< AI />

An AI copilot that
actually knows your data.

Ask in plain English. Pilot reads your ERPNext, runs the query, respects your permissions, and gives you the answer in seconds — not the dashboard you have to interpret yourself.

Book a free consultation

Pilot is online

You

Why hasn’t SINV-2026-00417 been paid?

→ tool: get_invoice_aging(name="SINV-2026-00417")

Pilot

37 days overdue. Customer Acme BV. Last reminder sent 8 days ago. Two earlier invoices to the same customer are also open.

Ask anything

Questions you’d normally pay an analyst to answer.

“Why hasn’t this invoice been paid?”

→

Pulls customer ageing, last reminder, related open invoices — answers in one paragraph.

“Which 10 items will I run out of next month?”

→

Reads stock levels, lead times, recent sales velocity. Returns a ranked list with reorder suggestions.

“Show me last quarter’s margin by project.”

→

Queries Timesheets, Sales Invoices, and Project Costs. Returns a sorted table you can drill into.

“Is there anything weird in this month’s expenses?”

→

Runs the anomaly detector across Purchase Invoices and Expense Claims. Flags outliers with a one-line reason.

“Draft a follow-up email for everyone over 30 days late.”

→

Generates an email per customer with the right amount and dates. You approve and send.

“Daily briefing, every morning at 8am.”

→

Scheduled alerts run for you and arrive over email, Slack, Telegram, or WhatsApp.

Dashboards don’t answer questions. People do. Until now.

Your ERPNext has every fact you need. The problem is the gap between ‘the data’ and ‘the answer’ — the time it takes a person to write a report, run a filter, cross-check, and write up. Pilot closes that gap. Ask in plain English; get the answer in seconds.

What you get

An assistant that reads your data, not just your prompt.

Plain-English Q&A

Type a question the way you’d ask a colleague. Pilot picks the right query, runs it, and explains the result.

Knows your schema

Pilot understands your DocTypes, fields, and relationships — including your custom fields. Answers stay accurate as your data model evolves.

Permission-respecting

Every query runs as the asking user. Pilot never returns a row the user couldn’t see on a list view. No data leakage by design.

Tool-calling, not guessing

Pilot calls real ERPNext functions (search, aggregate, draft document) instead of inventing answers. You can audit every tool call.

Scheduled alerts

Daily briefings, weekly anomaly scans, monthly margin reports — delivered to email, Slack, Telegram, or WhatsApp. Set once, run forever.

Multi-engine, BYO LLM

Run on Claude, GPT-4, Groq, Ollama — or all of them. Pilot routes between a fast model for short asks and a strong one for analysis.

How a question becomes an answer

Ask. Pilot decides which tool. You see the work.

Ask in chat

Type the question in the floating Pilot widget anywhere in the desk.

Pilot picks a tool

Pilot reads the question, picks the right ERPNext query or report function, and runs it under your permissions.

Answer + audit

You see the answer plus the exact tool call that produced it. Disagree? Re-ask. Every interaction is logged.

Optional follow-up

Ask Pilot to draft an email, open a document, or schedule the same answer as a daily report.

Built for

Where the answer matters more than the report.

Operators and managers

People who need an answer ‘now’, not after the BI team runs the query. Daily ops, weekly reviews, monthly reports.

Finance teams

Anomaly detection, cashflow questions, customer ageing analysis — without leaving ERPNext.

Anyone tired of dashboards

If your team opens a dashboard and then asks a colleague what it means, Pilot replaces both steps.

Under the hood

Your data stays in your ERPNext. Your LLM, your choice.

Pilot runs inside your Frappe site. It calls ERPNext functions directly — not a copy of your data in someone else’s cloud. You bring your own LLM (Claude, GPT-4, Groq, or a local Ollama model). Permissions flow through ERPNext’s standard role system: Pilot can’t see what the asking user can’t. The code is open source. Every prompt, tool call, and answer is logged for audit.

An AI copilot thatactually knows your data.