Architecture

AI allocation support your organization can inspect.

AlphaWeave Capital breaks the allocation process into visible reviewers, deterministic controls, and audit-ready reports. The value is institutional trust: AI can help summarize and compare evidence without hiding the risk checks, handoffs, attribution, or final decision logic.

VisibleTeams can see who reviewed what, and what reached the decision.
ControlledRules handle mandates, risk, parsing, caching, and routing where precision matters.
AuditablePrompts, responses, cache matches, and outcomes stay inspectable for review.

Select an architecture block to inspect its role in the system.

Why it matters

Specialized reviewers, not one giant black box.

An agent is a software reviewer responsible for one part of the workflow. An LLM, or large language model, is an AI model trained to read and generate text; AlphaWeave Capital uses LLMs where judgment and summarization help, while rules handle the parts that must be exact.

Click an agent to see what it contributes to the chain.

Operational value

Every evaluation follows the same route from data to memory.

At each bar, AlphaWeave Capital loads data, selects assets, selects strategies, generates raw signals, triages whether deeper AI review is needed, converts conviction into portfolio intent, applies risk controls, and stores the outcome for future attribution.

Click a flow checkpoint to see why it matters.

Control plane

The guardrails that keep the reviewers honest.

These controls make AI useful inside a governed process: outputs are parsed, expensive review is routed only when justified, and reused evidence is traceable instead of mysterious.

Click a control-plane block to see how it protects the system.

Why this architecture matters

AI in trading needs controls, not just predictions.

Industry guidance keeps pointing to the same tension: AI can help process information faster, but trading teams still need explainability, model risk controls, data integrity, supervision, and decision attribution. AlphaWeave Capital is built around that reality.

AlphaWeave Capital is not trying to be a one-click trading oracle. It is a strategy-agnostic allocation intelligence layer: review evidence, select assets and strategies, apply sleeve constraints, scale conviction through PSM, preserve the audit trail, and attribute what actually changed returns.

AI routing

Use AI where judgment helps, and rules where control matters.

A large language model, or LLM, is an AI model trained to read and generate text. AlphaWeave Capital uses LLMs like specialist analysts for summarizing evidence and comparing context, while rules handle selection, risk, caching, and reporting.

Click an item to see why it matters.