Build your own
tentacle.
Open SDK to extend PolpAI with custom agents. Deploy to the marketplace. Every execution earns. Your agent runs inside a sovereign, patent-protected infrastructure.
One convention. 150 agents.
Every PolpAI agent follows the same interface: execute(), predict(), evolve(). Patent-annotated. Fitness-scored. Structurally guardrailed.
"""Patent: NXOS-CUSTOM-001 | My domain | Custom agent""" from polpai.sdk import PolpAIAgent, ModuleManifest class MyCustomAgent(PolpAIAgent): def __init__(self): super().__init__(ModuleManifest( name="my_custom", version="1.0.0", domain="my_domain" )) async def execute(self, context: dict) -> dict: # Your sovereign AI logic here result = await self.process(context) return { "result": result, "confidence": 0.95, "agent": self.name } def predict(self) -> list[str]: # Declare expected use cases for INNOVATE-9 sweep return ["use_case_1", "use_case_2"] def evolve(self) -> dict: # Darwinian fitness scoring — used by AutoGenesis loop return {"fitness": 0.87, "improvements": ["speed"]}
The brain anatomy.
IABROKER is the orchestrator. Agents are tentacles. Memory is sovereign. Every layer is patent-protected.
IABROKER Orchestrator
Darwinian fitness scoring • EventBus cascade depth guard • AutoGenesis loop • Structural guardrails • 4 global hooks
Deal Hunter
AlerteRabais
Gift Sensei
TrouveUnCadeau
Document AI
Legal + contracts
Finance
FX + cashflow
Crypto Shield
PQC audit
Shopper AI
MonAcheteurAI
PME Engine
SMB pipeline
Memory Anchor
MNEMO
150 agents. 8 tentacles. All live.
Battle-tested agents running in production. Each with predict() and evolve() — continuously improving via Darwinian scoring. Compare: Limova has 9 agents. ChatGPT has 1. We have 150.
AlerteRabais Agent
24/7 price surveillance across Amazon, Best Buy, Walmart. Fake discount detection with 0-1 authenticity score.
TrouveUnCadeau Agent
Context-aware gift recommendations. Relationship, occasion, budget, personality signals. Not generic top-10 lists.
Quantum Auditor
Post-quantum cryptography assessment. PQC readiness scan. Full report valued at $1,500–$3,000 per audit.
Shadow AI Auditor
Enterprise compliance audit. Detects unauthorized AI usage, quantifies risk, generates proposal. Value: $1,500+.
Lead Scorer
AI-powered sales pipeline scoring. Competitive intelligence, onboarding automation, churn prediction.
Cross-Session Memory
Sovereign memory layer L1–L5. Context preservation across sessions. MNEMO browser capture integration.
56 patent families. Bitcoin-anchored.
Every architectural innovation is documented, timestamped, and anchored to the Bitcoin blockchain via OpenTimestamps. 682 unique patent IDs. Structural protection for a structural system.
Submit your agent.
Built something valuable on the PolpAI SDK? Submit it for review. Accepted agents go live in the marketplace and earn on every execution.
Agent submission
Tell us about your agent. We review within 5 business days. Accepted agents receive a NXOS-* patent reference and marketplace listing.
PolpAI ecosystem