👋 Looking to set up a customized Text AI Agent in your Structurely account?
Head over to our AI Agents settings page to do just that 👇
In this tutorial, we’ll walk through how to:
Create a Text AI Agent
Upload & add a Knowledge Base
Establish a drip cadence
Enroll your leads to communicate with the Text AI
⚠️ First, please be sure that you have requested access to the “Text AI Agents” feature from the Structurely team!
This option is currently in beta, so you may not have full access upon the initial setup of your account. If you find that your newly created Text Agent has seemingly disappeared, please let us know as this feature is likely not yet activated for your account.
👉 Follow this link to the Text AI Agents page within the AI Agents section of your Structurely Homechat account.
Select the “Create Agent” button at the upper right, and begin filling in each section as it is presented:
👉 The name of your agent is what will be utilized as the tag for lead enrollment later on, so we suggest something short & sweet that is likely tied to the specific use case.
Please note, this is NOT the name that your Agent will use to introduce themselves; that setting is found here, under “AI Settings”
You’ll want to use the Standard Text Prompt option under the System Prompt drop-down, and you can add a Lead Type if there is a fitting match available.
🛑 Be sure to save your progress using the “Create Agent” and later the “Update Agent” button as you make changes. If you do not save the changes made, your Text AI Agent will revert back to the last saved version. 🛑
👉 Persona defines how your Text AI Agent sees themself - how they communicate, their job title, and what type of role they are filling. You will have preset values within these fields, but you can manipulate, add, remove, or replace the values as you see fit.
Everything established here can be changed later on so you can do a bit of trial and error, too!
👉 The Greeting is the general template for the introduction that a lead will receive when we first make contact; it will automatically bring in the Conversational Intent that was set in the first step, but you can manipulate the layout of this content if you like.
We do recommend using the {name} field to personalize the Greeting, and it is required by Twilio to identify “yourself” so we also require using the {bot_name} as well as the {agency_name} though you are welcome to add those names in directly rather than using the formatted merge fields if you prefer.
👉 Knowledge Bases can be shared between your Text AI and Call AI Agents, and as such are often the simplest/quickest way to add your company’s FAQ’s and/or Q+A’s.
To create a Knowledge Base, you will actually need to maneuver to the “Knowledge Bases” page. Just be sure to update your Agent before you switch pages! ✅
You’ll want to name and create your Knowledge Base:
Then you can manually add Questions + Answers within the “Custom Answers” section or upload a DOC, DOCX, or PDF file using the button at the upper right 👇
Once your Knowledge Base has been created (and saved/updated, if relevant) you’ll want to maneuver back into your Text AI Agents tab and choose to “View / Edit” the Agent you started previously.
Now that you have an existing Knowledge Base saved in your account, you should have the option to Select a Knowledge Base (and can select the appropriate option from the dropdown) 👇
👉 Guidelines refer to the Qualifying Questions and predefined Answers that make up the underlying content of your Text Agent’s conversations.
The questions defined here will generally be asked in order, but may be manipulated by the Text AI Agent to better fit the flow of the conversation - our goal is to seem as human-like as possible, after all!
If you have already defined specific Q+A info within your Knowledge Base there is no need to repeat it here; it is fine to include additional Q+A’s, though, as long as the information does not conflict with what is present in the Knowledge Base.
👉 The Drip Cadence options allow you to decide how often your Text AI Agent will reach back out to a lead once they stop responding.
These are reset after each interaction from a lead; e.g. will be rescheduled on the specified timeframes each time we receive a new response.
When you go to “Add Drip” you will have 2 options for the scheduled cadence - either a Schedule Based Delay or an Offset Based Delay 👇
These should be fairly self-explanatory - just keep in mind that your offset based delays are ALL scheduled based on the single last response (with your AI’s hours of operation, naturally) so you will want to space them out accordingly.
You can also choose to toggle ON / OFF the “nudge lead” option which would normally go out 6-12 minutes after the first message if the lead does not respond.
We’ve found this to greatly increase our engagement rates (by up to 40% in some instances) and do highly recommend leaving this turned ON 👇
Last but certainly not least, you have the option of suppressing any further text communications if certain criteria are met; we do have a list of recommendations preset which you can adjust or add to if need be.
🎓 Our team is more familiar with the specific criteria determinations on the back end, and would be more than happy to assist you if there is a certain circumstance that you are trying to incorporate here! 🎓
👉 Options for Qualification allow you to add Disqualification criteria.
If one of these criteria were met, the Text AI Agent would smoothly end the conversation (for example, if you are a real estate agent that does not work with renters, you could specify the criteria as follows) 👇
📲 You can also choose how the Text AI will handle the outcome of its conversation here; standard protocol is to live transfer to a human, but you have the option of choosing any existing Call AI Agent to give the lead a call at that point, instead!
✏️ When you enroll your leads into communications with this new Text AI Agent you will want to use the Agent’s name (as specified in the General portion of the setup) as the enrollment tag within your CRM.
We recommend viewing the CRM-specific processes also found in the Structurely Help Center if you are unfamiliar with the tag-based enrollment.
If you have any questions you can send an email to help@structurely.com and we will get back to you!
🚀 The Structurely Team