Not a smaller SailPoint.
A different way of doing IGA.

📊 2 · Your data on day one

We don't demo a sandbox with fictional employees. The POC connects your HR feed and one of your systems in the first session — role-mining proposals, risk scores and audit evidence come from your own environment the same afternoon. If the value isn't visible in your data, you shouldn't buy it.

🇪🇺 3 · Sovereignty by construction

Not a compliance slide — an architecture. The tier-3 agent runs in your VPC and resolves connector credentials locally: secrets never cross the wire to our control plane. Bring-your-own-vault points us at HashiCorp Vault, Azure Key Vault or AWS Secrets Manager instead — we store the reference, your vault keeps the value. EU-hosted, EU-owned, no US parent company. The full architecture →

💬 6 · Business-readable, wizard-first

Role mining outputs plain-language proposals — "12 people in Finance share this access" — not cluster IDs. Policies are built in a visual wizard with a preview that names people, not just counts them: who starts matching, who stops, and what that population actually gets. A policy that quietly takes access away from ninety-three people should not read as −93. Config that classically needs a consultant dialect is a guided flow an admin walks through alone. The consulting workshop is the product.

🧩 7 · One rule model — central defaults, local overrides

Who approves, who reviews, who owns, what may never combine — in classic IGA that logic is re-authored inside every workflow and drifts apart. Here every rule family — approval chains, review rules, ownership rules and SoD rules with compensating controls — is a named, reusable object with a workbench per topic: define it once as the default for its type, override only where a system or team genuinely differs, and universal fallbacks guarantee nothing ever routes to nobody.

🔁 8 · The test is what a change costs

Every IGA platform works on the day it goes live. The question nobody asks in the demo is what happens when the rules change, when the people change, and when the controlling never stops — because that is where these programmes actually die.

Our answer is one architectural choice: the platform converges, it does not react to events. Reconciliation pulls the world toward the model on a clock, so a missed message costs you a run instead of a gap. A leaver's account is disabled by the next run — no event to miss, no ticket to chase.

On top of that sit the things that keep the asking honest:

• A question that has been answered closes itself. We measured 24 of 192 open items asking something that was already settled. They now supersede themselves, with the reason on the trail.
• A decision follows its owner at the moment the owner changes — not at the next sync. And a task somebody already took over by hand stays put.
• The "object owner" role is derived, not assigned. Own something and you have it; own nothing and you do not. Nothing else grants it, and the API refuses a manual add.
• The asking is no longer a login habit. A review round announces itself, an overdue decision nudges its owner from a scheduled job, and the person who asked hears the outcome — approved, rejected with the reason, live. "Reminders are automatic" is a sweep here, not a sentence on a status page.
• An attestation can say no. "Do you still own this?" offers "no — it is X's now", and the handover moves the pending decisions at that moment. The only answer a yearly campaign used to accept was yes.

And underneath, 45 background jobs keep the model converged — each one reporting its own health, because a job that quietly stops running is itself a finding.

🔌 Connector library size

The incumbents ship connector libraries in the hundreds; we ship 14 vendor templates — from Entra ID and Active Directory to Exchange Online, Google Workspace, Workday, AFAS, TOPdesk and Nmbrs — plus six generic engines (REST, SCIM 2.0, LDAP, SQL, SFTP-CSV and SMB/NTFS): 20 production connectors in total, of which SQL, SFTP-CSV and SMB read only. For mid-market estates that generic layer covers the long tail — but if you need a certified mainframe or SAP GRC connector today, the incumbents are ahead. The flip side: because every template is built on those generic engines, a new vendor template is days of work, not a product-roadmap quarter — we build them alongside onboarding customers, at no extra cost. The fourteenth is not ours: Nmbrs was contributed as a package against our template SDK, which is the proof that the long tail does not have to run through our roadmap.

🔐 Deep PAM

We tag privileged access, route it to admin accounts, and measure JIT coverage — but we are not a password vault or session recorder. If you need full PAM, pair us with a dedicated tool; SailPoint + CyberArk is a mature combo.

📈 Analyst coverage & 20-year references

We're not in a Forrester Wave and won't be for a while, and our focus is mid-market estates, not FTSE-100 with 50k+ identities. If procurement needs a magic quadrant, that's a real constraint — we compensate with a POC on your data in a day, which no quadrant can show you.

📱 Mobile app & marketplace

Approvals and review rounds reach you by mail now — one link, per tenant and off by default, and the decision itself still happens on a page that shows you the SoD warnings first. There is no native mobile app and no third-party extension marketplace; extensibility runs through config packs and the governed API.