Skip to main content

​
1. Overview

WorkBuddy is Tencent’s AI Agent productivity product built around “describe the request, run the task, deliver complete results”. Unlike traditional chat AI that only gives advice or text replies, WorkBuddy understands natural-language instructions, decomposes tasks, plans steps, and executes them. It supports multimodal work such as documents, spreadsheets, PPT, and data analysis, can read authorized local folders for batch processing, and delivers reviewable outcomes (document generation, PPT design, coding assistance, data analysis, and more). WorkBuddy ships with mainstream models such as Hunyuan, GLM, MiniMax, Kimi, and DeepSeek (via Tencent Cloud Token Plan), and also lets you connect any third-party model under Model Settings. With GravitexAI you get:

​
2. System Requirements

  • Windows 10 or later (Windows 7/8/8.1 not supported)
  • macOS 12 (Monterey) or later
Systems that do not meet the requirements above cannot launch WorkBuddy.

​
3. Prerequisites

  • WorkBuddy is installed on your machine
  • You have created an API Key in the GravitexAI console (format sk-xxxxxxxxxx)

​
4. Quick Start

​
Step 1: Install WorkBuddy

WorkBuddy provides Windows / macOS desktop clients. Download the installer from the official site and double-click to install — no CLI required: After install, open WorkBuddy, sign in, and start tasks from the new-task bar in one sentence, or follow the official Quick Start guides.

​
Step 2: Get a GravitexAI API Key

  1. Open the GravitexAI console
  2. Click Create Token, name it, and save
  3. Copy the key starting with sk-

​
Step 3: Open Custom Model in WorkBuddy

In WorkBuddy, click your avatar (bottom left) → Settings → Model → Add Model, then fill in:
Use the full endpoint https://api.gravitex.ai/v1/chat/completions (including /chat/completions, as shown in the screenshot). Using only https://api.gravitex.ai or https://api.gravitex.ai/v1 will fail.
WorkBuddy Edit Model configuration

​
Step 4: Advanced settings

For standard providers such as Tencent Cloud Token Plan, tool calling / image input flags are filled automatically. For Custom + GravitexAI they are not auto-detected — enable them based on the model’s real capabilities:
  • Match checkboxes to the actual capabilities of the model ID you entered; when unsure, enable fewer rather than more — the claude-sonnet-4-6 example above only enables Tool Calling
  • Only enable Image Input / Thinking Mode after confirming the model supports them
  • Mismatched flags (e.g. tool calling on a model that does not support tools) may cause errors

​
Step 5: Input / Output limits

Use Input / Output to set context length and max output tokens. Leave blank for provider defaults, or pick presets such as 32K/64K/128K/256K (input) and 8K/16K/32K/64K (output). Click Save, then select the model under the custom group in the chat UI.
  • WorkBuddy’s model dialog only supports OpenAI-compatible APIs (labeled at the top of the dialog).
  • GravitexAI provides a standard OpenAI-compatible API. Choose Custom to access the full model matrix (GPT / Claude / Gemini / DeepSeek / Zhipu / Kimi, and more).

​
Step 6: Switch models

Back in the editor, open the model dropdown — your added models appear in the list. Select one to start chatting.

​
5. Feature Cheat Sheet

​
6. FAQ

No. Official docs state model config (including API Key) is stored only in local workbuddy/models.json and is not uploaded. WorkBuddy acts as a relay to your GravitexAI endpoint; except for necessary transmission, security audit, troubleshooting, and legal retention, WorkBuddy does not read or store conversation content.
No. Custom model costs (tokens, subscriptions, etc.) are billed directly by GravitexAI and do not use WorkBuddy credits or plans. Monitor your GravitexAI balance and usage to avoid unexpected charges.
WorkBuddy’s custom model dialog currently supports OpenAI-compatible APIs only. Use GravitexAI’s OpenAI-compatible endpoint https://api.gravitex.ai/v1/chat/completions. Anthropic native endpoints are not supported yet.
Standard providers fill these automatically; Custom + GravitexAI does not. Enable based on the model ID’s real capabilities — when unsure, enable fewer options. The claude-sonnet-4-6 screenshot only enables Tool Calling. Mismatched flags may cause errors.
Use the GravitexAI model ID, e.g. claude-sonnet-5, gpt-5.5, deepseek-v4-pro, gemini-3.6-flash, kimi-k3 — see the model catalog. To switch models, only change this field and save; endpoint and API Key stay the same.
  1. Confirm the API key starts with sk- and has no extra spaces
  2. Confirm the endpoint spelling, especially /v1/chat/completions
  3. In the GravitexAI console, check the token is enabled and balance is sufficient
  4. Insufficient balance can also return 401