Appendix · Reference material

The detail behind
the claims.

Ana, capability depth, security, SharePoint sync, attrition-proofing, and the full nine-screen product walk-through — for prospects whose questions take us there.

← Back to the overview

Ana — the planning copilot
in every screen.

Not a sidebar chatbot. Embedded across Demand, Supply, Finance, Analytics, Signals, and BI — grounded in your real SKU codes.

What Ana does

  • Answers in 3–5 sentences, grounded in your real SKU codes.
  • Generates the Business Report — deterministic numbers plus an executive summary.
  • Generates the Analytics Report — 4–6 "what's working / why / watch" cards.
  • Discovers competitors and feeds the news view automatically.
  • Surfaces screen-aware suggested questions for every module.

What Ana will not do

  • Off-topic queries are deflected before any AI call — zero cost.
  • Never invents SKU codes, customers, or numbers.
  • Never references the underlying AI tech in conversation.
  • Per-tenant query budget enforced server-side — predictable spend.

The depth most planning tools
cut for shorter sales cycles.

These are the features that get scoped out of a 6-month implementation. We ship them on day one.

Out of the box — what you usually pay extra for

  • Batch-level inventory with expiry tracking — FIFO, FEFO, and near-expiry rationalisation built in.
  • BOM with approved alternate components — the engine routes around shortages without planner intervention.
  • True three-tier supply chain (Retail → Plant/DC → Raw Material) with daily simulation, not monthly buckets.
  • Forecasting designed for intermittent and lumpy demand (Croston, Theta, Prophet-lite) — not just steady-state retail.
  • Audit-trail-friendly architecture (versioned facts, structured upload jobs, encrypted secrets) ready for regulated environments.

Who it serves — industries already running on it

  • Regulated manufacturing — batch + expiry tracking, audit-friendly architecture, approved-alternate routing on BOM shortages.
  • FMCG / CPG — promotions calendar with elasticity, channel-level demand, AOP variance per division.
  • Electronics & industrial — long lead-time suppliers, BOM with substitutes, lumpy demand modelled by Croston.
  • Specialty retail — multi-DC with cross-product Tier 2 demand, seasonal pattern detection.

When a senior planner leaves,
the institution doesn't restart.

Planning judgement lives in the product, not in someone's head.

9–18months to train a replacement planner

By which time you've absorbed at least one stockout and one missed AOP. With Planamind, a new joiner onboards by reading the screen — not by shadowing for a year.

1

Recommendations engine

Pattern-classifies every SKU forecast (intermittent, sparse, biased, wrong-model, volatile, healthy) and ranks fixes by expected portfolio FA improvement. A-class candidates pre-validated on a 6-month holdout.

2

Ana — Safety Stock recommendations

Suggests the right safety-stock days per SKU using your service-level target per ABC class. Shows recommended vs current as a Δ; the planner adopts with one click.

3

New launch reference SKU

Picks the right "like product" by hierarchy similarity, history depth, and recency — with a plain-English narrative.

4

Business Report (auto)

Ana writes "what's working / why / watch" cards every 6 hours from your data. New joiners onboard in days, not months.

Engineered to clear the
first InfoSec review.

No on-prem footprint. No bespoke integrations. No surprises in the questionnaire.

T

Tenant isolation

  • Per-tenant subdomain
  • tenantId on every fact row
  • Admin surface fenced to root domain
  • Cross-tenant query paths impossible
A

Auth & RBAC

  • Clerk: SAML, OIDC, MFA, magic links
  • SOC 2 Type II
  • Roles: admin / planner / collab / viewer
  • Dimension-level scoping for partners
K

Secrets & data

  • AES-256-GCM at rest (per-IV)
  • TLS only in transit
  • Neon Postgres encrypted at rest
  • EU / US / APAC region pinning
I

AI usage

  • Aggregated context only — no PII
  • Off-topic deflected before the LLM call
  • Per-tenant query budget enforced
  • No model training on your data
Audit trail — every upload writes a structured UploadJob row · every forecast and supply run is project-versioned · every user action ties back to a Clerk identity.

Drop files in SharePoint.
Planamind picks them up.

For customers whose planning files already live in a Document Library — no manual re-uploads.

STEP 1

Drop file in SharePoint

Planner adds Excel to the Document Library — exactly as today.

STEP 2

Planamind reads via Graph

Microsoft Graph, Sites.Selected (per-site grant), Files.Read.All. Read-only.

STEP 3

Filename routes to processor

Master, transactions, supply, pricelist, budget, BOM — auto-classified.

STEP 4

Processed & surfaced

Errors surface as structured issues. No raw stack traces to planners.

What IT sees

  • Standard Azure AD app registration — no on-prem agent.
  • Sites.Selected (per-site grant only) + Files.Read.All. No Delegated permissions, no user impersonation.
  • Read-only by design — Planamind never writes back to SharePoint.
  • Client secrets stored AES-256-GCM encrypted at rest.

Operationally

  • Rotation is a 30-second operation in Settings.
  • 30 minutes from Azure registration to first sync.
  • On-demand or scheduled poll — your choice.

Nine screens, one product.

Real product UI captured from a live tenant — not mockups. This is what your planners see the day after your data lands.

BBI Dashboard
DDemand Planning
SSupply / Replenishment
FFinance / GP
AForecast Accuracy
XExternal Signals — Insights
MExternal Signals — Matrix
RReports
NAna — built-in AI

The next step is small

48-hour Proof of Value

NDA only. No procurement, no MSA, no SOW. Send your last 24 months of data tonight — we'll have a side-by-side readout on your A-class SKUs in two business days.

rishi@anamind.com ← Back to overview