Margince

Before you decide

Everything you'd want to know before trusting a new AI platform.

Margince is a new product entering a crowded market with strong claims. These are the questions worth asking, answered without a sales deck.

The product

The coreWhat does Margince actually do, in one sentence?

Margince is an AI-native CRM where the AI fills your pipeline overnight, and every action it takes waits for a human to approve it before anything irreversible happens.

How it works

The AI drafts. You approve. Records build from your email and calendar. Nothing is typed by hand. The morning brief shows you what moved overnight and what needs a decision today.

The differenceHow is this different from HubSpot with AI, or Salesforce Einstein?

Those are AI features bolted onto a CRM that still expects humans to type the records in - the AI works from the same stale data. Margince starts from the other end: the AI maintains the records, every field shows its source, and the vendor's cloud and model are replaced by your boundary and your model.

The short version

Bolt-on AI inherits the data problem. Margince fixes the data problem first, then the AI has something true to work with.

Honest answerIs this a finished product or a pre-launch idea?

Pre-launch, and we say so everywhere. Design-partner pilots run alongside their current CRM through the first half of 2026, and general availability comes in autumn 2026.

Why join the waitlist now

Preview access opens from the waitlist, in order - and the conversations that shape compliance, Betriebsrat and migration requirements start there too.

The architectureWhat does 'AI-native' mean versus an AI add-on?

AI-native means the system is designed around the AI doing the work: records fill from approved sources, every AI value carries its evidence or is left blank, and every outward action passes an approval gate. An add-on drops a chat box next to a database that still expects manual entry.

The horizonIs Margince only a CRM?

The CRM comes first - it is the first place AI works on business data, and where sovereignty is tested first. From there the platform grows along the same control model: an AI governance suite (EU AI Act monitoring, data-residency management, a Betriebsrat dashboard), then agents and a partner channel for European distribution.

Plainly

A direction, not a promise. What is being built today is the sovereign CRM - everything beyond it is roadmap, and we label it that way.

Data & sovereignty

Your boundaryWhere does our data live, and who has access?

You decide: on-premise in your data center, in a private cloud, or with a European host. Access is role-based, every read is logged, and in the self-hosted variants nothing has to leave your boundary.

The realityIsn't an EU data center enough?

Data residency is not the same as control. A US-owned provider can be compelled to hand over data under the CLOUD Act, wherever it is stored. Two EU-US transfer frameworks have already been struck down - Safe Harbor in 2015, Privacy Shield in 2020 - and the current one is under fresh challenge.

The margince answer

The boundary sits in the architecture, not the contract: your instance, your model, your keys.

Model controlWhich AI model does it use?

Yours. Margince does not force a vendor model or meter your AI usage. Use the model your security and procurement process already cleared, on your own contract, in your own environment where required - and change model strategy later without touching the CRM.

Compliance & Betriebsrat

By designHow are GDPR and the EU AI Act handled?

Data residency, role-based access, human-in-the-loop approval and the audit log are part of the architecture, not the fine print - built so Legal, Security and works-council review can sign off before go-live.

Betriebsrat firstHow is employee data handled?

The AI reads only the mailboxes and calendars you connect; private items stay out. What employee data is in scope is defined before go-live, together with your works council. Every read is logged with the reason it happened.

Built for DACH

An AI reading email is a Betriebsrat topic before it is an IT topic. Margince is built for that order - consent and logging are the default, not a setting.

EvidenceWhat does the audit log actually record?

Which data the AI read and under which permission, what it wrote or suggested, who approved it before anything happened, which version or setting changed its behavior, and how to get your data out if you leave. When the auditor asks, you answer with records.

Migration & setup

NoDo we have to replace our current CRM on day one?

Margince starts alongside your existing CRM - on top of HubSpot first, with Salesforce and Dynamics adapters following - and proves its worth before you migrate. Full replacement is an option later, never the entry condition.

Not SaaSWho sets it up and runs it?

You get the software, not just a login. An implementation partner sets it up, adapts it and runs the rollout; it runs in your environment or at your EU host. Instance, data and code are yours.

Pricing & licensing

FlatWhat does it cost?

Unlimited read seats are free. Acting seats - create, send, advance, approve - are €25 per month above the free 10. No charge per contact, no meter on AI usage: the AI runs on your own model and your own contract. Enterprise (100+ seats): talk to us.

Source-availableWhat license is the source under?

Margince ships with its full source under BUSL-1.1 - review it, adapt it, self-host it. Each release becomes Apache 2.0 24 months after it ships. The software stays yours even if everything else changes.

For partners

Your business stays yoursWhat does the partner model look like?

100% of your services revenue stays with you - Gradion builds and maintains the secure core and never competes for your consulting, hosting or implementation work. Gradion invoices the license directly to the customer, you invoice everything else. You receive 10% of the paid license revenue for 36 months, plus 15% on service contracts and 20% on projects you refer to Gradion. Your registration protects the referral for 90 days. No tiers, no certification, no duties.

Why now

Not 5,000 partners - a few selected launch partners, DACH first. AI governance is a service category nobody else owns yet.

DirectWhat happens after we reach out?

A direct conversation with the team building the product - no form marathon, no sales deck, no SDR sequence.

Not answered here?

Talk to the team directly.

Margince is pre-launch. The team building it is directly reachable, not a support queue, not a sales sequence.

No SDR spam. Direct reply within 48 hours.