Alma
Self-hosted, vendor-neutral AI agent governance with passive discovery, policy enforcement, and cryptographic audit.
Alma is one of the few enterprise AI governance platforms you can run entirely on your own infrastructure, with real access mapping and cryptographic audit. The Community edition is genuinely free with no seat limits, but commercial pricing is still unannounced. If you're a DevOps or security team comfortable self-hosting, it's worth serious evaluation despite the early stage.
Verified 2d ago · liveness 79/100 · cite: rightaichoice.com/tools/alma
- Platform and DevOps teams needing to inventory and govern AI agents
- Security teams requiring audit trails and least-privilege enforcement for AI
- CISOs looking for vendor-neutral, self-hosted AI governance
- Organizations with strict data residency or air-gap requirements
- Users seeking a fully managed cloud solution (cloud option still on roadmap)
- Teams wanting a plug-and-play no-code AI governance tool (requires self-hosting and configuration)
- Organizations that cannot adopt AGPL-3.0 (unless they buy a commercial license)
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Alma if you need a fully managed cloud solution, want a plug-and-play no-code tool, cannot adopt AGPL-3.0, or require commercial support with published pricing and SLAs before production use.
Business edition pricing is not published yet, so you can't budget for the commercial licence or the four add-ons (Regulated Operations, AI Runtime Security, Compliance Packs, Identity & Scale) until they are announced.
Alma's Community edition is $0 with unlimited users, making it attractive for teams that can self-host and adopt AGPL-3.0. Compared to cloud governance tools like Lakera or Robust Intelligence, which charge per-seat or per-model, Alma's self-hosted model can be cheaper at scale but requires more ops effort. Business pricing is unannounced, so it's hard to compare directly; expect it to be competitive with other enterprise governance platforms.
In short
Alma — Self-hosted, vendor-neutral AI agent governance with passive discovery, policy enforcement, and cryptographic audit. Best for Platform and DevOps teams needing to inventory and govern AI agents, Security teams requiring audit trails and least-privilege enforcement for AI, CISOs looking for vendor-neutral, self-hosted AI governance. Free to use.
What's new in Alma
Checked 2 days agoAcross the latest 4 updates: 2 feature updates, 1 launch and 1 pricing change.
Add-on modules introduced
Introduced four optional add-ons (Regulated Operations, AI Runtime Security, Compliance Packs, Identity & Scale) each priced by the job, requiring Business edition.
Support for Codex and Grok Build
Expanded vendor-neutral agent support beyond Claude Code to include OpenAI Codex and Grok Build.
Evals & sandbox module
Added an Evals & sandbox module to the product tour, enhancing testing and validation of agents.
Product rebrand to Olivares AI
Rebranded from Alma to Olivares AI with a new console design and expanded feature set.
What people actually say about Alma — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
69 mentions across 4 sources (Hacker News, Product Hunt, App Store, Lemmy) · researched Jul 3, 2026.
- +Deny-closed default with Cedar-based RBAC/ABAC is security-best-practice approach.
- +Passive discovery of agents and MCP servers without proxies or agents.
- +Tamper-evident audit ledger with cryptographic chaining ensures compliance readiness.
- +Emergency kill switch for instant session revocation adds critical safety.
- +FinOps module for per-agent budgets enables cost allocation and control.
- −No community feedback available to validate actual usability or stability.
- −Feature claims are extensive but unverified in real-world deployments.
- −AGPL license with 3-user cap may complicate evaluation and adoption.
- −Documentation and onboarding experience cannot be assessed from data.
- −Self-hosted setup may require significant DevOps effort for non-experts.
- • Self-hosting infrastructure and maintenance costs not included.
- • Commercial license pricing undisclosed; may require sales contact.
Viability Score
How well maintained and how widely used is Alma? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: August 2026
How we score →Key Features
- Passive agent discovery without proxies or agents
- Real-time read/write access map (permitted vs observed)
- Cedar-based RBAC/ABAC policy engine with deny-closed defaults
- Append-only audit ledger with Ed25519 cryptographic chaining
- Emergency kill switch for agent or model session revocation
- FinOps with per-agent and per-model budget enforcement
- Drift detection with confidence levels
- Claude Code integration via OpenTelemetry gen_ai telemetry
- MCP server governance with OAuth 2.1
- Policy Enforcement Point (PEP) for Claude Code hooks
- Workload identity federation with short-lived JWT tokens
- SAML 2.0 / OIDC identity federation
- Compliance evidence module (SOC 2, ISO 27001, NIST) with OSCAL export
- SLSA Build Level 3 provenance attestation on releases
- mTLS between all internal components
About Alma
Alma, from Olivares AI, is an open-core, self-hosted platform for governing AI agents across your enterprise. It is built around Claude but supports any AI model, including OpenAI, Gemini, Llama, and Ollama/vLLM, making it a vendor-neutral control plane. You deploy a single container or binary on your own infrastructure, with no account, no cloud dependency, and no calls home, so it runs air-gapped and keeps governance data under your control. The platform passively inventories every agent, session, model, and MCP tool without requiring proxies or agents, then builds a real-time read/write access map showing what each agent is permitted to touch versus what it actually reads and writes. This gives security and platform teams a continuously updated view of AI activity across their estate, including drift detection that surfaces least-privilege violations the moment they appear. Alma's policy engine uses Cedar-based RBAC/ABAC with deny-closed defaults, letting you write policy as code and enforce it before and after an action—blocking violations, not just logging them. The append-only audit ledger uses cryptographic chaining (Ed25519) to make every access tamper-evident, with export options for auditors. The built-in FinOps module attributes every dollar of spend to the specific agent, model, and team that caused it, with budgets and trends your finance team can act on. Compliance evidence is mapped to frameworks like SOC 2, ISO 42001, NIST AI RMF, and the EU AI Act, with OSCAL export for auditors. Alma integrates deeply with Claude Code via OpenTelemetry gen_ai telemetry, and includes a Policy Enforcement Point for Claude Code hooks (gating, context, observe). It also governs MCP servers with OAuth 2.1 flows. Security controls include mTLS between components, a layered collector with minimal privilege, and SLSA Build Level 3 provenance attestation on releases. The project is open-core under AGPL-3.0; the Community edition is the full product, free, with unlimited users.
Behind the Verdict
Alma stands out in the AI governance space by being fully self-hosted and vendor-neutral. Unlike many governance tools that are cloud-only or tightly coupled to a single model provider, Alma runs on your own infrastructure and supports any model—Claude, OpenAI, Gemini, Llama, or local ones like Ollama/vLLM. This is a huge differentiator for security-conscious enterprises that cannot send data to third-party clouds. The passive discovery is a genuine innovation: it inventories agents without needing to install proxies or agents, which reduces deployment friction and blind spots. The read/write access map is the core value—it gives you a living picture of what every agent can touch versus what it actually touches, making least-privilege violations visible immediately. Policy enforcement is not just logging; it blocks violations with Cedar-based deny-closed rules, and the audit ledger is cryptographically chained, so you can prove tamper-evidence to auditors. The FinOps module attributes spend to agents, models, and teams, which is rare in governance tools and valuable for chargeback and budget control. The Community edition being the full product with unlimited users is refreshing—most open-core tools cripple the free tier, but here you get everything. The main weakness is maturity: the platform is pre-1.0 with no public release yet, so it's risky for production use if you need stability. Self-hosting requires infrastructure management, which may not suit small teams. Commercial pricing is unpublished, so budgeting is hard. The AGPL-3.0 license is a concern for some organizations, though a commercial license is planned. There's no managed cloud option yet, so you must run it yourself. Overall, Alma is a promising choice for DevOps, security, and platform teams that need on-prem governance and are willing to be early adopters. It's less suitable for teams wanting a plug-and-play SaaS or those requiring immediate commercial support with SLAs.
Researching Alma? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Alma actually fits — and what changes day-one when you adopt it.
Security engineer deploys Alma on a single VM to inventory all AI agents running across the company.
Outcome: Within an hour, they have a live access map showing every agent, its model, and the resources it reads/writes, plus an alert on any unexpected write to a production database.
DevOps lead wants to enforce least-privilege on Claude Code sessions without breaking workflows.
Outcome: They configure Alma's PEP hook to deny write access to production PostgreSQL for a specific agent, and within minutes see the block in the audit ledger.
CISO needs to demonstrate AI governance to auditors but can't use cloud services.
Outcome: Alma runs air-gapped, generates OSCAL exports for SOC 2 and ISO 42001, and the hash-chained audit ledger provides tamper-evident evidence.
Use Cases
- Discover all AI agents running on your infrastructure without deploying any extra agents.
- Map every agent's read/write access to databases, object stores, and APIs in real time.
- Enforce least-privilege policies with Cedar-based rules that block or alert on violations.
- Audit agent behavior with a tamper-evident, hash-chained ledger for compliance evidence.
- Manage AI spend with per-agent and per-model budgets and cost attribution.
- Govern Claude Code sessions and MCP server tool calls with deny-closed enforcement.
- Respond to drift—unexpected agent permissions—with classified findings and automated denial.
- Run compliance drafts for SOC 2 or ISO 42001 with OSCAL export for auditors.
Models Under the Hood
as of 2026-08-18
Limitations
- The platform is in active pre-release development; no public release has shipped yet.
- Self-hosting requires managing your own infrastructure, and the product is designed to run air-gapped.
- The free Community edition is under AGPL-3.0, which may not be acceptable for all organizations.
as of 2026-08-21
Verification history
We have re-verified Alma 5 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Alma tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Community
$0/mo
Ideal for
Solo developers or small teams who can adopt AGPL-3.0 and want a self-hosted, full-featured AI governance platform for free, without seat limits.
What this tier adds
Free entry point: the complete AGPL-3.0 platform with unlimited users, one IdP, and community support—this is the starting tier.
Business
Price at launch
Ideal for
Organizations that cannot adopt AGPL-3.0 and need a commercial licence, maintained signed releases, and optional add-ons for regulated or large-scale deployments.
What this tier adds
Adds a commercial licence (AGPL exemption for runtime), a rolling signed release channel, and access to four optional add-ons; price at launch, per deployment, not per user.
Where the pricing makes sense
The company stage and team size where Alma's pricing actually pencils out — and where peers do it cheaper.
Alma's Community edition is $0 with unlimited users, making it attractive for teams that can self-host and adopt AGPL-3.0. Compared to cloud governance tools like Lakera or Robust Intelligence, which charge per-seat or per-model, Alma's self-hosted model can be cheaper at scale but requires more ops effort. Business pricing is unannounced, so it's hard to compare directly; expect it to be competitive with other enterprise governance platforms.
Setup time & first value
How long it actually takes to get something useful out of Alma — broken out by persona, not the marketing-page minute.
Setup typically takes under 5 minutes to get the access graph populated using the demo estate. For a single binary install, you can have Alma running in minutes; connecting real sources like PostgreSQL or Claude Code takes additional time to configure, roughly 30-60 minutes for a production deployment.
Switching to or from Alma
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From homegrown scripts: Alma's connectors for pgaudit and s3cloudtrail let you replace custom logging with structured access-map edges.
- ↗To a cloud governance tool: Export the audit ledger (CSV/JSON) to retain compliance evidence before decommissioning.
Integrations
Resources & Guides
- Documentationolivares.ai
Docs · Alma
Full product docs from olivares.ai
- Documentationolivares.ai
Docs · Alma
Full product docs from olivares.ai
- Documentationolivares.ai
Docs · Alma
Full product docs from olivares.ai
- Documentationolivares.ai
Docs · Alma
Full product docs from olivares.ai
- Documentationolivares.ai
Docs · Alma
Full product docs from olivares.ai
- Documentationolivares.ai
Docs · Alma
Full product docs from olivares.ai
Tutorials & Learning
![alma [tab] - Antonio Rey tremolo guitar tutorial](https://img.youtube.com/vi/laHff1WapaQ/mqdefault.jpg)
Featured Head-to-Head Comparisons
Alma vs Temporal Ai
Choose Temporal if your core pain point is building reliable, failure-tolerant AI agents and workflows that survive crashes and retries. Choose Alma if you already have AI agents running and need visibility, governance, least-privilege enforcement, and an audit trail—Temporal does not provide governance controls. They are complementary: Temporal orchestrates; Alma governs.
Alma vs Audioeye
Alma and AudioEye serve completely different markets—Alma is an open-core self-hosted platform for governing AI agent security and compliance, while AudioEye is a paid SaaS for web accessibility compliance. Your choice depends solely on whether you need AI governance or accessibility remediation; there's no overlap.
Alma vs Push Security
Choose Push Security if your primary threat is browser-based attacks (AiTM, ClickFix, session hijacking) and you need to control AI tool data leakage across any browser without migration. Choose Alma if you're a DevOps or security team governing AI agents (Claude Code, MCP servers) in a self-hosted environment and require least-privilege enforcement with an immutable audit trail. They are complementary: Push protects the user endpoint; Alma protects the agent infrastructure.
Popular in AI Governance & Guardrails
Mindgard
Automated AI red teaming & security platform for continuous agent and system protection
Poolside AI
Open-weight agentic coding models for regulated enterprises needing auditable, on-prem AI
Olas Network
Co-own and monetize autonomous AI agents on-chain with Olas.
Frequently Asked Questions
Used Alma? Help shape our editorial sentiment research.