Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 7 additions & 7 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ Open-source AI assistant for ERPNext. Ask business questions in plain English an
</div>


> **Note:** The current version is trained on ERPNext modules only. Like any AI model, it is still learning and handles a good range of ERPNext queries well, but will not get everything right. Accuracy improves over time with more training data and feedback.
> **Note:** The current version is trained on ERPNext modules only. Like any AI model, it is still learning and handles a good range of all ERPNext queries well, but will not get everything right. Accuracy improves over time with more training data and feedback.


## Table of Contents
Expand All @@ -26,7 +26,7 @@ Open-source AI assistant for ERPNext. Ask business questions in plain English an
- [Usage](#usage)
- [How It Works](#how-it-works)
- [FAQs](#faqs)
- [Links](#links)



## Key Features
Expand All @@ -35,7 +35,7 @@ Open-source AI assistant for ERPNext. Ask business questions in plain English an

2. **RAG-Powered Schema Retrieval** — changAI uses Retrieval-Augmented Generation to identify the most relevant tables, fields, and master records from your ERPNext schema and Master Data before generating a query. In Local Mode, this retrieval runs entirely on your own server using the downloaded embedding model, keeping your schema data on-premise. This improves accuracy across complex, multi-table questions.

3. **Permission-Aware Results** — Every query is validated through Frappe's native permission layer using the match_conditions API. Users only see data they are authorized to access, making changAI safe to deploy across teams with different roles.
3. **Permission-Aware Results** — Every query is validated through Frappe's native permission layer using the match_conditions API. Users only see data they are authorized to access

4. **Flexible AI Engine Configuration** — changAI supports multiple AI backends. Local Mode using Google Gemini is the recommended configuration for all users. Gemini is available on a free tier for testing via Google AI Studio and on an enterprise Vertex AI tier for production workloads. Remote Mode using Qwen3 via Replicate exists as an option in settings but is not currently in a stable working phase and is not recommended for production use at this time.

Expand Down Expand Up @@ -130,7 +130,7 @@ Both schema retrieval and SQL generation are handled by remote Replicate server
2. Under the Remote tab, fill in the following fields:
- Replicate API token
- Prediction URL
- Version IDs for the deployed models
- Version IDs
3. Save the settings
> 📺 **Full walkthrough:** [YouTube Setup Guide](https://youtu.be/fFxyAY_sVNs)
> 📖 **Full docs:** [docs.claudion.com](https://docs.claudion.com/Claudion-Docs/changaisetup)
Expand Down Expand Up @@ -207,9 +207,9 @@ Natural Language Result is returned as a human-readable answer
**Do I need to know SQL to use changAI?**
No. changAI is built for non-technical users. You type a question in plain English and the system handles schema lookup, query generation, and result formatting automatically.

**Which ERPNext versions are supported?**
**Which ERPNext versions are supported, Does it supports V15?**
changAI supports ERPNext v15, and v16 on Ubuntu with Python 3.14 or higher.
**Note** - Python 3.14 requires sudo apt-get install build-essential python3-dev before bench get-app
**Note** - Python 3.14 requires build-essential python3-dev before bench get-app

**Which modules does changAI cover out of the box?**
changAI ships pre-configured with the standard ERPNext schema, so modules like Accounts, Inventory, Purchasing, Sales, and HR work immediately after setup without any additional mapping. Custom doctypes and fields require a schema sync using an Anthropic Claude API key.
Expand All @@ -236,7 +236,7 @@ This can happen when the model fails to identify the correct tables or fields fo
Go to changAI Settings and click Download Embedding Model manually. Make sure your server has outbound internet access at the time of download. If you are on a restricted network, you may need to whitelist the Hugging Face domain.

**Is my ERPNext data sent to external servers?**
In Local Mode, schema retrieval runs on your server and only the SQL query context is sent to Gemini for generation. In Remote Mode, schema retrieval also runs on Replicate. If you enable custom schema sync, your schema structure is sent to the Anthropic Claude API. Review your data sharing and compliance requirements before choosing a configuration.
No. In Local Mode, schema retrieval runs on your server and only the SQL query context is sent to Gemini for generation. In Remote Mode, schema retrieval also runs on Replicate. If you enable custom schema sync, your schema structure is sent to the Anthropic Claude API. Review your data sharing and compliance requirements before choosing a configuration.

**How do I improve accuracy over time?**
Run the Train Data Automation feature to generate additional training data from your ERPNext modules. You can also re-sync the Master Data Schema after adding new doctypes or fields, and use the schema update feature to keep changAI aligned with any ERPNext customisations.
Expand Down
Loading
Loading