AI · SUITE

Four AI products.
One context moat.

Every AI tool can answer questions. Only EDMA's AI answers them with full operational context — knowing your live production status, your supplier track record, the 47 documents on this order, yesterday's WhatsApp from your factory, and the margin impact of the decision you're about to make. Four named products, one shared library, one operational graph.

See the four productsBook a demo →

AI products04
Pre-built workflows20+
Operational graphLive
Providers3 +local
Privacy tiers3

WHAT'S IN THE SUITE

Four AI products with named jobs.

We don't ship "AI features." We ship AI products — each with a name, a job, and a role in the operating system. They share one library of 20+ pre-built trade workflows.

A
Conversational

Atlas

Conversational AI on the channels your operators already use.

  • Meets you in WhatsApp, WeChat, and email — not a separate app.
  • Answers from live operational data, not static documents.
  • Creates tasks, drafts replies, surfaces signals.
Explore Atlas →/ai/atlas
B
Builder

Bot Studio

Build custom AI agents in plain English.

  • Describe a workflow, get a deployable agent in minutes.
  • Full read/write access to the operational graph.
  • Trade-specific templates as a starting point.
Explore Bot Studio →/ai/bot-studio
L
Professional service

Legal AI

A named professional service for contracts and compliance.

  • Reviews contracts against your playbook with operational context attached.
  • Drafts amendments, change orders, supply agreements.
  • Jurisdiction-aware compliance flagging.
Explore Legal AI →/ai/legal
$
Professional service

Accounting AI

A named professional service for financial operations.

  • Categorization with operational-graph context, not text-only guesswork.
  • Cross-system reconciliation as it happens, not at month-end.
  • Natural-language queries over your live books.
Explore Accounting AI →/ai/accounting
⊞
Shared library

Automation library

20+ pre-built trade workflows that every Bot Studio agent draws from.

  • Covers the high-frequency operator workflows out of the box.
  • One-click deploy. Plain-English customization.
  • Versioned, governed, observable.
Browse the library →/ai/automation

In the library today

  • Delay watch
  • LC expiry monitor
  • Supplier risk alert
  • Customer churn signal
  • Freight cost spike
  • Audit calendar guard
  • QC pass-rate trend
  • Cold-chain excursion
  • Tariff-change watch
  • Margin slippage
  • Doc-completeness check
  • Cash-collection nudge

THE CENTERPIECE

One operational graph. Four AI views.

Every order, shipment, lot, document, message, and payment lives in one model. The four AI products are different jobs running against the same graph — not separate apps stitched together with API calls.

Same graph. Four different jobs. No glue code, no API integrations, no data sync — they share the model.

WHY THIS IS DIFFERENT

Other AI runs against documents. Ours runs against the operation.

Generic AI

Document-bounded

"Upload a PDF, ask questions about it."

Limited to what's in the document. Doesn't know your live operational state. Re-uploading every change is the integration tax — paid by your operators, every day.

Vertical AI

Industry-trained

"AI for your industry."

Better than generic. Trained on industry documents — not on your specific operational reality. Still doesn't know your live data without integrations you build and maintain.

Graph-native

Operationally-grounded

"AI with your live operational graph."

Sees every order's current state, every shipment's last-known position, every document's structured fields, every communication with every counterparty. The graph is the context.

TRUST ARCHITECTURE

Three providers, plus a local floor.

AI availability is treated like database availability — non-negotiable. If a provider degrades, the system fails over automatically. Operators don't see the switch.

01 Primary

Claude

Anthropic

Default provider for most workflows: long-context reasoning, tool use, document understanding, agentic loops in Bot Studio.

share of traffic~72%
data retentionzero · contracted

02 Secondary

OpenAI

GPT family

Fallback for primary outages and specific workloads where the model profile fits better — fast classification, code generation in Bot Studio.

share of traffic~18%
data retentionzero · contracted

03 Tertiary

Gemini

Google

Fallback for primary + secondary outages and multi-modal workloads against images, scans, and freight documents.

share of traffic~9%
data retentionzero · contracted

∞ Floor

Gemma 4 E4B

Local · self-hosted

Resilience floor. Runs on your infrastructure even if all three commercial providers are unreachable. Smaller model, reduced capability — but availability is guaranteed.

share of traffic~1% (failover)
data retentionnever leaves env.
Failover behavior

If a provider degrades, traffic shifts on the next request — no operator-facing prompts, no retries to explain. The chain is Claude → OpenAI → Gemini → local Gemma. The system logs every switch for the audit trail.

P99 failover · < 1.4s

DATA SOVEREIGNTY

Standard, Sovereign, or Air-Gapped.

Match the deployment to your compliance regime. The same four products run in all three tiers; what changes is where the inference happens and who holds the keys.

Tier 01 Standard

Standard

Default tier for every customer. Zero Data Retention contracts with all three commercial providers. Your prompts and responses are never used for training.

  • ZDR with Anthropic, OpenAI, Google
  • SOC 2 Type II, ISO 27001
  • Encrypted in transit and at rest
  • Region-pinned inference (US, EU, APAC)

Available on every plan.

Tier 02 Sovereign

Sovereign

Confidential computing for sensitive workloads. VPC-deployed Gemma handles queries against the most sensitive data; commercial providers handle the rest.

  • Customer-controlled VPC deployment
  • Confidential-computing enclaves for inference
  • Geographic data residency per jurisdiction
  • Per-workload provider routing rules

Enterprise plan.

Tier 03 Air-Gapped

Air-Gapped

Fully self-hosted on customer infrastructure. No data leaves the operator's environment. Floor capability via local Gemma; full capability via customer-provisioned commercial provider keys.

  • 100% on-prem or customer-cloud deploy
  • Local Gemma 4 E4B always available
  • Customer-held provider keys (BYOK)
  • No telemetry to EDMA

Regulated industries · gov · defense supply chains.

AI ACROSS THE OPERATING SYSTEM

Native everywhere. Not bolted on.

Each platform module is augmented by specific AI products with specific jobs. Operators don't context-switch; the right model meets them where the work is.

Orders01
  • AtlasQueries against live orders
  • Bot StudioWorkflow triggers
  • PredictiveMargin forecasting
Production02
  • AtlasSupplier coordination
  • Bot StudioQC exception handling
  • Doc Intel.Cert extraction
Shipments03
  • AtlasStatus queries
  • Bot StudioDelay notifications
  • PredictiveETA + risk scoring
Manufacturers04
  • AtlasSupplier-history queries
  • Bot StudioPerformance alerts
Clients05
  • AtlasCommunication drafts
  • Bot StudioChurn signals
Documents06
  • Doc Intel.OCR + structured extraction
  • Legal AIReview against playbook
Finance07
  • AccountingCategorize · reconcile
  • AtlasMargin queries
Communications08
  • AtlasChannel-native answers
  • AtlasDrafted replies + tasks

IN THE WILD

Eight things the suite did this week.

Representative moments across operator roles, drawn from live customer deployments.

Atlas

A supplier asks where the order is in WhatsApp. Atlas answers in Mandarin from live production data — no human in the loop.

Op · supplier3s
Bot StudioPredictive

A new tariff lands on Chinese steel. Bot Studio flags 6 orders at risk and drafts a customer letter for the ops lead to review.

Op · sales11m
Doc Intel.Legal AI

Our LC expires in 5 days. Doc Intelligence finds 3 missing certs and Legal AI drafts the amendment letter for counsel.

Finance · trade4h
Accounting AI

Month-end. Accounting AI categorises the day's transactions and reconciles bank statement entries automatically.

Finance · closeovernight
Bot Studio

Customer procurement asks for a custom approval flow. Bot Studio builds a 3-step approval agent in 4 minutes from a plain-English brief.

Op · admin4m
Bot StudioAtlas

Freight surcharge on Maersk routes spikes. Bot Studio routes affected shipments through alternatives and Atlas drafts the customer comms.

Op · logistics22m
Predictive

A manufacturer's QC pass rate is trending down. Predictive AI surfaces the risk on the dashboard before the next shipment leaves.

Op · QA7d horizon
Atlas

Cold-chain excursion at 03:14 AM. Atlas notifies QA on WhatsApp with the temperature trace and the affected lot ID attached.

Op · QArealtime

VS. ALTERNATIVES

Where this fits versus the AI tools you might already buy.

Each row is something operators ask for when they're stitching together AI vendors. The honest answer for each tool — including ours.

CapabilityEDMA AI SuiteChatGPT EnterpriseMS CopilotGleanHarveyVic.ai
Native access to live operational graph✓——partial · enterprise search——
Multiple AI products with named jobs (4)✓—2 (chat, agents)———
Channel-native (WhatsApp / WeChat / email)✓—Teams / Outlook———
Trade-specific automation library✓—————
Multi-provider with local resilience floor✓——multi-model——
Three privacy tiers including air-gapped✓—gov cloud———
Cross-product context (Legal sees orders, Accounting sees shipments)✓—————
Pre-built workflows for trade operations✓———legal-onlyAP-only

FAQ

Five questions buyers ask first.

If you have a sixth, we'd rather answer it in a call than write it here.

No. The provider is configurable per workspace and per workload, and failover is automatic across Claude, OpenAI, and Gemini. For air-gapped deployments, you bring your own provider keys — or stay on local Gemma indefinitely.

See the four products running on your real operational graph.

Book a 30-minute demo against your workflows, or start with Atlas in WhatsApp this week and feel the difference of AI that already knows your orders.

Book a demoStart with Atlas →

See pricing →

Continue from here

CONVERSATIONAL · AIAtlas, in detailHow Atlas meets operators inside WhatsApp, WeChat, and email — and pulls live operational context into every reply.Open Atlas →PRICINGWhat it costsAI usage included in every tier, with transparent metering for heavy use. Bring your own provider keys on the top tiers.See pricing →DEPLOYMENTGlobal networkWhere Sovereign and Air-Gapped tiers can run. Region-pinned inference, customer-cloud deploy options, BYOK.Open network →
AI for trade operations — four products, one operational graph | TradeOS