PRODUCT PROFILE · RANK 2 OF 6
Neo review: holding AI agent actions for approval, with an agent-level audit chain
SHORT ANSWER
Neo intercepts AI agent actions at the endpoint and allows, blocks, redirects or holds them for verified human approval, with an audit mode before enforcement, and ties every action to the human or agent that caused it. It rolls out as a new endpoint sensor. It scores 6.85 and ranks 2 of 6, and trails on Before it runs, Capability rules, Rollout, Cloud reach, Named agents and Public proof.
By The Agent Desk · Agentic Endpoint Compare · Published 2026-10-11
Desk research from public vendor pages, press releases and standards pages read on 27 September and 11 October 2026. Editorial assessment; we did not run the products.
Rank 2 of 6 among agentic endpoint vendors on our published weights
Approval holds and an agent-level audit chain
What does Neo control?
An endpoint sensor that intercepts AI agent actions and allows, blocks, redirects or holds them for verified human approval, with every action tied to the human or agent that caused it.
How does Neo reach devices?
A new lightweight endpoint sensor with no kernel module, no reboot and no OS hooks; OS support and MDM rollout are not stated.
Source: Neo platform · Read 13 Oct 2026
Which verdicts can Neo return?
- Block or denyDocumented
- Ask a personDocumented
- Alert or warnNot described
- Mask or modifyDocumented
- Simulate firstDocumented
- Remove or scale backNot described
"Allow, block, redirect or hold an action for verified human approval; audit mode before enforcement."
Source: Neo platform · Read 13 Oct 2026
How does Neo score on each criterion?
| Criterion | Weight | Score | Why |
|---|---|---|---|
| Stop before it runs | 25% | 8 / 10 | Intercepts agent actions at the endpoint and allows, blocks or holds them for approval before they run; decision latency is not stated. Source: Neo homepage · Read 13 Oct 2026 |
| Verdict range | 15% | 7 / 10 Tied highest | Documents 4 of six verdict types: Block or deny, Ask a person, Mask or modify and Simulate first. "Allow, block, redirect or hold an action for verified human approval; audit mode before enforcement." Source: Neo platform · Read 13 Oct 2026 |
| Capability rules | 15% | 7 / 10 | Policies built with a built-in LLM, with examples such as Block secret reads and Hold exports over 10 MB, over an inventory of agents, models, skills, MCP servers and extensions. Source: Neo homepage · Read 13 Oct 2026 |
| Actor trail | 10% | 8 / 10 Tied highest | Every action is tied to the human or agent that caused it, in what Neo calls an immutable audit chain. Source: Neo homepage · Read 13 Oct 2026 |
| Rollout path | 10% | 6 / 10 | A new lightweight endpoint sensor: no kernel module, no reboot, no OS hooks; OS support and MDM rollout are not stated. Source: Neo platform · Read 13 Oct 2026 |
| Cloud and workspace reach | 10% | 5 / 10 | Names SaaS and desktop AI such as Slack AI, Notion AI and Salesforce Agentforce from the endpoint; agents in cloud workloads are not described. Source: Neo platform · Read 13 Oct 2026 |
| Named agent coverage | 5% | 5 / 10 | Names 3 of the six agents we track: Claude Code, GitHub Copilot and OpenAI Codex. Source: Neo platform · Read 13 Oct 2026 |
| Public proof | 10% | 6 / 10 | Came out of stealth on 20 July 2026 with $100M from Andreessen Horowitz and Bessemer; no documentation portal or pricing. Source: Neo news: launch with $100M (20 July 2026) · Read 13 Oct 2026 |
Which agents does Neo name?
- Claude CodeNamed
- CursorNot named
- GitHub CopilotNamed
- OpenAI CodexNamed
- WindsurfNot named
- Claude DesktopNot named
Neo names 3 of the six agents we track.
Source: Neo platform · Read 13 Oct 2026
- Neo's platform page also names Gemini CLI, ChatGPT Desktop, Perplexity, Mistral Le Chat, Slack AI, Notion AI and Salesforce Agentforce; Cursor is not named.
What does it cost?
Where is Neo strongest, and where does it fall short?
Strongest
- Verdict range: 7 / 10 (Tied highest)
- Actor trail: 8 / 10 (Tied highest)
- Holds an agent action for verified human approval before it runs, with an audit mode before enforcement.
- Ties every action to the human or agent that caused it.
Falls short
- Stop before it runs: 8 / 10; highest Bay 9 / 10
- Capability rules: 7 / 10; highest Bay 9 / 10
- Rollout path: 6 / 10; highest Bay 9 / 10
- Cloud and workspace reach: 5 / 10; highest CrowdStrike Falcon Guardian 8 / 10
- Named agent coverage: 5 / 10; highest Ent 7 / 10
- Public proof: 6 / 10; highest CrowdStrike Falcon Guardian 8 / 10
- Rollout through MDM and supported operating systems are not stated.
- No documentation portal or pricing, and agents in cloud workloads are not described.
Key facts
| Fact | Detail and source |
|---|---|
| Identity | Identity: Neo Security, Inc. at neo.ai; not Pulumi Neo. Source: Neo homepage · Read 13 Oct 2026 |
| Verdicts | Allow, block, redirect or hold an action for verified human approval, with an audit mode before enforcement. Source: Neo platform · Read 13 Oct 2026 |
| Policies | Policies built with a built-in LLM, with examples such as Block secret reads and Hold exports over 10 MB, over an inventory of agents, models, skills, MCP servers and extensions; Neoverse risk scores and on-device analysis. Source: Neo homepage · Read 13 Oct 2026 |
| Team and HQ | CEO Nick Warner and co-founder Eran Shirazi, with former SentinelOne, Wiz and Palo Alto Networks staff; headquartered in Boston. Source: Neo news: launch with $100M (20 July 2026) · Read 13 Oct 2026 |
| Funding | Came out of stealth on 20 July 2026 with $100M from Andreessen Horowitz and Bessemer. Source: Neo news: launch with $100M (20 July 2026) · Read 13 Oct 2026 |
How does Neo compare head to head?
| Pair | Result |
|---|---|
| Bay matched with Neo | Bay 7.20, Neo 6.85 |
| Neo matched with Onyx Security | Tied at 6.85 |
| Neo matched with Zenity | Neo 6.85, Zenity 6.75 |
| Neo matched with Noma Security | Neo 6.85, Noma Security 6.55 |
| Lasso Security matched with Neo | Lasso Security 6.40, Neo 6.85 |
| CrowdStrike Falcon Guardian matched with Neo | CrowdStrike Falcon Guardian 6.25, Neo 6.85 |
| Harmonic Security matched with Neo | Harmonic Security 5.90, Neo 6.85 |
| Ent matched with Neo | Ent 5.80, Neo 6.85 |
| Koi (Palo Alto Networks) matched with Neo | Koi (Palo Alto Networks) 5.75, Neo 6.85 |
| Neo matched with Prompt Security (SentinelOne) | Neo 6.85, Prompt Security (SentinelOne) 5.45 |
| Glow Security matched with Neo | Glow Security 5.40, Neo 6.85 |
| Bloom Security matched with Neo | Bloom Security 4.65, Neo 6.85 |
Frequently asked questions
Can Neo stop an agent action before it runs?
Intercepts agent actions at the endpoint and allows, blocks or holds them for approval before they run; decision latency is not stated. We score this 8 out of 10.
Which verdicts does Neo document?
Neo documents 4 of six verdict types: Block or deny, Ask a person, Mask or modify and Simulate first. "Allow, block, redirect or hold an action for verified human approval; audit mode before enforcement."
How much does Neo cost?
Not published. Neo publishes no pricing.