Demand Forecasting (Preview)

Forecast SKU demand with ranges, accuracy, and inventory handoff.

This connector is available in the following products and regions:

Service Class Regions
Copilot Studio Premium All Power Automate regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Logic Apps Standard All Logic Apps regions except the following:
     -   Azure Government regions
     -   Azure China regions
     -   US Department of Defense (DoD)
Power Apps Premium All Power Apps regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Power Automate Premium All Power Automate regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Contact
Name Saifs AI
URL https://saifs.ai/
Email taher@saifytech.com
Connector Metadata
Publisher Saifs AI
Website https://saifs.ai/
Privacy Policy https://saifs.ai/privacy_page
Categories AI;Data

Demand Forecasting - Saifs AI turns a sales or volume history into what each product (or region / channel) will sell next, with a range, not a bare number. Connect sales data through Microsoft Copilot Studio or Microsoft 365 Copilot. The MCP grids the history, picks a classical model (Naive / SES / Holt / Holt-Winters / Croston-SBA) by rolling-origin backtest, and returns ranked decisions, per-series forecasts with prediction intervals, accuracy, regime-change alerts, a data-quality report, and an inventory handoff (demand over lead time plus forecast-error sigma).

Who should use it

  • Retail buyers and merchandisers who need a per-SKU next-quarter outlook instead of last-year-plus-10%
  • Distributors and wholesalers who split forecasts by warehouse, godown, or region
  • Spare-parts and industrial planners whose demand is lumpy and should use Croston / SBA
  • Manufacturers and production planners who need a unit outlook by SKU
  • E-commerce and D2C operations planning the next quarter and watching for demand shifts
  • Pharmacy and perishable planners who need a demand range, not a false point estimate
  • Support and workforce planners forecasting ticket or contact volume
  • SMB owners using Tally or Excel exports without cleaning or renaming columns first
  • Inventory and purchasing managers who need demand over each product's supplier lead time
  • Copilot Studio and Microsoft 365 Copilot makers adding explained demand math to agents

How it helps

  • Next-period forecast per product (and optionally per region or channel) with a prediction interval (default 95%)
  • Automatic model choice by time-ordered backtest; intermittent series are routed to Croston / SBA
  • Seasonal models only when there is enough history; thin series are refused or given a wide mean-only fallback with a stated reason
  • Messy real-world files are accepted as-is (Voucher Date, Billed Quantity, Stock Item, Godown; values such as 12 pcs, n/a, 15-Jan-2024)
  • Inventory handoff uses each SKU's own lead time when provided; otherwise a default wait (14 days)
  • Spikes are flagged and capped; unreadable cells and holes in a dense file are missing, not sold-zero

Tools

  • demand_forecaster: about 2,500 rows or 200 products or fewer; send rows and get the full forecast result
  • demand_forecaster_get_engine: larger files; returns a ready-to-run numpy script so heavy data can stay on the user machine

Authentication OAuth 2.0. Users complete the OAuth consent flow when connecting the server. For Microsoft certification packages, OAuth client settings are referenced through Azure Key Vault (ClientId, ClientSecret, AuthorizationUrl, TokenUrl; RefreshUrl and Scopes when applicable).

Getting started

  1. Add Demand Forecasting - Saifs AI in Copilot Studio or Microsoft 365 Copilot.
  2. Sign in with OAuth 2.0 when prompted.
  3. Provide a sales or volume history with at least date and quantity; include SKU, region, channel, and lead times for stronger results. Copy rows exactly; do not rename or pre-clean.
  4. Ask what you will sell next, per SKU, by warehouse, or over a given lead time.

Results are decision support, not an ERP or S&OP replacement. Always use the forecast range, not only the midpoint. Thin history is held back with a stated reason rather than inventing precision. Published by Saifs AI.

Actions

Demand Forecasting

Forecast SKU demand with ranges, accuracy, and inventory handoff.

Demand Forecasting

Operation ID:
InvokeMCP

Forecast SKU demand with ranges, accuracy, and inventory handoff.