Each Chat account can support up to 20 unique bots, each backed by IBM Watson and/or ChatGPT. Bots can respond to students on any number of Live chat, SMS, Facebook messenger, Instagram, and WhatsApp channels.
This page will teach you how bots can be created, launched, and managed. Once you're live, learning about Gecko's bot reporting capabilities is helpful!
Creating and cloning bots
💡 Looking to create a new bot? Be sure to reach out to your account manager! Alternatively, contact the team at support@geckoengage.com
Once created, you can navigate between different bots by clicking Change Bot in the bottom-left corner of the page.
Answers
Think of Answers as the memory bank of the bot. Answers contain all of the questions the bot has been taught to look out for and the answers should any of those questions be asked. Learn more below:
Creating bot answers
Answers can be created/edited/deleted at any time by navigating to Bots > Answers. Click + Create Answer in the top-right corner to build a new answer.
There are three major sections to be mindful of when creating an answer:
1. Codifying
Answer Title: Internal only. We recommend using a Team Name - Answer Topic naming convention.
Answer Tags: Internal only. Use tags to describe what your answer is about to your teammates. Tags can be created in-line by typing your tag and pressing enter.
2. Question bank
Ask a question: Teach the bot ten unique questions/statements/phrases about your question. Aim to be as varied as possible with these questions. The bot will reference these questions when analyzing messages to determine the best answer to respond with.
3. Answers
Message - How should the bot respond to questions in the Question bank? Your reply can span multiple messages and supports rich text, emojis and template tags.
Skill - Learn more about skills below.
Once you're happy, click Save Answer in the page's top/bottom right corner. Finally, to teach your bot to start using your answer, click the Retrain bot card underneath your bot's portrait on the left of the screen.
Importing bot answers via CSV
Answers can easily be bulk imported into your bot from a CSV file. To import your CSV file, go to Bots > Answers and click Import Answers in the top-right corner. Upload your CSV, confirm changes and hit import!
As discussed in the video above, your CSV will need to be formatted as follows:
Column A - Answer title. The name of the answer you wish to import. No quotation marks are required. e.g.,
Admissions - How do I apply?
Column B - Questions. The questions your answer should look out for. We recommend a minimum of ten questions. Questions should be wrapped in quotation marks and comma separated. e.g.,
"How do I apply?","Am I able to apply to Gecko U now?","what are the steps for applying?","where do I go to apply","tell me about applying to ur school","i wanna apply but don't know how","apply now","information on applying","what undergrad programs do you have","help with applying"
Column C - Answer message. If you want your bot to reply with a message instead of a skill, use Column C and leave Column D blank. Answer messages must be wrapped in quotation marks. Each message block must be comma separated to send a series of messages. e.g.,
"That's a great question. The good news is that applying to Gecko U is easy!","To begin your journey, be sure to check out geckoengage.com and click 'Apply now'!"
Column D - Skill name. If you want your bot to reply with a skill, use Column D and leave Column C blank. Include the name of the existing bot skill you wish to trigger. No quotation marks are required. e.g.,
Apply to Gecko
Column E - Tags. Tags help codify your answers, making them easier to find and manage. Tags must be wrapped in quotation marks comma separated. e.g.,
"Admissions","Application","Apply now","Apply"
Finally, to teach your bot to start using your imported answers, click the Retrain bot card underneath your bot's portrait on the left of the screen.
Skills
💡 Skills can only trigger when used within an answer.
Your bot can make use of a range of different skills. Skills can be created/edited/removed by navigating to Bots > Skills or created in-line when building answers. Skills are an advanced way of responding to student questions within an answer.
Display list
The Display list skill displays a message above a pre-defined list of options to share with the student. You can display up to three unique list items at a time.
Head to Bots > Skills > +Create skill > Display List to build a new Display list. There are several fields to configure:
Skill Title - The internal name of this skill
Skill Tags - Internal tags to help find this skill
Message - What message should display above your list of options
List items - Define the label (what the list item text should say) and then pick from one of the following actions to occur if the list item is clicked/tapped by the student:
When you're happy with your configuration, click Save skill.
Suggest answer
The Suggest answer skill displays a message above multiple options. There is no limit to the number of options a suggest answer skill can provide:
Head to Bots > Skills > +Create skill > Suggest Answer to build a new skill. There are several fields to configure:
Skill Title - The internal name of this skill
Skill Tags - Internal tags to help find this skill
Message - What message should display above your list of options
List items - Define the label (what the list item text should say) and then pick from one of the following actions to occur if the list item is clicked/tapped by the student:
When you're happy with your configuration, click Save skill.
Data capture
The Data capture skill lets your bot capture any number of Chat custom fields.
To get started, go to Bots > Skills and click + Create Skill > Data Capture. Give your skill a name and pick the form you'd like to pull into the bot. From here, configure:
Confirm message - Let the student know they cannot ask any other questions of the bot when the Data capture skill is active. This text has a Yes/No option below it that students will pick from.
Exit message - What message should be displayed to the student once all custom fields are captured?
Configure fields - Define how your bot should phrase each question when capturing the custom field data.
Once you're happy, click Save skill.
Gecko form
Note: the Gecko form skill is only usable by clients with Gecko Forms. Interested in learning more? Contact your Sales or Account Manager here.
The Gecko form skill lets your bot capture responses for your Gecko forms. All workflows, conditional logic and integrations for your forms will work as expected!
Before starting, ensure your Chat x Engage accounts are integrated by logging into Chat and going to Settings > Integrations.
Head to Bots > Skills > +Create skill > Suggest Answer to build a Gecko form skill. Name your skill and pick the form you'd like to pull into the bot. From here, configure:
Confirm message - Let the student know they cannot ask any other questions of the bot when the Gecko form skill is active. This text has a Yes/No option below it that students will pick from.
Exit message - What message should be displayed to the student once all form fields are captured?
Configure fields - Define how your bot should phrase each question when capturing data in each field.
Please note that the bot cannot capture these Gecko form fields: Consent, File, Form Section, Matrix, Media, Organisation, Repeatable & Text block fields.
Workflow action
Head to Bots > Skills > +Create skill before clicking the Workflow action card.
Be sure to give your skill a memorable name in the Skill title field to enable you to identify it later on.
From here, pick any number of the five types of workflow that you'd like to trigger:
Assign conversation to agent/team: assigns the conversation to the specified teams and/or agents.
Trigger a message to the contact: this allows you to send a pre-configured message to a contact.
Add a note to the contact: adds an internal note to the student's contact record, which only other Chat agents can view.
Assign a tag to the conversation: add a pre-created tag to the contact.
End Conversation with Bot: when this workflow action triggers, permanently unassign the bot from the conversation.
When you're happy with the automation you've set up, click Save skill.
Bot settings
To access your bot settings, navigate to Bots and click the cog icon at the top-right of your bot's avatar.
Here you can rename your bot and pick its avatar.
In the General Settings section, you can pick which channels the bot should actively respond to new conversations. If you'd prefer, bot assignment rules can be set via workflows that allow for extra control using conditional logic. Bots can respond to students on any number of Live chat, SMS, Facebook messenger, Instagram, and WhatsApp channels.
At the bottom of the Settings page, you'll find a toggle to Show Advanced Settings.
Bot confidence
These options allow you to configure how your bot performs precisely. The options available to you and how they function change depending on the type of bot you are using:
IBM Watson bot
IBM Watson bot
Match breakpoint: The high confidence breakpoint required for an answer to be considered a match. Answers with a confidence % >= the Match breakpoint will be sent to the student without creating a training opportunity. Recommended value: 75%.
Low confidence breakpoint: The minimum confidence needed to reply to the student. Answers with a confidence >= the Low confidence breakpoint and < the Match breakpoint will reply to the student can create a training opportunity, if desired. Recommended value: 40%
Close conversations after period of inactivity: Pick how many hours after the student's last message Gecko should wait before closing the conversation. Choose from 8 hours, 24 hours, 48 hours, or 72 hours. Just so you know, most bot-reporting metrics require conversations to be closed before including them in the results.
Hybrid bot
Hybrid bot
Match breakpoint: The high confidence breakpoint required for an answer to be considered a match for existing IBM answers. IBM answers with a confidence % >= the Match breakpoint will be sent to the student immediately. Answers with a confidence % <= the Match breakpoint will invoke ChatGPT. Recommended value: 85%.
Close conversations after period of inactivity: Pick how many hours after the student's last message Gecko should wait before closing the conversation. Choose from 8 hours, 24 hours, 48 hours, or 72 hours. Just so you know, most bot-reporting metrics require conversations to be closed before including them in the results.
ChatGPT bot
ChatGPT bot
Close conversations after period of inactivity: Pick how many hours after the student's last message Gecko should wait before closing the conversation. Choose from 8 hours, 24 hours, 48 hours, or 72 hours. Just so you know, most bot-reporting metrics require conversations to be closed before including them in the results.
Bot training opportunity expiry
Gecko will tidy up your train table for you to keep your training opportunities as manageable as possible.
If you'd like to change any of the expiry dates, be sure to contact your account manager or the support team!
Low Confidence - This applies to IBM Watson-only bots. Low confidence matches are created when the bot replies to the student, but the answer's confidence % is >= the Low confidence breakpoint and < the Match breakpoint. By default, these opportunities will not be logged on the train table, but this behavior can be changed by increasing the value in this field to >= 1. Recommended value: 0 days
Very Low Confidence - Applies to all bots (IBM Watson, Hybrid, ChatGPT). Very low-confidence matches are created where the bot is not confident enough to return an answer to the student. Recommended value: 7 days
User complaint - Applies to all bots (IBM Watson, Hybrid, ChatGPT). User complaints are created when a user clicks '👎 Not Useful' or uses a reserved complaint message. Recommended value: 30 days
Finally, any unwanted bots can be deleted by clicking the Delete option at the bottom of the page.
Training
💡 Updated Feb 2024: To keep your training opportunity reviews as manageable as possible, we automatically clear opportunities on a regular schedule. These expiry times can be controlled and edited by navigating to Bots > Settings > Advanced Settings. Learn more here.
Training allows you to make your bot smarter and better than ever before. Completing an opportunity will add the student's message to the answer's question bank.
Navigate to Bots > Train to view the training table. Training opportunities can be managed by clicking on any message on the train table and resolving it in any one of five unique ways:
Confirming - If your bot answered correctly, click Resolve Answer to complete the opportunity.
Ignoring - Opportunities can be removed from the train table by clicking Ignore Miss. Helpful for questions you don't want to teach the bot.
Redirecting - Is there a better answer you'd like your bot to respond with? Search your existing answers in the Choose an Answer field. You can search by scrolling or by typing. Gecko searches your bot's answers across the answer title, answer tags, answer message, and skill name fields. Click the answer that best fits, and you'll see a preview of how the conversation could look in the future in the Example Conversation column. Click Resolve Answer to complete the opportunity.
Editing - is an excellent option if you have an existing answer that almost answers the student's question. Ensure the answer you wish to edit is selected in the Choose an Answer field, and click Edit in the top-right corner of the training pop-up window. This will take you to your answer and allow you to edit it.
Creating a new answer - Unable to find a suitable reply to the student's question, even after you've searched in the Choose an Answer field? No problem: click create an answer at the bottom of the training pop-up window to build a brand new answer. Learn how to create answers.
Training opportunities are a great way to level your bot up, but they are optional. Don't feel pressured into doing them! If you're finding too many opportunities are coming into your table, either ignore them (and Gecko will auto-clear them after a set amount of time!) or speak to your Account Manager about making changes to your bot settings.
Any questions? Start a live chat with a support team member, or feel free to explore the rest of our academy. Spotted an error or want to suggest a future article for the academy? Let us know here.