Introduce Fin AI Copilot to your team
Fin AI Copilot uses your trusted knowledge sources to provide support with:
Generating answers to share with customers,
Offering guidance based on internal resources and previous conversations,
Making it easy to browse and validate sources without leaving the conversation,
Streamlining the process of responding to customers using your preferred tone, phrasing, and language.
This article explains how you can test and roll out Fin AI Copilot to your team.
Changing habits: Fin-first approach
Using Fin AI Copilot should become the first port of call for all teammates looking for help with a conversation or ticket. Fin makes this an intuitive and easy experience as you don’t need to switch tabs or remember different places to search for information, you only have one place to look and it’s right there in the inbox.
If Fin doesn’t give you the exact answer you were searching for, it will provide the sources of the answer which can be viewed from the AI Copilot panel in the inbox for further investigation, without needing to change tabs and lose the context of the conversation.
Changing habits is hard - a good way to get more teammates to adopt this new way of working is by encouraging teammates to share their Fin experiences with each other.
Here’s how we did it at Intercom:
We created a tiger team of trusted and tenured agents who were involved in the test. We did a quick kick off meeting to walk them through the tool and created a dedicated Slack channel for them.
Over the course of a few days, we prompted them to share their Fin AI Copilot experiences here:
Success stories of when Fin worked well
Examples of when it didn’t
Content gaps they identified
Suggestions for new content
Our team leads also sought more anecdotal feedback in daily stand ups and 1:1 sessions with the tiger team.
Enable conversation history
Your past conversations and customer tickets provide a rich history of knowledge teammates have shared with customers, and this knowledge can be made available to Fin AI Copilot for all teammates to have quick access to answers and solutions other teammates have provided.
You can also control which teammates' conversations and tickets Fin AI Copilot has access to and ensure only answers from tenured support agents are available.
Note: Fin Ai Copilot will take content only from chat conversations and tickets. It will not pull content from e-mail conversations.
Add internal content
One of the key benefits of Fin AI Copilot is that it can help guide teammates with your team’s best practices and internal processes. To get the most out of Fin, you should add internal sources such as teammate onboarding articles, conversation handling advice, and troubleshooting tools.
Once that’s done, don’t forget to let your team know that Fin has been enabled with this knowledge and can help them if they’re stuck!
Ask good questions
Use these guidelines when interacting with Fin AI Copilot to improve the quality and relevance of the responses:
Ask questions that relate to your support content: Fin uses your knowledge sources to answer questions. Make sure you ask questions which might be answered within this pool of content.
Ask clear and detailed questions: When posing your first question, ensure you provide enough detail. If there are features or terms that are similar or have similar names, be specific to avoid confusion.
Provide context: Context helps Fin understand the background of your question, leading to more accurate and relevant responses.
Use follow-up questions: Since Fin AI Copilot is conversational, you can ask follow-up questions to refine the information or to delve deeper into a topic.
Understand the difference from traditional search: Unlike traditional search engines, where you use keywords and sift through the results yourself, Fin AI Copilot processes your query and sifts through relevant sources to generate a coherent answer.
Be specific to avoid vague responses: If your query is too vague or relies on only a few keywords, Fin has to pull from a broader range of resources, which may result in an off-topic response. Being specific helps narrow down the information pool, leading to more accurate answers.
Understand myths and anti-patterns
Asking questions which can’t be answered using available knowledge: Fin isn’t an AI assistant that can answer any question you might have. It can only answer questions based on content you’ve enabled in the Knowledge Hub. For example, it can’t answer specific questions about the customer or the company of the customer.
Giving instructions: Fin AI Copilot doesn’t work like ChatGPT where you can give it instructions or prompts e.g. “make this shorter”. However, you can modify Fin’s answer by clicking on the menu of the “Add to composer” button, such as making the answer more friendly or formal.
Asking what the conversation is about: Fin AI Copilot can’t summarize the customer conversation or tell you what it is about. If you want to summarize the conversation, you should use the AI Summarize feature instead.
Expecting that Fin AI Copilot needs ‘training’: Fin doesn’t automatically improve over time or with more usage. All it needs to perform well is good content and well formed questions and as such doesn't require any actual training.
Utilize the source filter
In cases where you know you’re more likely to find a good answer in a particular source, it’s worth using the source filter to hone in on the information you need, and therefore get a more relevant answer. For example, for questions on how to handle certain conversations, choosing "Internal articles" may be more useful.
Verify sources for accuracy
If the inline source for an answer is a past conversation, check the date range of the conversation, if it's a recent conversation (within the last few weeks), that’s a good indication it's still accurate. If it's an older conversation, check out another source (there will always be a few sources shared in the AI Copilot tab) to be sure.
If you're unsure if Fin’s solution is correct, it's always good practice to test it in your own product before sending the response to a customer.
Provide feedback for continuous improvement
Any time Fin AI Copilot is not able to provide an answer, have your team tag the conversation with “Fin Missing Knowledge.” You can then have teammates review conversations with this tag and establish whether there might be a knowledge gap that needs a new or updated resource.
Utilize Back-office tickets to enable teammates to submit requests for updates to articles straight from the Help Desk. If Fin AI Copilot gave an answer that didn't help, they can check the source(s) used and open a ticket to get the content improved without leaving the inbox or losing too much time.
Need more help? Get support from our Community Forum
Find answers and get help from Intercom Support and Community Experts