HEAD-TO-HEAD · FACTS ONLY
Bloom vs Neo: AI endpoint control facts side by side
Short answer
By The Device Desk · Endpoint AI Finder · Published 2026-10-11
Desk research from public vendor pages read on 27 September and 11 October 2026. Profiles follow our published method; we did not run the products.
Vendor pages read September and October 2026 · updated 2026-10-13
Where do their controls sit?
- Filled or hatched marker: Bloom
- Dashed teal ring: Neo, described
- DEV Filled marker: described on the product pages we read
- DEV Hatched marker: partly described (another component, a narrower scope, or method not detailed)
- DEV Dashed marker: not in our sources (it may exist; we did not find it)
What are the facts, side by side?
| Fact | Bloom | Neo |
|---|---|---|
| Owner | None | None |
| Positioning | An endpoint security company built around "Securing the AI-Native Endpoint": discovery of the software, extensions and AI tools running on devices, with policy enforcement across package and extension marketplaces. Offices in Wilmington, Delaware and Tel Aviv. | Agentic Software Control from Neo Security, Inc. (neo.ai, not Pulumi Neo): a live inventory of agents, models, skills, MCP servers and extensions on every endpoint, and runtime control that intercepts agent actions and allows, blocks or holds them for approval. Launched 20 July 2026 with $100M from Andreessen Horowitz and Bessemer; HQ Boston. |
| How it reaches devices | Deployment method not published on the pages we read (re-checked 11 October 2026). Source: Bloom Security homepage · Read 27 Sep 2026 | A lightweight endpoint sensor with no kernel module, no reboot and no OS hooks; analysis stays on the device. OS support is not listed on the pages we read; SACR (secondary) says it needs no MDM. Source: Neo: platform · Read 13 Oct 2026 |
| On-device controlDiffers | Partly describedDiscovers software, extensions and AI on the endpoint and automatically scales back overpermissioned MCP servers; how it reaches the device is not published. Source: Bloom Security homepage · Read 27 Sep 2026 | DescribedA live inventory of agents, models, skills, MCP servers and extensions on every endpoint, with runtime control that intercepts agent actions. Source: Neo homepage · Read 13 Oct 2026 |
| Browser extensionDiffers | Not in our sourcesNot described on the pages we read. | Partly described(secondary: SACR) user-mode endpoint and browser security. Source: SACR (secondary): CISO guide to endpoint control and prevention, July 2026 · Read 13 Oct 2026 · secondary source |
| Network or SSE proxy | Not in our sourcesNot described on the pages we read. | Not in our sourcesNot described on the pages we read. |
| AI or MCP gateway | Not in our sourcesNot described on the pages we read. | Not in our sourcesNot described on the pages we read. |
| Names the coding agents it coversDiffers | Not in our sources | MatchesNames Claude Code, Gemini CLI, GitHub Copilot and OpenAI Codex; Cursor is not named. Source: Neo: platform · Read 13 Oct 2026 |
| Decides on a tool call before it runsDiffers | PartlyAutomatically scales back overpermissioned MCP servers; decisions on individual calls are not described. Source: Bloom Security homepage · Read 27 Sep 2026 | MatchesIntercepts agent actions at the endpoint and allows, blocks or holds them for approval. Source: Neo homepage · Read 13 Oct 2026 |
| Controls what goes into browser AI | Not in our sources | Not in our sources |
| Rolls out through EDR or device management you runDiffers | Not in our sources | PartlyIts own lightweight sensor rather than a rollout through existing EDR or MDM. Source: Neo: platform · Read 13 Oct 2026 |
| Can warn or ask instead of only blockingDiffers | Not in our sources | MatchesHold an action for verified human approval. Source: Neo: platform · Read 13 Oct 2026 |
| Can block an agent, tool or actionDiffers | PartlyEnforces policies across npm, Chrome Web Store and Open VSX. Source: Bloom Security homepage · Read 27 Sep 2026 | MatchesBlock is one of its runtime actions. Source: Neo: platform · Read 13 Oct 2026 |
| Has a watch-only or simulation modeDiffers | Not in our sources | MatchesAudit mode before enforcement. Source: Neo: platform · Read 13 Oct 2026 |
| Records who or what took each actionDiffers | Not in our sources | MatchesEvery action tied to the human or agent that caused it, in an immutable audit chain. Source: Neo homepage · Read 13 Oct 2026 |
| Also covers SaaS, cloud or homegrown AIDiffers | Not in our sources | PartlyNames SaaS and desktop AI such as Slack AI, Notion AI, Salesforce Agentforce and ChatGPT Desktop. Source: Neo: platform · Read 13 Oct 2026 |
| Has a public price or tiers (vendor site or marketplace) | Not in our sources | Not in our sources |
| Built into | None | None |
| Works with | None in our sources | None in our sources |
| Price | Not published. Source: Bloom Security homepage · Read 27 Sep 2026 | Not published (demo request). Source: Neo homepage · Read 13 Oct 2026 |
Where do they differ?
- On-device control: Bloom partly describes it; Neo describes it.
- Browser extension: Bloom is not in our sources; Neo partly describes it.
- Names the coding agents it covers: Bloom is not in our sources; Neo matches.
- Decides on a tool call before it runs: Bloom partly matches; Neo matches.
- Rolls out through EDR or device management you run: Bloom is not in our sources; Neo partly matches.
- Can warn or ask instead of only blocking: Bloom is not in our sources; Neo matches.
- Can block an agent, tool or action: Bloom partly matches; Neo matches.
- Has a watch-only or simulation mode: Bloom is not in our sources; Neo matches.
- Records who or what took each action: Bloom is not in our sources; Neo matches.
- Also covers SaaS, cloud or homegrown AI: Bloom is not in our sources; Neo partly matches.
Choose Bloom if
- Your first worry is unvetted extensions and packages arriving on developer machines
- You want an inventory of everything on the endpoint before you write AI-specific rules
Choose Neo if
- You want to hold agent actions for human approval on the device
- You want an audit mode before enforcing rules
Which limits should you check?
Bloom
- Deployment method and pricing are not published.
- Decisions on individual tool calls and an activity record are not described on the pages we read.
- Limited company information on its own pages: they do not name investors; funding is known from press coverage (Ctech, 30 July 2026).
Neo
- Uses its own sensor rather than the EDR or MDM you run, and OS support is not listed on the pages we read.
- No price is published.