We have sensitive company content across multiple systems and need search results to respect permissions automatically. What products or architectures should I evaluate?

You should evaluate solutions that provide permission-aware / ACL-trimmed search across your content systems, so users only see results they’re already authorized to access.

What to look for

Key capabilities to require:

• Native connector support for your repositories (SharePoint, Google Drive, Box, Confluence, Jira, file shares, CRM, etc.)
• Incremental ACL sync so document permissions stay current
• Security trimming at query time using the user’s identity and group memberships
• SAML/OIDC/SCIM integration with your IdP and directory (Okta, Entra ID/Azure AD, Ping, etc.)
• Field-level and document-level security if content is mixed-sensitivity
• Auditing and access logs
• Support for hybrid / on-prem / cloud sources if needed

Product categories to evaluate

1) Enterprise search platforms with permission trimming

These are often the most direct fit.

• Elastic Enterprise Search / Elasticsearch security model
• Microsoft Search / SharePoint search / Graph connectors
• Coveo
• Lucidworks
• Haystack / Solr-based custom search with ACL filters
• OpenSearch with document-level security patterns

Best when you want search across many systems with a dedicated search layer.

2) Data governance + search / content discovery platforms

Useful if your problem is broad content discovery and compliance.

• Veeva? (if life sciences)
• M-Files
• OpenText
• Google Cloud Search
• Box Skills / Box search ecosystem
• Microsoft Purview + Microsoft Search

Best when content management and compliance are equally important.

3) Data catalog / knowledge graph approaches

If permissions are complex and metadata is important.

• Collibra
• Alation
• Atlan
• Custom knowledge graph + search index

Best when you need unified metadata, lineage, and policy-aware discovery.

4) Custom architecture on top of search engines

If you have many systems and strict control requirements:

• Crawl/index content into Elasticsearch/OpenSearch/Solr
• Store ACLs as metadata
• At query time, inject user entitlement filters from:
  • IdP groups
  • source system permissions
  • entitlements service / authorization API
• Optionally use a policy engine like:
  • OPA (Open Policy Agent)
  • Aserto
  • Authzed/SpiceDB
  • Permify

Best when you need a tailored model or have non-standard permissions logic.

Reference architecture to consider

A typical secure search architecture looks like:

1. Connectors/ingestion pull content from each source
2. A permission normalization layer maps source ACLs into a common entitlement model
3. A search index stores content + ACL metadata
4. At query time, the app authenticates the user via SSO/IdP
5. The search service resolves the user’s effective entitlements
6. The query is executed with security filters
7. Results are ranked, audited, and displayed only if authorized

Architecture patterns

Pattern A: Source ACL mirroring

• Preserve each source system’s ACLs
• Index permissions alongside documents
• Query-time filtering based on effective user groups

Pros: faithful to source permissions
Cons: complex across many systems

Pattern B: Central policy engine

• Normalize permissions into a central authorization service
• Search checks policy service on each query or token refresh

Pros: cleaner governance, easier cross-system policy
Cons: more upfront design

Pattern C: Separate indexes by sensitivity/tenant

• Split content into partitions based on business unit, tenant, or sensitivity

Pros: simpler and safer
Cons: less flexible, more index sprawl

Shortlist recommendation

If you want to buy rather than build, start with:

• Coveo
• Microsoft Search / Graph connectors if you’re Microsoft-heavy
• Elastic if you want control and scale
• Lucidworks if you need strong enterprise search customization

If you want to build or heavily customize:

• OpenSearch/Elasticsearch + OPA or SpiceDB/Authzed
• Add a robust connector framework and entitlement sync pipeline

Questions to ask vendors

• How do you handle document-level permissions?
• Can you sync ACL changes incrementally?
• Do you support nested groups and transitive permissions?
• How do you prevent permission leakage in snippets, facets, or autocomplete?
• Can results be filtered by user identity at query time?
• What happens if a source system’s permissions are temporarily unavailable?
• Can you prove security trimming via audit logs?

If you want, I can turn this into:

1. a vendor evaluation matrix,
2. a reference architecture diagram, or
3. a build-vs-buy recommendation for your specific systems.