Investor Workbench

The Vault is Bridgewater Associates’ internal data platform for indexing, organizing, and serving more than 100 million unstructured documents from external sources, as well as the firm’s memos and meeting transcripts. The material is cleaned, permissioned, and tagged to Bridgewater’s internal ontology for use by human investors and software agents. Its search system is designed to support both investor queries and agent workflows, with fast filtering across the document collection. Bridgewater developed a hand-built benchmark to improve search precision at the top 10 results from about 60% to more than 90%. The platform also supports analyst research, forecasts, signals, and proactive agents that surface relevant information to investors.

Weights & Biases (W&B) is the platform account used in the evaluation workshop, providing inference credits and sandboxes for running evaluation workflows. The workflow uses W&B's Weave tooling to preserve agent trajectories, including commands, messages, and failures, so reviewers can inspect the evidence behind leaderboard scores.

Workshop Studio is a hosted Jupyter notebook environment with pre-created AWS accounts for building and testing an agent. In the described workshop, it hosts notebooks for developing a customer-service agent with Strands Agents, testing tool-call controls, adding skills and steering handlers, and preparing deployment to Amazon Bedrock AgentCore.

ZooWork is an AI agent delivery platform for forward-deployed engineers and domain experts. It packages a role's knowledge, industry context, standard operating procedures, delivery standards, and tool permissions into Agent Skills, then uses them to build agents that plan and execute work, call connected systems, request human approval before sensitive actions, and deliver outputs such as drafts, documents, orders, and system updates. Its Agent Builder turns a conversational specification into a deployable agent, while the ZooWork Agent Runtime handles planning, task breakdown, execution, guardrails, feedback-based skill improvement, isolated sandboxes, autoscaling, GPU-on-demand execution, and observability. It connects to services including Google Workspace, Microsoft 365, Slack, Lark, Jira, and GitHub through MCP-compatible integrations, and agents can be used from ZooWork, Slack, Microsoft Teams, WhatsApp, or embedded applications through an API and SDK. The platform also includes ZooData, which converts URLs into structured, agent-ready data. The page describes tenant isolation, encryption by default, zero data retention, no training on customer data, role-based access, runtime checks on inputs, outputs, and tool calls, approval gates, and replayable audit records. ZooWork offers forward-deployed support for scoping and taking a workflow into production through a stated seven-day pilot process.

Vibe-Trading is an open-source AI personal trading agent and quantitative-finance research workspace developed by HKUDS. It turns natural-language finance questions into runnable analysis and backtesting workflows by connecting an agent to market-data loaders, portfolio and backtesting engines, report generation, persistent memory, and multi-agent research teams across asset classes. Its quantitative and risk-analysis capabilities include strategy backtesting, Heston stochastic-volatility pricing, hierarchical risk parity, copulas, and market-microstructure estimators. The project provides API and MCP interfaces, broker and market-data connectors, portfolio-management and broker-authorized trading workflows, examples, documentation, a demo, and a Shadow Account for testing. It supports read-only account sources such as official IBKR MCP and opt-in Binance USD-M snapshots, while scheduling actions require explicit confirmation and live-order safeguards fail closed on contradictory or invalid data. Implemented as a Python project, it uses a connector onboarding system that stores connection secrets in the operating-system keyring and is distributed under an open-source license.