Bots
Getting started
Learn how to turn your knowledge base into a ChatGPT-like bot within your favorite messaging or community platform. In minutes, with no code.
IngestAI enables you to quickly build and deploy ChatGPT-like bots that answer on queries based on uploaded knowledge-based (technical documentation, educational program content, company knowledge base, etc). Upload your knowledge base, open our WebChat, and get accurate and relevant answers about your product.
We encourage our new users to follow our simple Implementation Guide.
Implementation guide
1. Start small. Before sharing your bot with colleagues, try it out to ensure it works as intended. Starting small helps you identify potential issues and make necessary adjustments before showing them to others.
2. Start with simple sources of data or information and gradually add more. For educational materials, we recommend adding a Dictionary first, followed by Tutorial Guidance and Video Transcripts. This step-by-step approach ensures your bot answers queries accurately and prepares it for more complex data.
3. Make it Easy. Make the bot creation process easy by testing your library in our WebChat or creating a Telegram bot – both of which can be done in just 2-3 minutes or even seconds in the case of our WebChat. As you become more confident in using IngestAI, creating additional chatbots for other platforms becomes much easier. You can add Slack, Discord, WhatsApp, and Microsoft Teams, embed our WebChat into your website, or access it through our API Endpoints.
4. Analysis and Action. After creating your bot, share it with a small group of colleagues to obtain feedback and valuable insights. Use this information to improve your bot's accuracy and effectiveness (e.g. playing with the creativity level and detalization). Analysis of how your colleagues are using it and their feedback can help you create better bots that will provide meaningful assistance to your team and customers.
Before uploading your docs
Here, we will guide you through the process of uploading files to IngestAI Libraries. We support a wide range of file formats, including:
.txt,.csv,.doc,.docx,.abw,.djvu,.docm,.dot,.dotx,.html,.hwp,.lwp,.md,.odt,.pages,.pdf,.rst,.rtf,.tex,.wpd,.wps,.zabw,.azw,.azw3,.azw4,.cbc,.cbr,.cbz,.chm,.epub,.fb2,.htm,.htmlz,.lit,.lrf,.mobi,.pdb,.pml,.prc,.rb,.snb,.tcr,.txtz,.dps,.key,.odp,.pot,.potx,.pps,.ppsx,.ppt,.pptm,.pptx,.et,.numbers,.ods,.xlr,.xls,.xlsm,.xlsx
Before uploading your files, it is essential to follow the GIGA rule, which is well-known among data scientists. GIGA stands for "garbage in, garbage out," meaning the quality of the output is severely impacted by the quality of the input data. Therefore, it is crucial to remove irrelevant, outdated, duplicate or contradictory data to improve the output quality.
Our core technology focuses on quick retrieval of relevant information from your knowledge base, and then converting it into a conversational chat that mimics human interactions, using AI – within your favorite app or just embedded into your website.
By considering the GIGA rule and using our technology, you can ensure accurate and reliable data outputs. If you have any questions or need further assistance, please don't hesitate to contact our support team.
If you’d like to know more about data preparation, please read our blog Data Wrangling for AI Virtual Assistants and AI Chatbots. https://ingestai.io/blog/prepare-data-for-ai
Important! Basic rules for creating a file for libraries. Follow the rules below:
- Ensure that all content relevant to a specific topic is stored in the same Library. If splitting data to make it accessible from different chats or slash commands is desired, create separate Libraries and upload the content accordingly.
- Higher granularity results in more predictable (and less creative) responses, since it's harder for the AI to give different answers based on small, precise pieces of text.
- Data accuracy: Make sure these answers are comprehensive and detailed and do not consist of short, one- or two-word answers such as "Yes" or "No."
For example:
"Can I cancel my subscription during the trial period?" -> "Yes, you can cancel your subscription during the trial period" makes the information clearer
- Avoid Binary Responses:
For example:Columns with binary data showing whether a customer has children or higher education (represented by "0" or "1") should be changed to "have children" or "has higher education" as “1” and “doesn’t have children” or “doesn’t have higher education” instead of “0”
"number of children: 3", "work experience: 5 years” this is the correct data for AI
- If you want to download data from videos or images, first reformat them into text format using subtitles or additional applications. Don't try to embed images or videos in your uploads
- Excel or Google Sheets files should only have one tab, extra tabs AI bot will not see
In complex scenarios, such as integrating tabular structured data (Excel, Google Sheets, or relational databases like SQL) with text content (.docx, .txt, etc.) in the same Library, manual configuration by IngestAI may be required to achieve the best results. This customization service is currently available only in Business or Enterprise tariff subscription plans.
- Use a unique paragraph for each topic of up to 2000 characters
If you have paragraphs or rows in Excel or Google Sheets exceeding 2000 characters, we recommend using summarization or other prompt methods (available in IngestAI Prompt Engineering functionality) to reduce the maximum paragraph size to no more than 2000 characters.
Сompliance with all of the above rules will help to fully use modern AI technologies
How to upload your docs
Follow these instructions to create a library.
- Got to "Libraries" from the navigation bar.
- Click "Add Library" button.
- Name your Library and click Save.
- Drag and drop your docs or use our web crawler.
- Click Save and wait until the upload is complete and your docs are processed, this can take a few seconds or minutes, depending on the file size.
Advanced options
Creativity. A higher temperature (will result in more diverse and creative responses, sometimes sacrificing coherence. A lower temperature will lead to more focused and consistent responses but may appear repetitive and less creative. Adjusting the temperature allows you to fine-tune the balance between predictability and creativity in the model's output.
Library Detalization
Higher Library detalization allows for consistent context throughout the document even after it's been split into smaller sections. This is important because it ensures that the information provided in each section or document is understandable and seamless, allowing for effective communication of ideas. We recommend increasing Library Detalization in case of contextually rich documents, like learning materials and decrease it to a minimum in less contextually connected documents, like in case of spreadsheets where information is normally spitted by rows.
File Size: Freemium users are able to upload one file with limited size of 3Mb.
Summary
In summary, Libraries are organized compilations of data that make your chatbots context-aware. Libraries enable your chatbots to access specific data points and then generate responses based on user input and the retrieved data from your knowledge base. When initiating a conversation with a chatbot, you can access only one Library. However, the number of files you can upload to one Library depends on your monthly membership or subscription plan.
Web Chats
This function is used to work with libraries that are loaded with the correct files, which the AI-Bot can respond to and answer questions contained in the downloaded files
After creating the library https://app.ingestai.io/main/libraries, you have the option to send AI requests in the Web Chats tab https://app.ingestai.io/main/webchats .
To open a web chat, you need to go to the page in your personal account to the WebChat page, and go to the chat in the library you are interested in using the blue button. A detailed description can be found on our website at the link https://ingestai.io/docs/webchat-embedding
Links/URL Crawler
Both of these features are designed to make it easier to work with content that is located on Internet resources, sites, web pages, or online documents. This feature is only available to authorized users.
Links Crawler
Links Crawler - where users can provide URLs of Google Docs and Google Sheets (each link from a new line, 10 links max). Also, straight website links are supported in a current library (https://ingestai.io/docs/howto). The Links Crawler will analyze all content on the page and after that in WebChat (https://ingestai.io/docs/webchat-embedding) or in some of these bots (https://ingestai.io/docs/discord) there is an opportunity to ask questions about this content. This feature is available for all account types.
Web Crawler
Web Crawler - where users can provide URLs of Google Docs and Google Sheets. Also, straight website links are supported in a current library (https://ingestai.io/docs/howto). The crawler will go to your site and crawl all the internal pages it finds on each page. The crawl limit is 100 links. The Web Crawler will analyze all content on the page and after that in WebChat (https://ingestai.io/docs/webchat-embedding) or in some of these bots (https://ingestai.io/docs/discord) there is an opportunity to ask questions about this content. This feature is available only for PRO and BUSINESS account types.
After successfully creating the links in the library we can ask any questions about the content contained in this resource.
A detailed description of Links Crawler and Web Crawler can be found on our website at the link https://ingestai.io/docs/web-links-and-website-crawler
Bot integrations
IngestAI is a platform that lets any one without technical background and minimal technical skills build and launch contextually aware chatbots on numerous messaging platforms. It offers seamless integration with other tools. Explore how to integrate IngestAI with a platform of your choice - Slack, MS Teams, Discord, WhatsApp, Telegram, or embedding our web chat into your website.
Discord
Experience the convenience of integrating IngestAI with Discord, a leading collaboration platform for gaming and communities. Connect your chatbot directly to Discord to access a variety of conversational AI features and elevate your chatting experience to the next level. With IngestAI and Discord integration, conversing with your chatbot has never been easier or more engaging.
To integrate IngestAI with Discord, follow these simple steps: https://ingestai.io/docs/discord
Slack
Elevate your team's productivity and streamline communication by integrating IngestAI with Slack. Deploy your chatbot directly on Slack in minutes for seamless communication and efficient collaboration. With IngestAI's powerful integration, you can enjoy real-time messaging and access to numerous tools and platforms to enhance your chatbot experience. Interact with your chatbot effortlessly within the Slack platform and take your team's workflow to the next.
To integrate IngestAI with Slack , follow these simple steps: https://ingestai.io/docs/slack
WhatsApp
IngestAI's integration with WhatsApp offers a simple yet powerful solution for businesses to interact with their customers through chatbots. With a user-friendly deployment process, businesses can conveniently engage with their customers on this widely-used messaging platform. IngestAI's integration guarantees seamless communication with customers, ensuring efficient and consistent interactions with chatbots. Simplify your customer service by integrating WhatsApp with IngestAI to provide a smooth, reliable, and effective user experience.
To integrate IngestAI with WhatsApp, follow these simple steps: https://ingestai.io/docs/whatsapp
If Bot doesn't work
If the bot in the messenger does not work, first check if all the following points are met:
- If you are using Discord, make sure the bot is online.
- If the bot is offline, it's likely that your token has expired or that you accidentally refreshed it without updating your ingestai account settings. Make sure that the token in the settings of your bot in the personal account of ingestai and the current Bot is the same
- Make sure the files loaded into the current library are working and correct. To check this, you need to enter the current library and click on the magnifying glass icon next to the file you are interested in "show raw content". A pop-up window should open with the content contained in your file.
If the above points did not solve your problem, please contact our team.
Сan you understand youtube videos?
IngestAI currently doesn’t support uploading content from YouTube videos. Nevertheless: here's a brief tutorial on how to copy transcript from YouTube videos. It's literally one minute and you have your transcripts: https://t.co/rEIgdhh7nK
Prompts
Prompt: it’s functional which helps to create a behavior model for the bot to follow. In Prompt, you can create macros for further queries, Prompt can store query history, and create macros both with data from open sources and with local libraries
Save Your Prompts: You can now effortlessly save your prompts in our web interface or any messaging apps like Slack or Discord, making them accessible wherever you need!
Prompt Descriptions: Add a description to your prompts to easily remember the purpose it was designed for and stay organized.
Tagging System: Stay on top of your projects with our efficient tagging system - find the right prompt or group prompts by tags in just a few seconds!
Edit & Roll-back: With prompt versioning, you can edit your prompts and roll back to any previous versions, giving you full control over your content.
Save Requests & Answers: Keep track of all your requests and answers for each prompt and its versions - never lose valuable information again!
Grading System: Grade the answers your prompt generates and see the average grade of all generated answers to understand its effectiveness.
How to create a new Prompt?
- Need to be authorized. After that, go to the "Promts" tab. On this page will be all created Promts
- Click on the “+Add Prompt” green button for creating a new Prompt
- On the current page - the ability to create new macros using the option
In the Prompt Name field, add a new name
- In the Model selector, you need to select AI Generator. After this prompt, all requests will work with this AI generator.
- In the library selector, select a library from the previously created ones, and the AI will only work with content from this library, or leave "Direct prompt to AI" and the bot will use the exposed resources.
- Prompt Description textarea: This is not a required field, it can be Empty but can be used as an indication
- Prompt Header (first part of the prompt) - This is not a required field, but it is an important option if you want to get a custom response format. For example: "Crete a text using the following background". Leave it empty if you want to pass user's request to AI model without any changes. If you're using your own Library, context will be loaded in this section by default, but you also can customize it by using [[context]] variable in the text.
- Prompt Footer (part of the prompt that comes after user’s input or context from your Dataset/Library). This is not a required field, this function can also help in editing the implemented query. You can set the number of lines that will be in the answer, and language, you can also add links, or content for comparison
- Prompt Tags field: This is not a required field. This feature can help you mark a special Prompt. To add a new tag, enter a word in the "Tag" field and press "Enter".
- Click on the “Save” button
Make the Project available to other users
After the Prompt was created it can be changed to the public, In the edit state of the Prompt, a public Public Prompt toggler appears at the bottom of the page, after tern on it generates a URL that can be shared with anyone.
Existing features and settings for Prompt
Go back to the Prompts page. New options will appear on the left side: Run Prompt, Prompt Requests History, Edit Prompt and Delete Prompt.
Run Prompt: After clicking on this button, you will be redirected to the request page where you can enter a request to the AI using the previously created settings.
Prompt Requests History: In the Prompt request history option, you can see the history of all requests in this Prompt, mark the most liked answer from 1 to 5.
Edit Prompt: After clicking on the pencil button, the Prompt template can be edited.
Prompt can be deleted by clicking the delete button on the right side of the screen.
There is a prompt example?
"Prompt" function, there are several examples of how it can be created.
On the Prompt page second green button “Pre-Built Prompts” contains examples of how to compose Prompt
What is Playground in the Dashboard?
This functionality was created to get acquainted with our system. Each user can try to ask the AI a question and get an answer using the models of the open-source version of the AI or from their uploaded files via libraries. Where he can choose the path in which the response from the AI will be formed, it can be either a text format or an image using outdated AI bot models or the most modern ones.
How does the Playground work?
Need to be authorized, on first the Playground screen chooses Model of AI.
In the second selector, select a library, or leave "Direct prompt to AI" and the bot will use the exposed resources.
Fill in a question in textarea and click on the “Run” button
All responses are saved and can be found on the Prompt page.