D13 - Competitive landscape (informative)¶
1. Positioning map¶
Prompt/tool policy Network egress
───────────────── ──────────────
Enterprise SaaS Zenity Boundaries Formal Endpoint
Lasso Intent Deputy C1 Egress
Pillar Agentic Endpoint Forcepoint Agent Gateway
AppOmni AgentGuard Permit HTTP Egress Proxy
Local OSS AgentFence (MCP tools) Leash (egress audit proxy)
Agent Guard (fs/MCP) Fence / srt / nono / hazmat
Microsoft MCP gateway seekrit-proxy (cred broker)
DenyOut sits in local + server egress, consumer-simple UX, secret-in-URL / R-20 emphasis, brand: default deny outbound.
2. Product notes¶
| Name | Overlap | Differentiation vs DenyOut |
|---|---|---|
| C1 Egress | Deny-by-default egress, creds out of agent | Enterprise; DenyOut = OSS/local-first seatbelt |
| Formal Endpoint | Laptop+sandbox+pod egress MITM | Broad platform; DenyOut narrower & simpler |
| Permit egress proxy | Workflows, HITL, creds | AuthZ platform; DenyOut no account MVP |
| seekrit-proxy | Cred broker + allowlist | DenyOut focuses on deny/exfil; may integrate later |
| Leash | Agent egress audit/enforce | Similar; compare UX/name; avoid clone - emphasize R-20 scanners + run UX |
| Fence / srt / nono | OS sandbox + proxy | Complementary; DenyOut wraps/policy layer |
| AgentFence | MCP tool allow/deny | Complementary (tools vs raw HTTP) |
| **AgentGuard *** | Crowded name, varied scopes | Name avoided on purpose |
| Zenity / Pillar / Lasso / Forcepoint | Enterprise runtime | Different buyer; DenyOut can be control they lack on DIY agents |
3. Strategic implications¶
- Do not compete on "full AI SPM platform" first.
- Do compete on time-to-protected-run and honest R-20 story.
- Integrate with sandbox runtimes rather than reimplementing seatbelt.
- Keep policy language compatible enough to map to enterprise proxies later.
4. Name collisions checked (2026-11-13)¶
denyout.com- registered by us.- Sister TLDs appeared free at check time - consider defensive registration.
- No major product branded DenyOut found in search.