I'm facing issues with training the bot on Slack, Discord, URLs, or PDFs. What could be the problem?
For URLs - You can train it on documentation or support URLs, GitHub Repos, Substack, Medium, and most SaaS/product/publicly accessible websites. You might face issues with Notion links, private/gated links, and some other websites. We will be adding support to more URL types soon.
If the issue persists after deleting and re-adding the URL, please drop us a message and we will reach out to you. You can also reach out to us at support@threado.com
For Slack and Discord - The bot can only train on public channels unless you give access to private channels i.e. add the Threado bot on private channels (see how).
For PDFs - While training, the formatting of the PDF could be an issue.
Why is the bot not being able to find the right answer?
The reason might be that the bot is not trained on the right content. Revisit your question and make sure the bot has been trained on what is being asked. After you've retrained the bot, test out relevant questions from the Test tab before installing or embedding it.
If the bot continues to not answer the question even after being trained on the right content, this might be an internal error. Please reach out to us at support@threado.com, we will look into the problem asap and get back to you.
How can I prevent the bot from responding to every question on Slack or Discord?
The bot auto-responds to queries on channels where it has been installed. We would recommend having a dedicated channel just for asking questions and communicating the same to your community members as well. This way, you can avoid the bot answering questions on every channel.
Why is the URL training taking time?
The time taken for the bot to crawl and train on URLs depends on the size or depth of the website/URL. Larger websites that have thousands of URLs are likely to take a bit of time for the bot to completely train on. You will be notified via email once the training is complete.
Why is the URL training not initiated?
The structures of some websites are different from the usual ones, because of which it takes time to crawl these sites and pull all the data. In case it doesn't initiate in a couple of minutes please feel free to reach out to us at support@threado.com.
Why is the URL training failing?
The most likely reason could be that the URL is gated or has blocked crawling on that page. Make sure that the link is public otherwise, the bot will not be able to train on it.
Other reasons could be that the page (although indexed), does not exist anymore, is a broken link, or throw a 404 error. You can manually check the link to make sure it's a valid link. If this isn't the issue, try re-training the bot on the URL and reach out to us if the problem persists.
How can I train my workspace/server members about the Threado AI bot?
You can share this article with them that shows how end users can resolve their queries using Threado AI.
Have a dedicated channel for queries and tell your workspace members to post all their queries in that channel only.
Feel free to ask them to check out Threado AI themself.
How do I track unanswered conversations?
Navigate to the Open Conversations tab which filters conversations that are open i.e. questions that haven't been answered yet. You can now respond to these questions manually and ensure the best possible product support for all users.
I'm not able to save my customizations. What should I do?
This is an internal issue. Please reach out to us at support@threado.com and we'll figure out what the problem is.
How can I train the bot on video?
You can't directly train the bot on video but you can convert those videos into transcript PDFs and then upload those PDFs for the bot to be trained on.
Can I train on gated URLs or URLs that require login?
No, it's not possible to train the bot on private content, gated URLs, or resources that lie behind login walls.
Can I ask questions privately to the bot on Slack or Discord?
Yes, you can enable private messages for the bot when installing it on Slack. This allows the users to ask questions directly to the bot without having to publicly post them in channels.
You can also ask the bot questions by tagging it in channels where it's been installed.
As of now, this option isn't available for Discord.
Can I have multiple chats with the chatbot?
Yes, you can create separate threads to chat over different contexts with the bot. This comes in handy when you have multiple questions or if you come up with more queries before the previous query gets resolved.
How are the bot responses counted and billed?
Each query or question that a bot responds to is counted as a response.
In the free plan, the bot will answer up to 50 queries and you're allowed to train the bot on 1 community channel, 1 PDF, and 1 URL.
On the Basic plan, the bot will answer up to 500 queries and you're allowed to train the bot on 10 community channels, 10 PDFs, and 10 URLs.
If you want a custom limit for the bot, please reach out to us at support@threado.com
During your free trial, you will be on the Basic plan and will automatically be moved to the Free plan when the trial ends.
What happens when the limit is exceeded?
Depending on whichever plan you are on, the bot is limited by the number of responses it can give and the number of data sources it can be trained on.
When the response limit of the bot is reached, it will stop responding to any new queries on Slack, Discord, or the Web.
When the training limit of the bot is reached, it can no longer be trained on new data sources.
For more info on limits, checkout out pricing plans.
Where to check the bot response usage count?
You can check the usage count on the Billing page under Settings. An indicator will show you how many queries have been answered and how many more are left depending on your current plan.
📌 Important Note: For additional concerns or any issues pertaining to the bot, feel free to reach out to us at support@threado.com. We will be happy to assist you.