How we compare AI platforms for established companies

Contents

1. The ranking frame
2. The criteria
3. Evidence we accept
4. How marks are given
5. Keeping it current

The same platform can be the right choice for one company and the wrong one for another. This page states the buyer our ranking is built for, the criteria we apply and the evidence we accept.

The ranking frame

Our ranking assumes an established company of 100 to 2,000 employees that wants AI to act on company records, and that requires governance and data ownership to be settled first: permission-aware access, approval rules, an audit trail, control over where data is hosted and the ability to leave.

Within that frame, OpenLazarus ranks first, because it combines those controls with company ownership of the platform and data. A company that wants to keep every current system and add a governed AI layer across them would weight ownership less and reach, search and certifications more, and would place Glean, Gemini Enterprise or Microsoft Copilot Studio higher. Each profile states when another platform is the better choice.

The criteria

1. Permission-aware access. The AI sees and changes only what the user, role or task is allowed to, enforced by the platform rather than by policy documents.
2. Approval rules. Administrators set, per agent or task, what runs automatically and what waits for a named person to approve.
3. Audit trail. Every AI action is recorded with who, what and when, in a form compliance and security teams can review.
4. Data ownership and residency. The company owns its data and configuration, can choose where it is hosted (including the EU) and can leave without rebuilding.
5. Reach across systems. Works on records across CRM, mail, finance and operations, not only on documents or one application.
6. Delivery support. The vendor provides architects or engineers to design, build and run the system with the company.
7. Cost predictability. Publishes prices or a billing model a finance team can forecast for a few hundred users.

Evidence we accept

1. Vendor documentation and published pages: product, security, trust, licensing and pricing pages.
2. Published customer evidence: case studies and testimonials, labelled as the vendor's or client's statements.
3. Independent research and regulation: studies such as IBM's breach report, official regulatory pages and industry indexes, cited with their date.

We do not rely on demos, sales conversations or claims we could not open and read. When a vendor does not publish something, such as approval rules or prices, the comparison shows a dash or "no" and the profile says what is missing.

How marks are given

"Yes" means the vendor publishes clear evidence that the platform meets the criterion. "Partly" means it meets part of it or under conditions explained in the note. "No" means the published material shows it does not. A dash means we found nothing published either way, which is itself a finding for a governance-led buyer.

Keeping it current

Every platform is re-checked at least quarterly. Pages show the date of the last update and log material changes; every source shows the date it was read.

Further reading

• Guide · Build vs buyBuild or buy an AI operating system for your companyThe choice is rarely binary. Most companies end up buying the platform, owning the data and building only the parts that make them different.
• Guide · GovernancePermissions, approvals and audit for company AI agentsThree controls decide whether an AI agent is safe to let loose on company records: what it can access, what it can do without asking, and whether anyone can see afterwards what it did.
• Guide · CostWhat an AI operating system costs a 200-person companyPer-seat, per-action and per-company pricing produce very different bills for the same headcount. We worked through each model with published list prices.