@ATBASHai

ATBASH

ATBASH is building a pre-execution control layer for AI agents in sensitive workflows, allowing compliance teams to define red lines and block irreversible actions before they execute. The project has a named founding team (Yosef Shaftal, Or Perelman), a working demo showing verdict/block functionality, and is in selective CLI/SDK rollout. Account is extremely new (April 2026), small following (~900), and addresses a genuine emerging problem in agentic AI security. No token is planned per their bio.

𝕏 @ATBASHai ↗ t.coAIpromising

＋ Watch

AI Analysispromising

Confidence

72%

ATBASH is building a pre-execution control layer for AI agents in sensitive workflows, allowing compliance teams to define red lines and block irreversible actions before they execute.

The project has a named founding team (Yosef Shaftal, Or Perelman), a working demo showing verdict/block functionality, and is in selective CLI/SDK rollout.

Account is extremely new (April 2026), small following (~900), and addresses a genuine emerging problem in agentic AI security.

No token is planned per their bio.

Green flags: Very early stage: <900 followers, account <3 months old, addressing emerging AI agent security gap · Named founding team with working product demo (verdict system visible on site) · Clear technical differentiation: pre-execution boundary control vs post-hoc guardrails · Selective CLI/SDK rollout to real enterprise users (Risk/Compliance teams) · Substantive technical discourse in tweets (identity laundering, prompt injection patterns)

Red flags: Extremely nascent: only 2 months old, limited public traction outside scout circle · Enterprise B2B model may limit organic visibility/community growth · No public roadmap or transparency around funding/backers

Token

No · pre-launch

Chain

—

Stage

testnet