HaROS · Comnous AI

AI at the center of operations

Comnous AI is not bolted on — it has been inside the platform from the start. Demand forecasting, dynamic allocation and anomaly detection run right inside the operational flow. Not an add-on, but the way the system works.

In brief

Comnous AI comes down to three capabilities: forecast, allocate and detect. All three run on Data Fabric (a single data layer). Stock, order, product and customer data are seen at once, and decisions are made automatically inside the operation. With Comnous AI Assistant you ask this data questions in natural language.

Comnous AI

Forecast · Allocate · Detect

Comnous AI is built on three capabilities. All three share the same data: forecasting feeds allocation, and allocation feeds detection signals.

01

Forecast

predict

HaROS's demand forecasting engine, Kahin, analyzes sales, historical data, live stock status and traffic to produce forecasts at SKU level. Season planning and automatic replenishment run on this forecast. You see in advance which stock will be needed, when and where.

Example: Before a campaign, the recommended stock level for each warehouse is calculated automatically; replenishment tasks are opened in advance for SKUs that would fall short.
02

Allocate

allocate

Each incoming order is automatically assigned to the best-fit point where the stock is. Channel priority, location, stock status and shipping cost are evaluated together; the decision is made by defined rules and AI together.

Example: When the same SKU is in all three warehouses, the order is fulfilled from the warehouse that best fits delivery time and cost. If needed, the order is split and shipped from more than one point.
03

Detect

detect

Catches anomalies in operations in real time: short-picks, phantom stock, unusual returns or fraud signals. Price optimization works in this layer too; deviations are flagged before the problem grows.

Example: When stock that shows in the system but is not on the shelf, or an unexpected short-pick pattern, is noticed, the relevant team gets an alert.
Capabilities

Comnous AI working inside operations

Capabilities you use in daily operations. All run on a single data layer; no separate tool or integration is required.

→ Forecast

Demand forecasting

Kahin produces demand forecasts by SKU, channel and location; they are updated with season and campaign signals.

→ Forecast

Automatic replenishment

The forecast flags stock that would fall short; replenishment tasks are generated automatically without manual follow-up.

→ Allocate

Dynamic allocation

The order is assigned from the best-fit point where the stock is, evaluating channel priority, cost and delivery time together.

→ Detect

Price optimization

Price recommendations based on demand, stock and channel data; they update with data instead of a fixed list.

→ Detect

Anomaly and fraud detection

Short-pick, phantom stock, unusual return and fraud patterns are flagged in real time.

→ Data Fabric

One data layer

Stock, orders, products and customers meet in the same data. Because Comnous AI works with all of it, decisions are consistent.

Comnous AI Assistant

Query operations in natural language

The operations assistant that works inside HaROS

Comnous AI Assistant is the operations assistant that works inside HaROS. You ask about stock, order and performance data in natural language and get answers without browsing screen after screen for a report.

You do not need to move data to a separate reporting tool. The assistant works on the same Data Fabric; everything it sees is real-time operations data.

Today
Current capabilities

Capabilities available today: knowledge and data

These capabilities make it easier for teams to reach corporate knowledge and to prepare operations data for target systems. Scope is determined by connected systems, user permissions and the approval rules defined at setup. They are separate from the autonomous agent vision in the roadmap and are available today.

→ Knowledge

Chat with corporate knowledge

Ask questions in natural language over your connected files and knowledge sources; the answer is given based on the relevant source.

→ Knowledge

Task-specific bots and skills

Set up assistants with knowledge scope and tools defined for different business needs.

→ Data

Dataset workspace

Review and edit datasets on a table; changes are tracked with version history.

→ Data

Sourced enrichment suggestions

Information taken from web addresses you specify is offered as suggestions with their source; you approve them.

→ Template

Template-ready Excel preparation

Map source columns to the target template's fields, check allowed attribute values and get a downloadable draft file.

→ Template

Missing field and mapping report

Reports missing required fields and mapping mismatches; shows whether the file is ready to upload.

→ Analysis

Analysis and reports on your data

Query your datasets, see results with their scope and turn them into Excel output.

→ Control

Permission and approval mechanisms

Operations in connected systems run with access rules, human approval when required and an activity log.

→ Control

Usage and budget tracking

You track Comnous AI usage and its budget on dedicated screens.

Example use: from a product dataset to a marketplace template

  1. 1. Source and templateThe team provides a product dataset and the target marketplace's upload template.
  2. 2. MappingThe assistant helps map source columns to template fields and checks allowed attribute values.
  3. 3. GapsMissing required fields and mapping mismatches are reported; whether the file is ready to upload is indicated.
  4. 4. DraftA checkable Excel draft is produced; the user completes the gaps and reviews the result.
Roadmap
In development — not in use yet

Autonomous AI agents vision

The capabilities below are on the Comnous AI roadmap. Development is ongoing; they will be added to the platform as they are completed.

Roadmap

Comnous Agent Studio

A layer for building AI agents that will run your operation-specific workflows on their own. Its foundation today is Comnous AI Assistant; fully configurable agent creation is in development.

Roadmap

Comnous Agentic AI

Assistants that do not just anticipate operations but take action on their own within the limits you define.

Frequently asked questions

Is Comnous AI part of the platform, or was it added later?

Part of the platform. Forecast, Allocate and Detect run on HaROS's single operational data layer (Data Fabric) — it is not a separate product or an add-on. AI added afterward stays outside the operational flow; in HaROS, decisions are made right inside the flow.

Which decisions does Comnous AI automate?

Decisions such as demand forecasting, automatic stock replenishment, dynamic order allocation (from the best-fit point where stock is), price optimization signals and anomaly/fraud detection (short-pick, phantom SKU, etc.) can be automated.

Is my data safe?

Yes. Comnous AI works only with your own operations data, and every decision it makes can be audited.

Is Comnous AI Assistant available today?

Yes. Comnous AI Assistant is available today; you query operations data in natural language. Comnous Agent Studio and Comnous Agentic AI on the roadmap are in development.

See AI in your own operation

In a short discovery call, let's show forecasting, allocation and detection on your own data.

Request a demo