Introducing Apla, the AI Account Planner Agent.
Introducing Apla, the AI Account Planner Agent.
Join the Webinar
This agent will send out a welcome message to email addresses when prompted to. Then it will handle any follow-up questions that are sent via email. If it doesn't know the answer, it will escalate to a human to answer via Slack. Then once an answer is provided it will update the Frequently Asked Questions knowledge base, and reply to the original question email with a response.
This agent is great for handling support questions over email. The core building blocks are: Sending emails, checking knowledge for answers to questions, escalating unknown questions to a human or team on Slack, and updating knowledge with new information. All of these can be repurposed for a wide range of use-cases.
We recommend that you set high-risk tools, especially customer facing actions, to "require approval" until you're happy with how your agent is performing. Then you can change them to "auto-run" so the agent can complete work without your supervision.
❓
This tool will update a Frequently Asked Questions knowledge table with a new question and answer pair. You can adapt this tool to add any information to an existing knowledge table.
Build or use this tool ->
🔎
Given a question, this tool will search an Frequently Asked Questions knowledge table to see if there is an answer to that question. If there is, it will return the answer. If there isn't, it will return "I don't know".
Build or use this tool ->
📧
This tool sends an email with Outlook. You can follow the same process for sending an email with Gmail instead, just swap out the Outlook tool-step with a Gmail tool-step. You can guide the kinds of emails your agent composes via the core instructions in your agent settings.
Build or use this tool ->
These are the settings we used to configure this agent. Every setting is completely customisable. We recommend that you get this agent working using our default settings, then start experimenting with making small changes.
Create & Configure an agent ->
Email FAQ Follow-Up Agent
This FAQ follow-up agent sends out a welcome message, and handles follow-up questions asked by recipients. It checks to see if there is already an answer to the question in the frequently asked questions table. If not, it’ll escalate to a slack channel and ask for an answer there. Then it will update the FAQ table with the new Q&A pair, and respond via email with the answer.
Your Microsoft outlook account must be connected under "Integrations/Connections" for the send email tool to work, or Gmail if you're doing it that way instead. In the "Abilities" section of agents settings, you also need to connect to your Slack account under the "Escalate to humans" option.
GPT-4o
When you receive an email, do the following:
1. Extract the questions from the email.
2. Retrieve the answers to those questions from knowledge using Check knowledge for answer to question .
2a. If an answer to that question doesn't exist, Escalate to manager via Slack and ask for an answer.
2b. Once you get an answer back, Add new FAQ to knowledge .
2c. If an answer to that question does exist, carry on with the flow.
3. Compose an email which replies to the questions (or replies in general if there were no questions).
4. Send email (Outlook) back to the sender of the original email.
Double-check that you have followed all of the instructions. Did you reply to the original email via an email?
--
If you are asked to send a welcome email, this is the email you should send:
Welcome to Relevance AI!
We're excited to see the impact your AI workforce has! Feel free to ask us anything, just reply to this email (e.g. request a demo video).
Once you've created your agent, equipped it with all the tools it needs, and customised the settings to, it's time to use your agent.
In this case, we have added an Outlook trigger, which means your agent will receive every email that gets sent to the connected Outlook account. Try this out by sending an email to that account with a question as part of the body, you should see a new task pop-up in the agent task history.
After that, if you have set your tools to approval mode, you will see "approve/reject" task requests where your agent asks for your permission to send a response to the question over email, or escalating to slack.
The video at the top of this page shows you how the agent handles FAQ questions.
