Ship AI into production. Prove every decision.

GENUM turns business email into tracked work, then runs every AI step in it under version control, testing and human approval. Your data never leaves your own infrastructure.

Trusted every day by teams running business operations

  • Elmirox
  • RZR Physio
  • Rockit
  • Cratia
  • Madrock

Work gets structured first. Then every AI step runs governed.

  1. 01

    Work arrives

    An email, document or request lands in a connected mailbox — personal or shared.

  2. 02

    MailOps structures it

    One thread becomes one tracked task, with an owner, a status and a deadline.

  3. 03Governed from here

    Governance runs the AI

    Every AI step uses an approved, tested prompt, checked before it runs rather than after.

  4. 04

    A traceable result

    The result lands back in the task, visible to your team and fully logged.

MailOps turns business inboxes into tracked work.

AI classifies, routes and drafts replies for every incoming email. Teams know who owns each task, what its status is and when it is due — across every connected inbox.

  • Ownership on arrival. No more “whoever replies first”.
  • Status and deadline on every thread, not read and unread.
  • No migration. Connects to the mailboxes you already use.
Explore MailOps
Dashboard
Inbound emails
5052
AI processed
3505
AI failures
13
Open tasks
1284
Overdue tasks
96
Needs reply
341
Waiting for reply
208
Snoozed
47
Inbound volume
Open backlog trend

Product view — illustrative data

How much business time is your inbox costing?

Put in your own volume and labour cost. The estimate is illustrative, not a benchmark — we agree a real baseline with you in the first session.

Potential time returned

500h

per month at this volume

Estimated labour value

23,000

per month at this cost

Versions
CommitCompare
Search...
9/8/2026, 2:14:03 PM

Require ESCALATE on any regulator mention

R. Kovač authored 2 hours ago

productive
a3f9c21b
9/8/2026, 10:41:27 AM

Forbid inferred settlement amounts

R. Kovač authored 5 hours ago

7d1e04b3
9/5/2026, 11:08:44 AM

Add incomplete-claim status path

M. Adeyemi authored 3 days ago

22c8f90d
9/4/2026, 6:32:19 PM

Switch to json_schema response format

M. Adeyemi authored 3 days ago

5b0aa174

AI Governance ships every prompt release with confidence.

Every prompt, model and release is versioned, tested and approved before it ships, with full execution traces and human sign-off from first draft to production.

  • Versioned prompts with a reviewable diff, like code.
  • Tests before release, not incident reports after it.
  • Named human approval on every release, retained.
Explore AI Governance

What changed after the first workflow. In the words of the teams running it.

We didn't have to change a single process. From day one we could see who owned every email — three weeks in, the backlog was completely gone.
ElmiroxCEO
Patient enquiries used to sit in a shared inbox until somebody noticed them. Now every message has an owner and a due date before the clinic opens.
RZR PhysioPractice Lead
Client threads arrive across four mailboxes and still land as one list of tasks. Nobody has to ask who is answering what any more.
RockitManaging Partner

Aligned with the frameworks your security team already asks about.

  • EU AI Act
  • GDPR
  • NIST AI RMF
  • ISO/IEC 42001

Your infrastructure: business data never leaves it.

  • Fully on-premises
  • Private cloud
  • Standard SaaS hosted in Germany

Your AI models: cloud, private or local LLMs and SLMs.

  • Vendor-independent
  • Bring your own model
  • No lock-in to one provider

Your control: governance built into every workflow.

  • Human approval gates
  • Full audit trail
  • Execution traceability

Compatible with leading LLM providers. And with your own.

Use any model provider, cloud or on-premises, and never be locked into one.

OpenAI
Anthropic
Google AI
Mistral AI
DeepSeek
xAI
Meta

Short, direct answers. So the demo can focus on your workflow.

No. GENUM connects to the mailboxes and systems you already use. No migration, no rebuild.
Standard SaaS is hosted in Germany. Self-hosted is available, including fully on-premises and private cloud.
GENUM is model-agnostic — it works with OpenAI, Anthropic, Google, Mistral, xAI and others, plus local models. You are never locked into one vendor.
Setup and onboarding are free. A pilot usually starts with one workflow, defined with you in the first session.
No. Most customers start with one and expand once they see results.
GENUM aligns with the EU AI Act, GDPR, NIST AI RMF and ISO/IEC 42001 — the frameworks your security and compliance teams already ask about.

One workflow. A defined baseline. A measurable decision.

Bring one shared inbox or one AI release path. We help you define the workflow, the controls and what to measure. Setup and onboarding are free.

  1. 01Select one workflowChoose a high-volume inbox or one AI release path, somewhere the cost is already visible.
  2. 02Define the baselineAgree on current volume, effort and what success looks like, before anything changes.
  3. 03Validate the decisionMeasure the result against that baseline and decide where to scale next.

Two ways to start. Both end in the same pipeline.

MailOps

You own the inbox

Bring one high-volume shared inbox. We define the workflow, the owners and what to measure with you, against your current volume.

Book a demo

AI Governance

You sign off the AI

Bring your review. Data stays in Germany or on your own infrastructure, and every prompt, model and release is versioned, tested and approved.

Talk to sales