Each version ships a usable fund.
Value first, armor when there's something worth armoring, and a learning loop once there's history to learn from.
Replatform
A real web platform: browser-based Admin Console with role-based access, kill switches, and live Boardroom streaming, on an event-driven core with a tamper-evident, hash-chained audit store. Cost-metering taps go into the LLM router from day one, and every agent loads into a typed slot through a signed manifest.
Prediction + Behavioral Desk
A dedicated forecasting swarm: regime detection, time-series foundation models and a probability-weighted market model issue predictions into a registry, then get scored against realized data. A new Behavioral Desk adds prospect-theory, attention and extrapolation signals, and the same math turns inward as bias gates on our own agent panel.
Execution + Stewardship
From one ticker to a screened universe worked by a production-grade execution engine with a timing overlay: trading windows, calendar-risk days, participation caps and order-to-trade hygiene. A new Stewardship swarm manages the book after purchase: living theses, deterministic exit discipline, hedging and a drawdown governor. Paper first, always through the gates.
Governance, Mandate & Graph
A compliance swarm audits every agent against the rulebook and files findings with citations. Humans steer any agent through an instruction console, and every autonomous campaign runs inside a human-signed Mandate compiled to OPA policy. A shared knowledge graph and per-swarm lineage DAG become the fund's memory, and the Budget Governor + Bursar report net-of-cost P&L.
LoQOS + Foundry
The fund becomes a machine. LoQOS, an immutable, reproducible OS image, runs every agent in its own micro-VM with kernel-level network and policy enforcement. Validated "golden rules" and mandate envelopes compile into a policy kernel no agent can route around, and The Foundry begins running governed self-improvement ratchet loops on the Prediction and Research swarms.
Fund-in-a-Box
The learning loop closes across every swarm: a neural engine trains on the fund's own audit history, The Foundry runs mandate-governed experiments on all seven swarms, and the golden-rule lifecycle matures. Multi-book support, a first slice of live capital. Install the image, get a hedge fund.
Three swarms become seven.
Specialists, not a monolith. Every swarm evolves, and every swarm is watched.
Ingestion
Agent-supervised data pipelines that detect drift, gaps and staleness, filing incidents instead of failing silently, now including search-interest, options-chain and futures-basis feeds.
Prediction
Probabilistic forecasts scored daily against realized data, plus a Behavioral Desk of prospect-theory, attention and extrapolation signals. Skill earns trust; decay loses it.
Research
Desks, debates and adjudicated house views, grounded in a shared knowledge graph and policed by bias gates that catch herding and overconfidence in our own agents.
Execution
Deliberately not an LLM: a deterministic engine with a timing overlay that only accepts tickets clearing every gate, every policy and every trading-window rule.
Stewardship
The post-purchase layer: living theses re-underwritten on cadence, deterministic exit discipline, a hedging module and a drawdown governor watching the whole book.
Compliance
Continuous rule checks plus auditors that replay agent histories against regulation, and manipulation-pattern detectors run on our own flow as examiner-ready proof.
Neural Engine, Foundry & Bursar
The audit trail becomes training data. The Foundry runs governed self-improvement inside mandate envelopes, and the Bursar meters every dollar into net-of-cost P&L.
Where humans stay.
We automate everything a fund does that machines do better, and we're explicit about what they don't.
| function | who runs it |
|---|---|
| Research, forecasting, backtesting, order flow, monitoring | AUTOMATED (gate-checked) |
| Reconciliation & NAV | AUTOMATED |
| Exception resolution | HUMAN |
| Compliance judgment & regulatory sign-off | HUMAN |
| Capital allocation across strategies | HUMAN (machine proposes) |
| Mandate approval (objective, budget, legal scope, promotion criteria) | HUMAN (agents autonomous inside it) |
| Crisis discretion & kill switches | HUMAN (by design) |
Each release lands with a recorded, auditable run.
Paper-trading preview today. Reach out to follow along or talk investment.