
Agentic Onchain Treasury (AOT) enters internal testing phase
Aarna.ai's AOT system is now running internal tests with key features operational:
- Real-time agent reallocation of treasury funds
- Onchain policy execution automated through smart contracts
- Built-in yield logic coded directly into the system
- Tokenized vaults and policy engines under active development
The team continues refining the TARS agentic framework that powers the treasury management system.
Aarna Launches Full-Stack Treasury Management Layer for DeFi's $21B in Idle Capital
**The Problem**: Over $21 billion sits in crypto protocol treasuries, with 35-40% parked in stablecoins earning little to no yield. Most treasuries operate through disconnected tools - multisigs, spreadsheets, and manual governance votes. **The Solution**: Aarna introduces **Agentic Onchain Treasury (AOT)** - a complete infrastructure layer that replaces fragmented treasury management with: - **Smart contract-enforced policies**: Allocation bands, diversification rules, and risk triggers operate transparently onchain - **24/7 autonomous execution**: TARS agents monitor markets continuously and rebalance automatically, achieving 25-40% higher yields - **Institutional-grade controls**: Every decision is recorded onchain with explainable reasoning, creating an immutable audit trail - **Modular vault system**: Unified access to fixed income, indices, and growth assets under one risk framework **Why It Matters**: Traditional treasury management relies on human oversight during business hours while crypto markets operate 24/7. AOT eliminates structural inefficiencies through programmable logic and transparent execution. Accessible from 100 USDC deposits with no bridges or technical complexity required. [Read the full breakdown](https://www.aarna.ai/blogs/defi-needs-a-full-stack-treasury-management-layer)
Canton Network's Eric Saraniecki on Rethinking Blockchain Architecture for Institutional Finance
A new podcast episode features Eric Saraniecki from Canton Network discussing blockchain infrastructure design. The conversation examines DeFi's multi-leg transaction logic and Canton Network's approach to building globally composable financial applications. Saraniecki challenges the assumption that decentralization is universally optimal. He argues that different system layers require different configurations: - Institutions need operational autonomy - Systems must connect through a neutral, composable network - Architecture should be pragmatic rather than ideological The discussion focuses on how blockchain infrastructure can scale while accommodating institutional requirements without sacrificing interoperability.
🔧 aTars MCP Expands Beyond Crypto
**aTars MCP is expanding its real-time data pipeline beyond crypto assets.** The system pre-computes derived signals on every update and exposes them as typed MCP tool endpoints, allowing Claude AI to make grounded, reasoned decisions. The architecture includes: - Continuous ingestion of 1-min OHLCV candles - Pre-computation of 40+ indicators (TA-Lib + custom) - Perplexity AI for daily news sentiment - MCP server with structured tool endpoints - Claude as reasoning and visualization layer The team emphasizes that **the challenge isn't building the MCP server—it's the data pipeline behind it**. By pre-computing answers before Claude asks, they eliminate on-demand calculation lag. aTars currently covers 9 major tokens (BTC, ETH, SOL, XRP) with 90-day rolling windows and ML-ready feature matrices. The endpoint is available at [https://mcp.aarna.ai/mcp](https://mcp.aarna.ai/mcp).
AI Spots Market Contradiction in 3 Seconds Using Live Data Pipeline

**When prices fall but sentiment stays bullish, something's off.** On March 17, 2026, both BTC and ETH dropped while news sentiment remained bullish—a contradiction that Claude AI identified in just 3 seconds. The aarna team built this capability using a live data pipeline that feeds **40+ real-time indicators** directly into Claude via MCP (Model Context Protocol). This architecture enables instant analysis of market discrepancies that human traders might miss. **How it works:** - Real-time data from multiple sources flows through MCP - Claude processes 40+ indicators simultaneously - Contradictions between price action and sentiment are flagged instantly This approach mirrors historical patterns where market sentiment diverged from price movements, like the 2025 flash crashes when DeFi TVL remained resilient despite broader market dumps. The system demonstrates how AI can process complex, multi-source data streams to identify market inefficiencies faster than traditional analysis methods.
aarna's tvPTmax Vault Adds New Principal Tokens to Portfolio Allocation
**Portfolio Update** aarna's tvPTmax vault has added fresh Principal Tokens (PTs) to its latest allocation, continuing its strategy of sourcing fixed-yield opportunities across Pendle markets. **Key Features** - Delivers consistent, predictable APY through automated PT selection - Operates fully onchain with transparent execution - Uses TARS agent to scan for 8-12% fixed yields - Employs dual-threshold rebalancing: 80% in stable positions, 20% for opportunistic allocation - Rebalances on deposits >$10k and periodic 14-day reviews **Recent Performance Context** - TARS generating 25-40% higher yields than baseline allocation - December delivered ~9% effective annualized USDC APY - Zero concentration breaches or liquidity stress events - Now integrated with Portals for one-click token zapping The vault provides fixed carry without impermanent loss or liquidation risk, using structured capital allocation across diversified PT markets with institutional-grade risk controls. [View full report](https://engine.aarna.ai/)