, and handle simple tasks and queries so your customer service teams don't have to. Instead, they can use their time responding to the more interesting, complex inquiries that actually need human intervention.

That said, there is an initial time investment to getting set up with the ChatGPT API, as well as ongoing maintenance and improvements.

Vertical AI customer support tools like Outverse, that are built to solve a specific purpose, are much simpler to set up and maintain, as you don't need to worry about configuration or maintenance.

Multilingual support

The ChatGPT API supports multiple languages, allowing your support chatbot to communicate fluently with customers worldwide, irrespective of language barriers. Klarna's AI assistant, for example, is available in in 23 markets and can speak more than 35 languages.

How the ChatGPT app can help customer service teams

‍

‍

First off, we'd recommend paying for a ChatGPT Plus subscription ($20/month) for access to GPT-4, which is much more accurate and useful than the free version — you can build custom GPTs to help you with specific tasks, and have a library of other GPTs to choose from. Read more on custom GPTs.

While the ChatGPT app can't directly assist your customers, you can paste customer messages into the chat and ask for advice on how to respond or interpret the message.

ChatGPT as a sounding board for human customer service agents

The ChatGPT app can act as a collaborator to your customer support agents by providing feedback and alternatives on difficult customer interactions.

It can advise on how to respond to customer complaints in a way that's empathetic and honest, and it can help you manage customer expectations.

If you use the ChatGPT app to draft messages in this manner, we'd always recommend sense-checking and editing for tone (and to make it sound more human) before sending on to a customer.

Use ChatGPT to handle translations for your customer service team

ChatGPT can help you provide multilingual support to customers. This is particularly useful for companies who don't yet have the support volume to warrant hiring support agents who speak multiple languages, but still want to respond to customers in their native language.

Create templates and frameworks

ChatGPT can generate feature request templates, build customer surveys, and suggest customer support or measurement frameworks for customer service teams to use.

Do you actually need to configure your own ChatGPT setup?

If you're looking to automate simple workflows and provide customer support via an AI customer support assistant, building your own solution with ChatGPT may be overkill for your use-case. There are plenty of purpose-built tools that can help you achieve these goals without complex setup or sprawling costs.

Using vertical AI tooling for customer support means you don't need to worry about technical implementation, development, or maintenance. With Outverse (pictured), you can deploy an LLM-powered AI customer assistant — that can provide answers from across your knowledge base, customer forums, and changelogs — without any technical know-how.

Preparing to introduce AI into your customer support stack? Here's what to do first

Update your knowledge base

You can't layer AI assistance over outdated or thin knowledge base articles and expect it to be useful. An AI assistant armed with incorrect information can be dangerous: a customer would much rather receive no answer than an incorrect one that wastes their time.

In the run-up to incorporating AI into your customer support strategy, update your knowledge base and invest in building out your documentation. Even if you don't do any other prep before launching an AI assistant, make sure you do this.

Gather data on your support queries

What percentage of queries could be solved by customers reading an existing knowledge base article? What percentage should be solvable, if only the knowledge base article existed? And what percentage are less neatly solvable by your knowledge base, but instead require a human agent in the loop (i.e. customer success queries for highly specific use cases, bug troubleshooting, etc.)?

This information will help you determine where AI can be most valuable to your customer service team, and help you identify gaps in your knowledge base that need addressing.

Benchmark

Benchmark existing metrics now, so you can gauge the impact of deploying your AI solution. A few things you might want to benchmark:

• Average resolution time
• First response time
• Customer satisfaction score
• Customer sentiment

Repurpose or surface customer knowledge

Think deeply about repurposing other sources of customer knowledge — if there’s a goldmine hidden in sales collateral, onboarding materials, your SaaS community, or elsewhere. How can you make this available to large language models?

Design the AI-to-human hand-off process

If your AI assistant can't answer a customer request, how would you like the hand-off process to be when they're transferred to a human agent?

Customers should feel supported, not passed around. What info do you need (from your system, from the customer…) to handle the AI > human agent transition smoothly?

Using Outverse for your AI customer support assistant

For startups and scaleups, Outverse is a great way to scale support with AI self service. The Outverse AI assistant is trained on your support content, and requires no technical setup from your customer support team. With Outverse AI, you can:

• Deploy an AI assistant to directly respond to your customers and increase your self-serve capacity
• Help your customers find answers from across your entire Outverse space and other support materials
• Review individual customer interactions with your AI support agent
• Get smart insights on support trends

Book a demo with the Outverse team to find out how Outverse AI can help your startup scale customer support.