Does ERPNext Have AI Features? 8 Real Ones You Can Use in 2026

If you are evaluating ERPNext and asking whether it has AI, the answer is yes, just not the way listicles put it. ERPNext v16 runs on Frappe Framework v16 (released on 12 January 2026), and it gives you three main routes to AI. Raven, an open-source chat app for Frappe now maintained under Frappe's GitHub organisation, has AI agents that create documents and read receipts. Frappe Assistant Core connects ERPNext to Claude or ChatGPT over the Model Context Protocol, so you query and update data in plain English. And community apps such as changAI add plain-English reporting inside ERPNext.

ERPNext's AI is assembled from open-source building blocks plus a model you pick, not shipped as one turnkey feature. That is why it carries no licence fee and why you decide which model sees your data. I am Manoj, and I lead AI and integration work on ERPNext and Frappe at MithTech, a team of Frappe specialists in Bengaluru. This is what works now on the systems I run.

The 8 ERPNext AI features you can use in 2026

1. Natural-language chat over your live data: Frappe Assistant Core (MCP) with Claude or ChatGPT.
2. AI agents that create and update records: Raven agents, on OpenAI models or a local model.
3. Plain-English reporting for non-technical staff: community apps such as changAI.
4. Receipt and invoice data extraction: a Raven agent reads the image or PDF and drafts the ERPNext document.
5. Expense claims from photos: send a receipt to a Raven agent, get the Expense Claim back.
6. Purchase invoice drafts from supplier PDFs: an agent reads the file and prefills the Purchase Invoice for a human check.
7. AI steps in automations: n8n workflows with an LLM node classify, extract or reply, then update ERPNext.
8. Private AI on your own server: Raven bots on a local model through Ollama, LM Studio or LocalAI, so nothing goes to an AI provider.

The rest of this post covers what each costs, how to set them up safely, and how they compare with Zoho Zia, SAP Joule and NetSuite AI. For the setup path, jump to How do I connect ChatGPT or Claude to ERPNext.

Does ERPNext have AI features?

Yes, in 2026, mostly through add-on apps and model integrations. What you can do today: chat agents that create and update ERPNext documents (Raven), plain-English questions and reports on your data (Frappe Assistant Core with Claude or ChatGPT, or changAI), receipt and invoice extraction, and AI steps in n8n automations.

For a buyer, the point is where the intelligence comes from. ERPNext is open source, so instead of one vendor's AI you attach the model and app you prefer. The table maps each capability to how ERPNext delivers it.

Skimmable summary: chat, agents, plain-English reporting and document extraction all come from open-source apps (Raven, Frappe Assistant Core, changAI) and a model you choose, not from the core.

Is AI built into ERPNext natively, or is it an add-on?

It is an add-on. Conversational agents, plain-English queries and document extraction all come from separate open-source apps you install (Raven, Frappe Assistant Core, changAI) plus a model you connect. We could not find an AI feature in the ERPNext or Frappe core release notes up to v16, so treat "native AI" claims in third-party write-ups as unverified until Frappe documents one.

That is the open-source model working as intended. A proprietary suite bundles one vendor's AI and bills it into the subscription. ERPNext lets you add the AI you want, swap the model when a better one ships, and choose where it runs. The trade-off is that someone has to assemble and configure it. For the wider open-source case, see my open-source ERP guide for India.

Skimmable summary: ERPNext's AI is add-on only; you get model choice and control over the data, at the cost of some assembly.

What can ERPNext AI do today?

It can hold a plain-English conversation over your live data, create and update documents on instruction, read receipts and invoices, and run AI steps in automations. Teams use it for expense claims, purchase invoice entry and ad-hoc reporting. What it can do depends on the app and model you connect.

The uses I see pay off:

1. Expense claims from a photo. An employee sends a receipt image to a Raven agent, which reads the amount and creates the Expense Claim.
2. Purchase invoice capture. An agent reads a supplier PDF and drafts the Purchase Invoice for a human to check, which cuts keying time.
3. Ask-your-ERP reporting. Through the MCP bridge, a manager asks Claude or ChatGPT for "last month's top five overdue customers" without opening Report Builder. See natural-language reporting in ERPNext.
4. Plain-English queries inside ERPNext. changAI-style apps turn a question into the right query, so nobody needs SQL.
5. AI steps in automations. An n8n workflow classifies an incoming email and updates an ERPNext record; see ERPNext and n8n for the automation layer.

Skimmable summary: receipt-to-expense-claim, purchase invoice capture, ask-your-ERP reporting, plain-English queries and AI automation steps, each set by which app and model you connect.

How do I connect ChatGPT or Claude to ERPNext?

Install the Frappe Assistant Core app, which exposes ERPNext to any MCP client, add your site's endpoint as a custom connector in Claude or ChatGPT, and sign in with your ERPNext login over OAuth 2.0. Then scope which DocTypes the model may read or write through the AI user's roles. For agents inside ERPNext, install Raven and configure an AI bot. Both apps are open source; you pay for model usage.

The full steps (the bench install, adding the connector, scoping permissions, ChatGPT plan limits) are in my guide to connecting ERPNext to ChatGPT and Claude with MCP.

The path I follow on a client site:

1. Choose the approach. Frappe Assistant Core for a Claude or ChatGPT copilot over your data; Raven for agents inside ERPNext; n8n for AI in automations.
2. Install the app with bench get-app on self-hosted ERPNext, or from the Apps tab on Frappe Cloud.
3. Connect the model. For the MCP bridge, add the site as a custom connector in Claude or ChatGPT and sign in with OAuth. For Raven, add an OpenAI key or a local model URL in Raven Settings.
4. Scope permissions. Start read-only, then grant write access only on the DocTypes the AI needs.
5. Test on staging with real questions and a few write actions before anyone uses it in production.

Skimmable summary: Frappe Assistant Core plus an OAuth connector in Claude or ChatGPT, or Raven for agents inside ERPNext; scope permissions and test on staging first.

Is ERPNext AI free?

The apps are: Raven, Frappe Assistant Core and changAI carry no licence fee. The model behind them usually is not. Connect Claude or ChatGPT and you pay that provider for your plan or usage; run an open-source model on your own server and you pay for the compute. There is no per-user AI charge, which is the main difference from suite AI.

In budget terms: app, no fee; model, usage-based. For expense claims, invoice capture and occasional reporting, model usage scales with volume and model choice, rather than a per-seat AI charge for every user. For where AI fits in a project, see estimate your price.

Skimmable summary: no licence fee for the apps; you pay for model usage or your own compute, not per seat.

Where does your data go when you use AI with ERPNext?

It depends on the model, and it is the question to settle first.

• A cloud model API (Claude, ChatGPT, Gemini, or a model on AWS Bedrock, Azure or Google Cloud): the text the model reads (your question, plus the records, report rows or file contents it pulls) goes to that provider's servers, under its terms for your plan.
• changAI sends questions to Google Gemini by default; its schema retrieval can run on your server.
• A model on your own server (Raven with Ollama, LM Studio or LocalAI): nothing goes to an AI provider. If that server is a rented cloud machine, the data sits with your hosting provider, as the rest of ERPNext already does.

Whatever the model, ERPNext's permissions still apply: scope the AI user to what you are willing to send.

How does ERPNext AI compare to Zoho Zia, SAP Joule and NetSuite AI?

The difference is delivery and control. ERPNext lets you attach any model through open apps and pick where it runs; Zoho Zia, SAP Joule and NetSuite AI are built into the vendor's cloud and billed with the subscription. Suite AI is more polished out of the box; ERPNext AI gives you model choice but needs assembly.

If you already pay for a proprietary suite and want AI that works without assembly, the vendor's AI is the easier path. If model choice, data residency and no per-seat AI charge matter more, ERPNext's approach fits, provided someone sets it up properly.

Skimmable summary: ERPNext AI is model-agnostic with no app licence fee; suite AI is more polished but tied to the vendor's model and subscription.

Should an Indian SMB use ERPNext AI in 2026?

Selectively: start with one or two well-bounded uses rather than "AI everywhere". Receipt-to-expense-claim and purchase invoice capture are easy to bound and check; open-ended agents with broad write access are where projects go wrong. The value comes from tight scope, staging tests and a human check on anything the AI writes.

Treat AI as an accelerator on a clean ERPNext, not a substitute for one. If masters, workflows and GST setup are messy, AI makes the same mistakes faster. Get the ERP right, then add AI to the two or three tasks that take the most manual time. Start small, measure, then expand.

Frequently asked questions