Behavioral Security for AI Agents On-Chain
SentinelCRE monitors AI agents the way Hexagate monitors smart contracts. Per-agent behavioral baselines detect compromise before damage is done.
The Opportunity
$168.6M lost in Q1 2026 across 34 DeFi hacks. AI agents are increasingly autonomous — and nobody monitors their behavior. When an agent is compromised, it acts with the same credentials and permissions as a legitimate one. Identity systems can’t tell the difference.
SentinelCRE is the first behavioral monitoring layer for AI agents on-chain. Hexagate monitors contracts. Hypernative monitors protocols. We monitor agents — per-agent behavioral baselines that catch compromise before damage is done.
ERC-8004 integration — we read agent identity and write behavioral incidents to the Reputation Registry. SentinelCRE is infrastructure, not just a product. Every agent interaction enriches the cross-protocol intelligence network.
Three Tiers. One Behavioral Engine.
Passive monitoring to active prevention. The data moat deepens with every agent onboarded.
Monitor
$2K/moPassive behavioral monitoring, per-agent baselines, anomaly alerts, dashboard access. Know when something is wrong.
Protect
$5K/moActive prevention via SentinelGuardian, throttle mode, policy enforcement, ERC-8004 reputation writes. Stop it before damage is done.
Enterprise
$10–20K/moDedicated behavioral analysis, custom red team engagements, cross-protocol intelligence feeds, SLA-backed response times.
Red Team Assessment — $15–30K one-time. AI-powered adversarial testing of your agent fleet. Find the gaps before attackers do.
Cryptographic Proof of Protection
The first agent security vendor whose metrics are independently verifiable on-chain.
Every blocked attack is logged as an immutable on-chain event with the attempted dollar value, the attack vector, and the defense layer that caught it.
This creates a cryptographic ledger of attacks that anyone can audit — investors, protocol customers, regulators. No trust required.
Hypernative: "We detected 500 risks" — trust us
Forta: "Our bots flagged 500 threats" — trust us
SentinelCRE: Every block is a transaction hash. Verify it yourself.
// Anyone with a block explorer can verify this event
Why Now
$168.6M lost · Q1 2026
34 DeFi hacks in a single quarter. AI agents are executing with increasing autonomy — and nobody is monitoring whether their behavior has changed. Exploits are getting faster and more sophisticated.
200K+ ERC-8004 agents
The agent standard is live. 200K+ agents registered with on-chain identity. But identity without behavioral monitoring is a credential waiting to be exploited.
$30T agent economy by 2030
Projected agent-managed value by 2030. 92% of security professionals are concerned about AI agent security. The behavioral monitoring category is wide open — first mover owns it.
The Security Stack — Complement, Not Compete
Hexagate monitors contracts. Hypernative monitors protocols. Nobody monitors agents. We fill the gap.
Hexagate
COMPLEMENTMonitors smart contracts across 75+ chains. $15B TVL protected. Does not monitor agent behavior — complementary layer.
Hypernative
COMPLEMENTMonitors protocols with 300+ risk types. 200+ customers. Alerts on protocol-level anomalies, not per-agent behavioral drift — complementary layer.
Forta
POTENTIAL INTEGRATIONDecentralized bot network for known attack patterns. Pattern-based, not behavioral. No per-agent baselines — potential integration partner.
SentinelCRE
THE GAPMonitors AI agents. Per-agent behavioral baselines, frozen-origin drift detection, dual-AI consensus, on-chain enforcement. The gap nobody else fills.
SentinelCRE: Per-agent behavioral baselines + frozen-origin drift detection + dual-AI consensus + on-chain enforcement + ERC-8004 reputation writes. The agent monitoring layer the security stack is missing.
The Moat
What stops someone from forking this?
Per-Agent Behavioral Data
Every agent monitored deepens our behavioral models. A fork starts with zero baselines. The gap widens with every agent onboarded.
Cross-Protocol Intelligence
Attack patterns detected at one protocol automatically strengthen defense for all others. Network effect — the more protocols that join, the stronger everyone gets.
ERC-8004 Reputation Oracle
We write behavioral incidents to the on-chain Reputation Registry. Protocols query our verdicts. Integration begets integration — infrastructure lock-in.
Chainlink Relationship
1st place at Convergence + Build Program. CRE volume discounts, co-marketing, ecosystem distribution.
Frozen-Origin Baselines
Behavioral scoring with frozen origin baselines for autonomous AI agents — a novel approach protected as a trade secret. Provisional patent filing planned.
On-Chain Proof of Track Record
Every blocked attack is a transaction hash. Our protection history is a public, verifiable ledger. New entrants have zero track record to show.
Traction
Monitoring MVP live. Agent Discovery Scanner built. ERC-8004 integration complete. Self-security architecture deployed (multisig, timelock, kill switch, safety floors).
1st Place
Chainlink Convergence 2026
112
Solidity tests passing
96.1%
Sandbox catch rate (51 scenarios)
ERC-8004
Agent standard integrated
$4B
TVL protected (sandbox)
The Ask
Raising
$750K – $1.5M
Pre-seed via SAFE
Valuation Cap
$8M – $12M
SAFE + token warrant
Runway
18+ months
AI-augmented solo founder model
Use of Funds
- Senior Security Engineer (month 3) — leads agent behavioral analysis and red team engagements.
- Sales / BD Lead (month 5) — DeFi protocol pipeline beyond founder relationships.
- Provisional patent on frozen-baseline behavioral risk engine for AI agents.
- First design partners — DeFi protocols with active AI agent fleets.
- Platform development — agent discovery, behavioral monitoring dashboard, ERC-8004 reputation oracle.
Milestones to Series A
- 3 DeFi protocols on Monitor tier
- 1 anchor Protect-tier customer
- Provisional patent filed
- First $250K in ARR
- Technical co-founder hired
The Vision
The agent economy needs behavioral security infrastructure.
Autonomous AI agents will manage trillions in on-chain value. When they’re compromised, identity systems can’t tell the difference — but behavioral baselines can.
SentinelCRE is building it.
Team
Willis (ProjectWaja)
Founder & Lead Developer
10 years in project management before transitioning to full-stack engineering. Built 30+ projects across Solidity, TypeScript, React, and Python — spanning DeFi security, AI, mobile, and algorithmic trading. The operational background means SentinelCRE is built for real protocol workflows, not academic threat models.
Won 1st place at Chainlink Convergence 2026 against 500+ projects — solo, with AI augmentation. Now productizing via Chainlink Build Program. Focused exclusively on AI agent behavioral monitoring for DeFi.
DeFi Security Focus
Full-stack Solidity + TypeScript + Python. Built for real DeFi protocol workflows and agent threat models.
AI-Augmented Velocity
Solo founder operating at team-scale velocity with AI. 2x runway vs traditionally-staffed startup.
Builder, Not a Talker
112 tests passing, 51 attack scenarios, ERC-8004 integrated, monitoring MVP live — working product, not a whitepaper.
Advisors
Josh Chin — 20-year cybersecurity veteran, NetForce founder, CMMC credentials. Cybersecurity advisory, enterprise introductions.
Sam Friedman — Senior Technical Account Manager, Chainlink Labs. Internal advocacy, ecosystem introductions, Build Program access.