Company profile
Pretxt
Runtime "AI supervision" layer that watches AI agents while they work, checks each action against a company's own processes and policies, and steers drifting agents back on track (or escalates to a human) instead of…
- Status
- PrivateResearch map: Seed / private · Aspen, CO + Dublin · $5M seed led by Deciens Capital (Howdy Partners, Service Provider Capital, NSV Wolf Capital) · out of stealth Nov 17, 2026 · early access
- Valuation
- Undisclosed
- Latest funding
- Seed · $5M Nov 17, 2026SiliconANGLE Source
- HQ
- Aspen, CO + Dublin / US-Ireland (Startup Radar)
- Last updated
Categories
Funding & valuation
| Date | Round | Amount | Valuation | Source |
|---|---|---|---|---|
| Nov 17, 2026 | Seed | $5M | Undisclosed | SiliconANGLEDeciens Capital led; Howdy Partners, Service Provider Capital and NSV Wolf Capital participated. |
Research map notes
- Research map · AI Agent Security · Undisclosed $5M seed led by Deciens Capital with Howdy Partners, Service Provider Capital and NSV Wolf Capital (Nov 17, 2026, Business Wire / SecurityWeek). Valuation not disclosed. Source (recorded Nov 19, 2026)
Latest signals
- Startup Radar · AI Security · snapshot Nov 20, 2026 17 Nov 2026: emerged from stealth with a $5M seed from Deciens Capital (lead per SiliconANGLE), Howdy Partners, Service Provider Capital and NSV Wolf Capital; product in early access. Founded by AppOmni founder Brian Soby (CEO) and Aaron Costello (CTO); says it cut policy-violating agent actions on Zapier's AutomationBench from nearly 1 in 5 to zero and runs ~60x faster than LLM-as-judge tools (company claim). Valuation Undisclosed (Business Wire/SecurityWeek/SiliconANGLE). Source
Competitors
Same-category peers from CyberMerge Research maps, Startup Radar and What's The Use tags.
- Pluto SecurityPrivateEndpoint AI-workspace security: visibility and real-time control over the AI tools employees use, the…
- RecoPrivateSaaS and AI-agent context graph plus browser/network discovery and prompt/tool-call controls — securing…
- IslandPrivateEnterprise browser plus AI Protect: prompts, tool calls, Island MCP Gateway — agentic control plane for…
- EndeemPrivateGives every AI agent its own identity instead of borrowing an employee's login: per-request short-lived…
- Asymptote LabsPrivateRuntime security for the AI agents employees already run: Beacon, an MIT-licensed telemetry layer, records…
- BayPrivateRuntime control for AI agents: inventories the assistants, coding agents, custom agents and MCP servers on…
- Eve SecurityPrivateAgentic AI runtime security (EveGuard): Agent-in-the-Loop observability, intent/data policy enforcement, and…
- Harmonic SecurityPrivateWorkforce AI security: classifies every AI task, discovers why it is used, and controls it in real time…
- HiddenLayerPrivateModel and agent discovery, AI supply-chain scanning, attack simulation, runtime firewall, AIBOM.
- KontextPrivateRuntime authorization for AI agents: sits between the agent and target systems, checks each requested action…
- MetanoPrivateAI agent security platform: finds every AI agent, MCP server and skill in use (approved or not) across…
- OuterlimitPrivateDecentralized security and authorization layer for agentic AI: extends Zero Trust to the agent action layer…
On the desk
- Startup Radar · Nov 17, 2026 Pretxt came out of stealth with a $5M seed to supervise AI agents while they work and pull them back when they drift off the rules. SiliconANGLE
Figures appear only as recorded in CyberMerge data, with source and date. “Undisclosed” means no published figure in our data. Editorial, not investment advice. Last updated . Spot an error? Tell the desk.
