2024 - GPTBots

08-30

Knowledge Base 2.0

Supports the "Retrieval Enhancement" feature, enabling enhancement and rewriting of the user's latest question based on "context." This can address the need for knowledge retrieval in consecutive dialogue sessions, making knowledge retrieval more precise;
Knowledge selection is adjusted to the "knowledge base" granularity, making knowledge management more convenient. The knowledge base supports the "on-demand invocation" mode, better supporting Bots with multiple knowledge bases, and retrieval is optimized for precision;
Knowledge reference source display supports 3 styles (minimalist mode, dropdown list, citation tag), meeting the needs of different types of customers;
The knowledge base has optimized the methods and capabilities for adding knowledge;
- PDF has added "advanced recognition" capability, improving the recognition accuracy of PDF files;
- Spreadsheet files support multi-sheet recognition, support for image processing within tables, and support for data display and editing in table form;
- Web scraping has added a "sitemap" mode;
- Scheduled updates support updates at the Google Drive file, folder, and shared folder dimensions;

Telegram Platform Support

Supports publishing Bots to the Telegram platform, with automatic synchronization of introductions, welcome messages, and guide questions;

AI Search Mode Support

Supports publishing regular Bots as AI Search mode, supporting multiple sessions, features like knowledge source identification and object extraction;

08-15

Flow 2.0

Optimize the style of Flow component and linking animation, support automatically following the currently active node when running flow;
Support to view the input/output details of a component;
Support for debugging individual components;
Support customized input data structure for LLM component, branching component, and conditional judgment, so that you can define the data structure more flexibly and make the AI response more stable and accurate;

Flow - Conditional Judgment

New Conditional Judgment component supports LLM to make comprehensive judgment on conditional rules and user input;
The rule descriptions of the conditional judgment support natural language, code and regular expressions;

Human Services - WebHook

The human service component supports WebHook service, which allows developers to forward human service messages to any business scenarios (DingTalk, internal business system, WeCom, etc.);

06-11

Bot Config

Supports integrating human customer service via Intercom for the Bot;
Supports configuring user attributes, automatically collecting user attribute information during conversations, and using this information as variables in welcome messages, product recommendations, user identity memory, and other business scenarios;

Knowledge Base

Optimized the parsing and storage capabilities of knowledge document data, automatically supporting vector storage of images and implementing LLM's image-text responses;
Optimized the document segmentation mechanism, preserving the semantic integrity of each knowledge block through codes, tables, and other preservation mechanisms to the greatest extent;
Google Drive supports setting automatic document synchronization;
Supports re-ranking recall data based on multiple Rerank models;
Supports custom keywords for table-type document knowledge blocks;
Optimized the document-to-Q&A conversion effect;
Optimized the efficiency and stability of vector retrieval;

Bot Integration

Bubble widgets and iframes have optimized custom UI configuration options, supporting more powerful UI customization capabilities;
Bubble widgets support proactive messages, set triggering conditions to display proactive messages, enhancing user interaction rates;
Supports more refined security frequency control settings to prevent waste of points. Also supports Webhook and GA4 event callbacks;

Tool

The official OpenTool has added nearly 20 new tools, including Google, arXiv, Hubspot, img2txt, YouTube, etc., covering scenarios such as search, business, images, videos, customer management, academia, marketing, and more. Welcome to explore!

05-17

LLM

Added support for the "gpt-4o" model.
Added support for several LLMs including Qwen, Llama, Mistral, and Claude.

Bot Publish

Support for using custom domains on "Share, Iframe, Bubble component," allowing users to access the Bot through a custom domain.

04-29

Official Tool

Code Interpreter: It can automatically write and execute code according to user requirements, completing complex tasks that require code execution such as data analysis, mathematical calculations, and file conversions.
Hubspot Contacts: This tool can synchronize customer contact information collected by the Bot into your Hubspot CRM and create contacts. Using this tool requires authentication with your Hubspot API Key.
Semrush Keyword Overview: This tool allows you to query and analyze keyword performance in search engines through Semrush's keyword analysis service, including key data such as competitiveness, search volume, and CPC. You can use this tool directly without authentication.
Semrush Related Keywords: This tool allows you to find related keywords through Semrush's keyword analysis service and provides you with extensive data for analysis. You can use this tool directly without authentication.

Bot Templates

Added a batch of Bot templates, including but not limited to customer service, lead generation, data analysis, SEO, marketing, etc., helping you quickly leverage the capabilities of GPTBots to truly enhance your business with AI.
Optimized the process of creating Bots through templates.

Bot Conversations

Supports submitting files as conversation input and reading file content as conversation context.

Bot API

Send Message: Added support for files as input parameters.

04-08

Beginner's Guide

Added a beginner's guide task, which, according to system instructions, completes tasks, earns credits rewards, and also helps new users better understand the core usage and value of GPTBots.

Bot Configuration

Supports creating Bot copies.
Redesigned the Bot configuration interface for improved interaction.

Bot Knowledge Base

Supports retrieving documents from "Google Drive" and adding them to the knowledge base.
Supports uploading up to 200 documents at once and optimized the uploading experience.

Bot Chat Records

Recorded message history by "Q&A" dimension and supports data export.

Bot API

Send Message: Supports using a specific knowledge document ID as an input parameter to retrieve knowledge only within the scope of this round of conversation as reference information for the LLM to summarize responses.

03-18

Bot Configuration

Supports using AI to generate "Identity Prompts": If you think your written identity prompts are not effective, you can let AI write them for you. AI will use the most scientific method to write the most suitable identity prompts that LLM can understand.
AI-generated "Welcome Messages": AI can optimize a better welcome message based on your current one.
Supports configuring multiple "preset questions" to guide users to ask questions when starting a conversation with the Bot.

Bot Knowledge Base

Upload Files: Increased the single upload file limit to 200 and optimized the file uploading process to make it clearer.
Scheduled Updates: Web page content always changes, and manually updating to get the latest web page content into the knowledge base is very troublesome! Now you can set scheduled updates, and the system will automatically update the web page content in the Bot based on your settings, ensuring that your Bot's web page knowledge data is always up-to-date.
Supports adding or deleting knowledge slices.

Bot Integration

Discord: Simplified the process of integrating Bots into Discord, improved the Bot's online duration within Discord, and supported streaming returns within Discord.

Bot API

Generate Suggested Questions: This function is separately packaged as an API, through which the system can recommend three questions to users based on the context.
Retrieve Referenced Knowledge: This function is separately packaged as an API, through which you can obtain the knowledge data used in the Bot's response.
Retrieve Conversation List: Added the ability to query the conversation list for a specific 'user_id'.
Send Message: Added 'short_term_memory' and 'long_term_memory' boolean parameters as input parameters, using these parameters can control whether the current message carries memory as context.
Upload Files: Upload files to the Bot knowledge base.
Query Document Status: Query the current embedding status of the document.
Re-embed: Re-embed all documents in the current Bot that are in 'failed' status.
Bot Reply Feedback: Allows users to 'vote' on the Bot's reply content, either 'like' or 'dislike', to provide feedback to the Bot developers to understand the user's experience and thereby optimize the Bot's prompts or knowledge base.

Bot Chat Records

Added a 'Q&A' list and supports downloading and training Q&A.

Dialogue Components

Optimized Bot dialogue components for debugging, sharing, iframe, and other scenarios to enhance user experience.

System Stability

Performed performance optimizations on the system to enhance the product's performance, including APIs, knowledge base, LLM, etc.

02-19

Lower ChatGPT Point Consumption

Lowered the credit consumption of the OpenAI's gpt-3.5-turbo model.

Design and Interaction Experience Optimization

Thoroughly upgraded the official website and client. Not only optimized the visuals but also enhanced the interaction experience, allowing you to browse and use GPTBots more efficiently and comfortably.

Bot/FlowBot Creation

Supports creating Bots/FlowBots with templates. GPTBots officially provides a variety of Bot/FlowBot templates in different fields, allowing you to easily create Bots/FlowBots according to your needs at any time and start using them immediately, enjoying the convenience brought by AI.

01-12

LLM

Added support for the model 'ChatGPT4.0-turbo-vision-1106-128K'.

Bot Configuration

Multimodal Input: Added support for 'images' as input information.
Multimodal Output: Added support for 'images' and 'voices' as output information, 'images' support displaying multiple, 'voices' support setting 60+ languages, and support setting the output language of the Bot.

Flow Components

OpenAI LLM: Added support for the model 'ChatGPT4.0-turbo-vision-1106-128K'.
Branch Judging: Added support for the model 'ChatGPT4.0-turbo-vision-1106-128K'.
Boolean Judging: Added support for the model 'ChatGPT4.0-turbo-vision-1106-128K'.

Bot API

Dialogue Generation: Added image information to the return fields.
Added 'Text to Speech' API, supporting converting AI output information into playable voice.

Bot Dialogue

Input: Supports using images as input information.
Output: Supports playing AI output information through voice.

Bot Integration

Added 'Slack', 'Discord', and 'Zapier' integrations, allowing you to integrate Bot services on these three platforms.
Added 'WhatsApp by EngageLab', supporting opening a 'Meta Business' account through the 'EngageLab' platform and creating a 'Meta App' to integrate with the Bot. 'EngageLab' is an official certified BSP (Business Solution Provider) by Meta, making it more convenient to interface with Meta applications than directly integrating through the Meta platform.

Note: 'Zapier' is in the process of being released and will be completed later this month.

Organization

Time Zone: Allows setting different time zones for different organizations, and the time information within the organization management module will be displayed based on that time zone.
Points: Adjusted credits management from the original account dimension to the organizational dimension, making credit management more efficient and usable.
Permissions: Supports setting Bot permissions for 'roles'.

Subscription (Add-on Service)

Monthly Points: Supports subscribing to monthly credits for organizations, to prevent the current organization's plan from running out of credits.
Custom LOGO: You will be able to customize the LOGO of the Bot dialogue component, either by removing it or replacing it with your company or organization's LOGO, to enhance brand value.