What Novus AI Helps With

Novus AI helps teams investigate business questions without starting from a blank spreadsheet. Users can ask about revenue, margin, stock or returns and review the data behind the answer.

It uses ClickHouse reporting data and defined business measures to support financial and operational analysis.

How Novus AI Works

1. Defined Business Measures

Questions use a Semantic Catalogue: a set of defined measures and reporting dimensions, rather than model-generated queries against raw database tables. These include:

  • Financial & Accounting Measures: Revenue, cost of goods sold (COGS), accounting gross profit, and gross margin percentage calculated at the company and fiscal-month grain.
  • Ecommerce & Sales Metrics: Gross sales, discounts, net sales, order volume, and units sold grouped by sales channel, platform, store, product category, SKU, and city.
  • Inventory & Operations Measures: Available quantity and real-time inventory valuation sliced by warehouse and location.
  • Courier & Return Metrics: Return counts and return value categorized by courier partner, return reason, channel, and destination.

If a requested measure is not available in the reporting data, the system returns a limitation rather than an unsupported calculation.

2. Comparing Reporting Periods

For a question such as “Why did gross profit drop last month compared with the same period last year?”, the analysis:

  1. Uses the company’s fiscal calendar to select the comparison periods.
  2. Calculates contributor shares using decimal arithmetic.
  3. Isolates price, volume, and product-mix effects across product categories.
  4. Shows the source reporting data so users can review the result.

3. Drafts for People to Review

Novus AI can prepare drafts from text or documents, such as vendor price-list items or customer quotation responses. Drafts remain in Draft Status for a team member to check. Posting and approval follow role-based access control (RBAC) and the ERP’s approval rules.

4. Access to Company Data

Novus AI works within the signed-in user’s company and role permissions. Users can only request ERP data they are permitted to access. AI-assisted work remains subject to the same posting and approval controls.

People Remain Responsible for Controlled Actions

Novus AI supports analysis, document drafts and plain-English audit explanations. A draft or recommendation is not an approval: the responsible user still needs to review the work and follow the ERP’s controls.

How Novus AI Uses ERP Data

Capability A Chatbot Without an ERP Connection Novus AI by Dataicraft
Data Authority Uses the information supplied in the conversation Defined business measures backed by ClickHouse reporting data
Calculation Accuracy Calculations need to be checked against source records Deterministic SQL/analytical engine calculations
Security & RBAC No built-in knowledge of Novus user permissions Uses the signed-in user’s company and role permissions
Transaction Execution Does not itself apply Novus approval workflows Draft preparation; controlled actions follow ERP approval rules
Auditability Evidence depends on the information provided Investigation records and source data for review

Frequently Asked Questions: Novus AI

No. Novus AI is strictly barred from modifying posted entries or approving outgoing payments. Four-eyes approvals and human sign-offs remain mandatory under the ERP security architecture.
Novus AI currently connects to Ecommerce analytics, Accounting profit-and-loss variance reports, Inventory valuation marts, and plain-English audit trail explanations, with additional operational domains expanding on a governed roadmap.
Provider access, data retention and training terms need to be confirmed for your deployment. Ask Dataicraft which AI services are used and what their service agreements cover.
Dataicraft

Published by Dataicraft

Dataicraft (Private) Limited develops Novus ERP. Updated October 2026.