Skip to main content
Apragya AIFree quotation builderQuotation builderQuote builderno signup
AI agent

AI Agent for Demand Forecasting: Reorder From What Is Selling, Not From Habit

For purchase managers, store owners and planners in Indian trading and manufacturing businesses: keep fast movers in stock without filling the godown with slow ones.

A purchase manager checking a nearly bin in a hardware godown

An AI agent for demand forecasting predicts what will sell and turns that into reorder suggestions. In many Indian businesses purchasing still runs on habit: the same quantity every month, topped up before Diwali, with a phone call when something runs out.

Habit works until demand shifts. A new product line takes off, a customer switches suppliers, or a monsoon delays deliveries, and the godown is full of the wrong things.

This page covers the ready-made forecasting and reorder agents in Apragya, an illustrative example, and how purchase keeps the final say.

An example: a Madurai hardware and sanitaryware distributor

Take an illustrative distributor stocking 4,000 items across two godowns, supplying plumbers, builders and retail shops, with suppliers in Gujarat and Rajasthan. Lead times range from four days to three weeks, and demand swings with the building season.

From sales history to a purchase order

  1. Sales trend read

    Recent consumption and seasonality per item.

  2. Stock checked

    On hand, reserved and days of cover.

  3. Lead time applied

    Shortfall against the supplier's delivery time.

  4. Reorder suggested

    Quantity, urgency and a one-line reason.

  5. Purchase decides

    The buyer raises the order, or holds.

The arithmetic is exact; the agent makes the call and explains it.

Ahead of the building season the agent raised urgency on CPVC fittings, whose sales had climbed for six weeks, and suggested holding a slow-moving range of taps. The buyer ordered the fittings early and avoided the usual mid-season shortage.

A full crate of fittings with a green ribbon next to a set-aside crate of taps
Fittings ordered early for the season, and the slow range of taps held back.

How inventory forecasting works in Apragya

Step 1 of 5

Demand Forecaster

Predicts demand per product from sales trends, seasonality and movement.

  1. Demand Forecaster: Predicts demand per product from sales trends, seasonality and movement.
  2. Reorder arithmetic: Consumption rate, days of cover and shortfall are computed exactly.
  3. Reorder suggestions: An agent decides urgency and quantity, with a rationale.
  4. On a screen: Recommendations are saved where the buyer can act on them.
  5. Buyer acts: Raise a material request or purchase order, or hold.

What forecasting by habit costs

Stock-outs on fast movers

Sales are lost while the reorder is still being discussed.

Slow stock piling up

Working capital sits in items that are not moving.

Lead times ignored

Orders placed too late for the supplier's delivery time.

Seasons missed

Festive and seasonal peaks are planned from memory.

Choosing inventory forecasting software

Good forecasting separates the exact from the judgement. Stock, consumption and lead time are numbers the system should compute precisely. How urgent a shortfall is, and whether a spike will last, is where an agent helps. Look for both, and for suggestions a buyer can see and act on.

Reordering by habit, against Apragya

FeatureBy habitApragya
QuantitySame as last monthBased on recent consumption
Lead timeRememberedApplied to the shortfall
UrgencyWhoever shouts loudestRanked with a reason
Slow moversReordered anywaySuggested to hold
DecisionBuyerBuyer, with the evidence

Checked against public information on 16 September 2026.

What sets these agents apart

Ready-made

Demand Forecaster and reorder suggestions come with Inventory.

Multi-godown

Stock and cover by warehouse.

Batches and expiry

Near-expiry stock is considered before new orders.

Manufacturing link

A Demand Planner proposes production plans from open orders.

Per company

Forecasts stay within each company's stock.

Metered AI use

AI Credits are tracked per company.

Forecasts use stock levels and stock movements, and suggestions turn into purchase orders.

Getting started with stock replenishment

  1. Make sure sales and stock movements are recorded for at least a few months.
  2. Set supplier lead times and reorder levels for key items.
  3. Run the Demand Forecaster on one product group.
  4. Compare its suggestions with your buyer's plan.
  5. Act on the suggestions for that group for a month.

One product group is enough to see whether the suggestions match your market.

What it costs to start

INR 3,999
Starter, a month
Up to 10 users, CRM and HR plus Inventory and Warehousing
2,000
AI Credits a month
Included on Starter
Free
Up to 3 users
CRM, HR and Vendor Onboarding

Demand forecasting agent, answered

Is this ready-made?
Yes. Inventory includes a Demand Forecaster and an AI reorder suggestions agent. Manufacturing adds a Demand Planner for production plans.
Does it place purchase orders?
No. It suggests. A buyer raises the order.
How much history does it need?
The more the better. A few months of sales and stock movement is a sensible start, and a year shows seasons.
Can it handle new products?
New items have little history, so treat early suggestions with care and set reorder levels by judgement.
Does it account for festive and seasonal peaks?
It reads seasonality from past sales, so a year of history helps. For a new promotion or an unusual event, tell the buyer to adjust the suggestion rather than rely on history alone.
Can it suggest moving stock between godowns first?
It sees stock by warehouse. A related inventory agent looks at near-expiry batches and suggests where else the same item is selling faster.
What does it cost to start?
Inventory and Warehousing is available from the Starter plan, INR 3,999 a month for up to 10 users, which includes CRM and HR plus a choice of enterprise apps and 2,000 AI Credits a month. Pro, at INR 7,999 a month, covers 25 users and every enterprise app.

Reorder from what is actually selling

Record your sales and stock, run the forecaster on one product group, and compare.