Trishool (SN23) Partners with Velantris, Secures First Proof Point for Halo

Trishool, the decentralized AI safety infrastructure provider operating on Bittensor Subnet 23, has announced a commercial partnership with Velantris, an enterprise AI agents startup. This deal is aimed at deploying Halo’s constitutional safety layer inside autonomous agent workflows already engaged with major financial institutions and Tier-1 system integrators.

Under the agreement, Halo, Trishool’s open-source constitutional AI guardrail architecture, will be integrated into every Velantris client engagement under one of two deployment paths:

a. Direct deployment using Velantris’s standard guardrail configuration, and

b. Custom deployment tuned to the specific policy framework and risk profile of the individual client.

Velantris is currently building autonomous agent swarms designed to automate enterprise functions, beginning with quality assurance. 

Its commercial footprint already includes:

a. Live engagements with Tier-1 system integrators,

b. Active client work with major banks, and

c. Exposure to the broader enterprise agents market, estimated at over $500 billion.

“Autonomous agents do not get past compliance without a guardrail layer the bank can read line by line,” Trishool noted, framing the deal as evidence that the regulatory window for ungoverned AI deployment is closing rapidly.

Why Halo Clears Enterprise Compliance Where Other Guardrails Fail

For Bittensor, the partnership is a notable data point in the network’s broader push to demonstrate that subnet-built infrastructure can find real-world enterprise adoption beyond crypto-native use cases. Rather than competing for individual developers or small teams, Trishool is positioning Halo as backend safety infrastructure inside the platforms whose end customers cannot accept uncontrolled agent behavior under any circumstances.

That distinction matters commercially because enterprise compliance teams require guardrails that meet a specific set of conditions:

a. Open source, so internal compliance and security teams can inspect every line,

b. Constitutional, with explicit rule structures rather than opaque pattern-matched behavior, and

c. Customizable, with risk policy living inside the agent rather than bolted on externally.

Halo’s architecture is the only currently-deployed safety layer that meets all three conditions simultaneously, which is the structural reason it can move through enterprise procurement processes that other guardrail layers cannot.

Trishool has indicated that the Velantris partnership represents the cleanest current expression of its target customer profile, which includes:

a. Agent-native startups deploying autonomous systems into regulated enterprise environments,

b. Platforms whose end customers cannot tolerate uncontrolled agent behavior, and

c. Companies that cannot afford to build a guardrail layer themselves.

The Companies Behind the Partnership

Trishool (Bittensor Subnet 23) is building open-source constitutional AI safety infrastructure including Halo, its flagship guardrail model. Its architecture is designed for inspectability, customization, and integration into platforms whose end customers require auditable agent behavior.

Velantris, on the other hand, is an enterprise AI agent startup developing autonomous agent swarms for enterprise function automation, with current focus on quality assurance workloads. The company’s client base includes Tier-1 system integrators and major banks operating in live production environments.

Further partnership announcements are expected in the coming weeks, with Trishool indicating that additional enterprise integrations are already in the pipeline.