What we built, what we learned, what is now coming together. The backdrop: why “ops” became the real product.
 2025 made one trend impossible to ignore: Stablecoins are now too large for DeFi to keep treating yield as a collection of ad hoc clicks. As of writing this, DefiLlama’s stablecoin dashboard puts the total stablecoin market cap at $308.374b.
When capital sits in stables at that scale, the bottleneck shifts from “finding yield” to running yield: selection, routing, risk limits, reporting, exits, and governance.
At the same time, fixed-rate markets matured into real venues. Pendle’s DEX volume over the last 30 days sits around $934m on DefiLlama. That activity matters because it signals allocator intent: a growing preference for predictable carry and instrumented risk, not just floating rates.
We took that as the design constraint for the year. The goal was not to add more strategies. It was to reduce operational burden while keeping non-custodial control and verifiability intact.
<h3 id="strong-building-the-rails-from-vaults-to-capital-primitives-strong">building the rails: from “vaults” to capital primitives</h3>
aarnâ’s product work in 2025 concentrated on the stablecoin-first foundation, then progressively added composability and structure.
The key step was moving from “a vault that earns” to “a vault position that can be used.” The âtv111 and âtvUSDC rollout in June framed this clearly: âtv111 as the structured stablecoin vault, âtvUSDC as the ERC4626 composable representation.
This may look incremental from the outside. Internally, it forced a different standard of engineering: tighter accounting, clearer NAV semantics, and a more explicit separation between the yield source layer (lending and basis opportunities) and the tokenized exposure layer (what other protocols and users can actually compose).
In parallel, we continued hardening the core infrastructure and security stack. aarna’s tokenization platform has gone through independent audits, with all findings resolved before deployment, table stakes for software intended to function as on-chain treasury infrastructure, not an experimental yield interface.
<h3 id="strong-the-uncomfortable-part-802-pmf-and-learning-to-respect-regimes-strong">The uncomfortable part: 802, pmf, and learning to respect regimes</h3>
One of the most valuable outcomes of 2025 was also the most uncomfortable: we got sharper about what works in production and what does not.
The 800 series was designed to express distinct market logic.
The âtv802 vault’s live performance diverged from our initial assumptions due to a combination of factors, including a limited executable token universe and overly constrained risk management during execution.
This period reinforced a core principle: preserving principal and delivering consistent, risk-adjusted returns is more important than pursuing high-risk outcomes.
Incorporating these learnings, we launched âtv808 a contrarian crypto asset vault designed for investors seeking asymmetric upside by capitalizing on undervalued market conditions.
While âtv808 is still being refined to improve discipline and consistency, this approach has already validated our hypothesis: within a limited token universe, longer-term positioning in undervalued markets with fewer rebalances produces better outcomes than frequent, hard exits.
In live conditions, the system's execution layer performed successfully. However, the lesson was not 'AI works' or 'AI fails,' but rather that translating AI signals into constrained, high-return strategies proved exceptionally challenging in volatile market conditions.
This experience reinforced that regime sensitivity is not optional, and that PMF in DeFi is as much about operational reliability and constraint design as it is about signal quality
We treated 802’s challenges as a forcing function. We tightened how strategies are expressed, how constraints are enforced, and how users reason about system behavior.
PMF, for us, is reached when users can delegate complexity-trusting the system to operate within a clearly defined mandate without needing to understand every underlying venue or micro-decision.
<h3 id="strong-fixed-rate-execution-as-the-turning-point-tars-and-tvptmax-strong">fixed-rate execution as the turning point: âTARS and âtvPTmax</h3>
The most important architectural development of the year came late, and it marked a shift from building yield products to operating fixed-rate execution as a system. Rather than offering another abstraction over Pendle PTs, âTARS and âtvPTmax formalize the execution and operational layer required to run fixed-rate strategies with precision and scalability at treasury scale.
Pendle’s Principal tokens enable predictable carry, but using them effectively demands continuous operational decisions, tenor selection, liquidity management, roll timing, and exit discipline.
âTARS is aarna’s policy-driven execution system: a multi-agent engine that evaluates fixed-rate opportunities, enforces maturity, liquidity, and risk constraints, and executes on-chain actions within defined bounds.
âtvPTmax is the ERC-4626 vault managed by âTARS, allowing users to deploy USDC into diversified Pendle PT exposure without managing individual maturities or rolls. Together, they move fixed-rate yield from discretionary, operator-led workflows to policy-bound execution, where treasury behavior is encoded, repeatable, and auditable by design.

<h3 id="strong-partnerships-and-integrations-expanding-the-capital-surface-area-strong">partnerships and integrations: expanding the capital surface area</h3>
In 2025 aarnâ’s partnerships were selected to expand what capital can do inside the system, across fixed-rate execution, stablecoin operations, liquidity access, and agent-compatible environments.
Capability-driven partnerships
Pendle became a foundational partner for fixed-income composability. PT/YT markets enabled predictable carry, underpinned âtvUSDC tokenization, and formed the execution substrate for agentic fixed-rate strategies such as âtvPTmax.
Sonic and Monad were chosen as execution environments aligned with agentic strategies. Sonic offers deep liquidity, Pendle composability, and yield stacking, while Monad provides high-throughput, low-latency execution suited to adaptive, agent-driven workflows.
Core product integrations: the capital rails
Underneath the agentic architecture sits a deliberately chosen DeFi stack that enables capital sourcing, execution, and settlement.
Money markets and yield sources such as Aave, Compound, Morpho, Dolomite, Silo, and Moonwell provide the base layer for lending, leverage, and yield routing.
Liquidity and execution venues, including Pendle, Balancer, Uniswap, 1inch, Odos, and Aerodrome support fixed-rate markets, vault liquidity, routing, and efficient onchain execution.
Together, these partnerships and integrations form the capital rails of aarnâ, not as isolated touchpoints, but as a coherent stack that allows policy-bound, agentic treasury operations to function reliably across markets and chains.
<h3 id="strong-validation-and-ecosystem-signals-without-pretending-it-is-done-strong">Validation and ecosystem signals, without pretending it is “done”</h3>
We do not treat validation as a single moment. We treat it as a growing set of external signals that indicate the direction is coherent.
One such signal in 2025 was Circle’s USDC Developer Grants. In its June 10, 2025 cohort announcement, Circle listed aarnâ Protocol (Singapore) among the recipients.
That matters because it aligns directly with the product thesis: stablecoin-native infrastructure designed to make on-chain finance operationally usable.
Another signal is simple observability. The system is live, onchain, and independently tracked, like by DeFiLlama.
It is evidence of a working system that can be observed, audited, and iterated in public.
At the product level, âtvUSDC has consistently ranked among the top-yielding Pendle pools, sustaining 20%+ APY during peak periods.
More importantly, the protocol has crossed $1.5M+ in lifetime TVL organically, without mercenary liquidity or short-term emission tactics, suggesting capital participation driven by structure and risk posture, not incentives.
<h3 id="strong-the-people-dimension-the-aarnauts-and-the-culture-required-for-deep-work-strong">the people dimension: the aarnauts, and the culture required for deep work</h3>
aarnâ is built by a compact, distributed core team across Singapore and India, oriented toward shipping infrastructure that must operate under real market conditions.
The operating bias is correctness, durability, and clarity qualities that only become visible once systems are live and exposed to stress. Work is driven by production feedback rather than abstract design.
That standard extends beyond the core team to the broader ecosystem of aarnauts who engage with the protocol as operators, researchers, educators, early users, and interlocutors.
Through the <a href="https://x.com/aarnasays/status/1937224159622496528?s=20" rel="noopener noreferrer nofollow">Vanguard Program</a>, 400+ micro-contributors participated across education, research, referrals, and structured feedback, with contributions manually reviewed to maintain signal quality rather than volume.
In parallel, <a href="https://x.com/aarnasays/status/1968449647460143161?utm_source=chatgpt.com" rel="noopener noreferrer nofollow">aarnâ Quintets</a> brought together top DeFi thought leaders in long-form collaborations that lived across narrative, product understanding, and ecosystem discourse, enabling deeper conversations about structured yield, agentic execution, and allocator trust rather than one-off commentary.
If 2025 had a defining internal lesson, it was building people and systems that behave predictably under stress, with constraints that are explicit, enforced, and understandable. That standard demands a culture oriented toward deep work, restraint, and long-term system integrity over short-term optics.
<h3 id="strong-how-it-comes-together-aot-placed-where-it-belongs-in-the-story-strong">how it comes together: AOT, placed where it belongs in the story</h3>
Everything we built, tested, and iterated from vaults to policies, from agentic execution to operational feedback was converging toward a single system. That system is the Agentic Onchain Treasury (AOT).
AOT is not a vault catalog, and it is not an “AI strategy.” It is the operating architecture for on-chain capital, where humans define policy, autonomous agents execute within those constraints, and the blockchain provides continuous verification.
Policy encodes risk limits, maturity bounds, allocation intent, and liquidity thresholds.
âTARS, the agentic execution layer, deploys and rebalances capital across markets like Pendle PTs, always within policy rules. Vaults such as âtvPTmax are expressions of this system, user-facing access points that rely on AOT and âTARS to operate safely and predictably.
The point is not automation for its own sake. It is the culmination of 2025’s learning: moving from experimental products and manual interventions to policy-bound, auditable, resilient treasury operations.
Every insight from early vault deployments, risk testing, and user feedback feeds directly into AOT, making it a system that behaves predictably under stress, enforces constraints, and lays the foundation for infrastructure-grade on-chain capital management.
<h3 id="strong-2026-what-we-are-building-toward-strong">2026: what we are building toward</h3>
If 2025 was about laying rails and proving they work in real conditions, 2026 is about making them treasury-grade: clearer mandates, tighter constraint enforcement, better reporting, deeper liquidity pathways, and distribution that does not depend on short-lived incentives.
The objective is simple to state and hard to execute: make onchain treasury operations feel boring, predictable, and verifiable. That is what the stack is being built to deliver.

aarna 2025 year-end update

Table of Contents

aarnâ protocol

Other articles published on Jan 16, 2026

Join our free newsletter for daily crypto updates!