Gomus AI Documentation

Gomus AI is an AI-powered knowledge platform built on a state-of-the-art RAG (Retrieval-Augmented Generation) engine. Upload documents, crawl websites, and build knowledge bases that your AI assistants and agents can query with precision. Every answer includes chunk-level citations so you can verify the source instantly.

Create an Account

Your First Knowledge Base

Your First AI Chat

Knowledge Bases

A knowledge base is a collection of parsed documents and web pages that serve as the context for your AI chats and agents. Gomus AI parses your files into semantic chunks, builds embeddings, and indexes everything for high-precision retrieval.

Create a Knowledge Base

Navigate to the Knowledge Base tab and click Create Knowledge Base. Each knowledge base has its own configuration for chunking method and embedding model.

Configure a Knowledge Base

After creating a knowledge base, configure it by selecting:

• Chunking method — How files are split into retrievable chunks
• Embedding model — Which model converts chunks into vectors
• Similarity threshold — Minimum relevance score for retrieval (default: 0.2)
• Vector similarity weight — Balance between vector and keyword search (default: 0.3)

Chunking Methods

Gomus AI offers multiple built-in chunking templates optimized for different document types:

You can also override the chunking method per file after upload.

Embedding Models

An embedding model converts your document chunks into vector representations for semantic search. Available embedding models:

Upload & Parse Files

After configuring your knowledge base:

1. Click + Add file to upload documents from your device
2. Click the play button on each file to start parsing
3. Once parsed, click the file name to view individual chunks
4. Double-click any chunk to add keywords, questions, or make manual edits

Retrieval Testing

Before setting up a chat, test your knowledge base retrieval:

1. Open your knowledge base and go to Retrieval Testing
2. Type a test question in the Test text field
3. Review the retrieved chunks and their similarity scores
4. Adjust Similarity threshold and Vector similarity weight as needed

Web Crawler

Turn any website into a searchable knowledge base. Base+

1. In your knowledge base, click + Add file and select Web Crawl
2. Enter the root URL of the website you want to crawl
3. Configure crawl depth and page limits
4. Gomus AI will crawl all linked pages, extract content, and index it

The crawler respects robots.txt rules and you can schedule periodic re-crawls to keep content fresh. Crawled web pages and uploaded documents can coexist in the same knowledge base for unified retrieval.

Chat Assistants

Chat assistants let you have AI-powered conversations grounded in your knowledge bases. Each answer is backed by citations from your documents, reducing hallucinations and ensuring accuracy.

Create a Chat Assistant

Chat Configuration

Each chat assistant has its own configuration:

• Model — Choose any available LLM (can differ per assistant)
• Knowledge bases — Select which knowledge bases to query
• System prompt — Customize the AI's behavior and persona
• Empty response — Response when no relevant information is found
• Multi-turn optimization — Uses conversation history to refine queries (enable under Prompt Engine)
• TopN — Number of chunks to retrieve (reduce to speed up responses)
• Similarity threshold — Minimum relevance score for chunks

AI Search vs AI Chat

Use AI Search for quick one-off queries and retrieval debugging. Use AI Chat for production conversations.

Agent Builder

The Agent Builder is a visual drag-and-drop editor for designing multi-step AI workflows. Connect retrieval, generation, web search, conditional logic, and custom components into powerful pipelines — no code required. All Plans

Create an Agent

Agent Components

Pre-built Templates

Gomus AI provides ready-to-use agent templates:

• Deep Research — Multi-step research with web search + knowledge base retrieval
• E-Commerce Customer Support — Intent classification + knowledge-grounded responses
• Ingestion Pipeline — Custom document parsing and indexing workflow
• Sandbox — Code execution for data analysis tasks

Available Models

Gomus AI provides 40+ pre-configured AI models from leading providers. All models are ready to use — no API keys required from your side.

Chat Models (LLMs)

Embedding Models

Reranking Models

Models by Tier

API Reference

Gomus AI provides RESTful APIs and an OpenAI-compatible endpoint to integrate AI capabilities into your applications. Base+

Authentication

All API requests require an API key. To get your key:

1. Click your avatar (top right) and select API Key
2. Click Create new key and copy the generated key
3. Include it in the Authorization header of every request

Authorization: Bearer YOUR_API_KEY

Base URL

https://app.gomus.org/api/v1

Error Codes

Chat Completion (OpenAI-Compatible)

Gomus AI supports the OpenAI chat completion format, making it easy to integrate with existing tools and libraries.

POST /api/v1/chats_openai/{chat_id}/chat/completions

{
  "model": "model",
  "messages": [
    {"role": "user", "content": "What does the contract say about termination?"}
  ],
  "stream": true
}

Request Headers

Content-Type: application/json
Authorization: Bearer YOUR_API_KEY

cURL Example

curl -X POST https://app.gomus.org/api/v1/chats_openai/{chat_id}/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "model",
    "messages": [{"role": "user", "content": "Summarize the key findings"}],
    "stream": true
  }'

Parameters

Datasets API

Manage knowledge bases and documents programmatically.

Agents API

Billing & Plans

Credit System

Gomus AI uses a credit-based system to track AI model usage. Different models consume different amounts of credits per request:

• Free models (Nova Micro, Llama 1B/3B) — Minimal credit usage
• Standard models (Nova Pro/Lite, Claude Haiku) — Moderate credit usage
• Premium models (Claude Sonnet/Opus) — Higher credit usage

Credits reset monthly. You can monitor your usage from Account Settings. If you exhaust your credits, you can wait for the monthly reset or upgrade your plan.

Frequently Asked Questions

What is Gomus AI?

Gomus AI is an AI-powered knowledge platform built on a state-of-the-art RAG (Retrieval-Augmented Generation) engine. It combines deep document understanding with visual AI agent orchestration and web crawling, all in one platform. Upload documents or crawl websites, then get precise, cited answers from your data.

What makes Gomus AI different from ChatGPT?

ChatGPT answers based on its training data. Gomus AI answers based on your documents. Every answer includes citations pointing to the exact source chunks, so you can verify accuracy. This dramatically reduces hallucinations and makes it suitable for professional use cases where accuracy matters.

Is my data secure?

Yes. All data is processed and stored on AWS infrastructure in the EU (Frankfurt, eu-central-1). Your documents are never used to train AI models. All data is encrypted at rest and in transit. The platform is designed with GDPR compliance in mind.

Which file formats are supported?

Gomus AI supports a wide range of file formats:

Can I use the API to integrate Gomus AI into my app?

Yes! Gomus AI provides RESTful APIs and an OpenAI-compatible chat completion endpoint. See the API Reference section for details. API access is available on all plans.

How does the Web Crawler work?

Paste the root URL of any website. Gomus AI crawls all linked pages, extracts content, cleans HTML into structured text, and indexes everything into a knowledge base. You can then query the entire site through chat, just like uploaded documents. The crawler respects robots.txt and you can configure crawl depth and filters.

What is the Agent Flow Builder?

A visual drag-and-drop canvas where you design multi-step AI workflows. Connect components like knowledge retrieval, LLM generation, web search, conditional logic, and custom tools into a pipeline. No coding required. See Agent Builder for details.

Can I switch AI models per conversation?

Yes. Each chat assistant can use a different AI model. You can also change models within the same assistant at any time.

Do you support multi-turn conversations?

Yes. Enable Multi-turn optimization in your chat assistant's settings (under Prompt Engine) to let the AI use previous messages as context for better follow-up responses.

What happens when I run out of credits?

When your monthly credits are exhausted, AI model requests will be paused until credits reset at the start of your next billing cycle. You can upgrade your plan at any time for additional credits. Free-tier models with minimal credit usage may still be available.

Contact & Support

• Email: support@gomus.org
• Website: www.gomus.org
• App: app.gomus.org

© 2025-2026 Gomus Digital Agency. All rights reserved.