Welcome to

Brookstreet is an award-winning investor augmented by Brookstreet Intelligence 2.0, our AI platform. The firm was established by a select group of Shipping and principal-capital families. Our 2.0 Thesis is built on optimising the Efficiency Frontier through Diversification across Stages, Markets, Regions, and Instruments. We deploy Hybrid Growth Capital with a clear emphasis on ROCE and DPI. We invest in Artificial Intelligence (AI) innovations that deliver transformative commercial solutions in Digital Transformation, Energy Transition, Longevity, and Dual-Use Defence Technologies. Headquartered in London’s Mayfair, we operate worldwide, partnering with founders and investors across the USA, Europe, MENA, and Asia.

Apply For Investment
Join Our Network

Brookstreet Intelligence 2.0

One Institutional Brain. Multimodal, Multi-Model, Multi-Agent Intelligence Orchestration

Brookstreet Intelligence 2.0 is our proprietary intelligence, research and execution architecture, embedded across the firm. Developed around the PEVC investment lifecycle, it now provides a common intelligence layer across Private Markets, AI & Technology and Public Markets.

One institutional memory connects frontier models, specialised agents, tools and proprietary data. Agents can work independently or in coordinated councils—researching, challenging assumptions, stress-testing analysis and structuring outputs before conclusions reach the human decision layer.

Across Private Markets, the architecture supports sourcing, diligence, valuation, investment committees, portfolio monitoring and exits. Across AI & Technology, it supports research, product development, experimentation and venture creation. Across Public Markets, it supports research, monitoring, quantitative analysis and systematic workflows.

Institutional intelligence: persistent, connected, challenged and capable of execution.

Investment Architecture

Access to frontier AI is rapidly becoming commoditised. Durable advantage comes from combining proprietary data, institutional memory, domain expertise, governance and execution architecture, so intelligence can be applied consistently across investment strategies.

Brookstreet Intelligence 2.0 treats models as interchangeable components in a broader operating system. GPT, Claude, Gemini, Grok, Kimi, GLM, open models, and future frontier models can be connected through APIs, selected for individual tasks or coordinated in model councils.

Above the models sit Brookstreet’s orchestration, agents, tools, workflows and memory. Security, permissions, auditability, compliance and human oversight surround them. This allows one architecture to serve different investment strategies without building a separate intelligence stack for each asset class.

The Fund teaches the AI, and the AI improves the Fund.

Artificial Intelligence at Brookstreet (& Key Technologies)

  • 2017–2022, Origins: We began exploring AI in 2017, following the publication of Google’s “Attention Is All You Need”, which introduced the Transformer architecture. We continued that exploration as investors through 2022. By 2022–23, we had developed our own views on how AI could be applied across private equity and venture capital.

  • 2023, First Steps: Fortune covered Brookstreet’s AI strategy for PEVC fund management. Our understanding of frontier technologies was shaped by investments in Behavioral Signals (Silicon Valley-based USA), applying AI to speech and vocal-behaviour analysis; Autonomo (EU-founded, Germany), applying computer vision; and Perceptual Robotics (UK-based), combining autonomous tech with Physical AI and the emergence of AGI. We also worked with portfolio companies backed by frontier-technology investors such as IQT. This interest was grounded in our own backgrounds: like many investment bankers and fund managers who entered the City of London in the 2000s, we came to finance after studying science and technology.

  • 2024, Connectivity: As frontier models became natively multimodal and the Model Context Protocol (MCP) emerged, by Anthropic engineers, we moved beyond single-model use toward tool-connected, interoperable AI architecture.

  • 2025, Orchestration: With the arrival of reasoning models and agentic systems, we began experimenting with “Orchestrations”, model councils, specialised agents and autonomous multi-step workflows.

  • 2026, Brookstreet.AI Rapid Acceleration Initiative

  • 2026 January, Production: We unified these components into a production-grade institutional stack: proprietary memory shared across engines, multi-LLM councils, tool connectivity and autonomous orchestration under human oversight.

  • 2026 March, Release: Brookstreet Intelligence 2.0 was formally released as Brookstreet’s central intelligence, research and orchestration environment.

  • 2026 April, Validation: Brookstreet Intelligence 2.0 was profiled in “Merging Agentic AI, MCP and Human-Led Insight in Private Equity” co-developed with Third Bridge.

  • 2026 June, AI Leadership: Christopher Willmott, former CFO of JPMorgan Global Fund Services and an experienced leader in automation and technology transformation, joined the Brookstreet Board. His appointment strengthened the Board-level governance supporting Brookstreet Intelligence 2.0, alongside Brookstreet’s Principal assuming the role of Chief Intelligence Architect. By this stage, Brookstreet Intelligence 2.0 had evolved into a single-brain, multi-council, multi-agent architecture embedded throughout the investment-management lifecycle.

  • 2026 July, Brookstreet Intelligence Academy: Brookstreet launched its full six-volume AI Mastery Course, designed to take private-capital professionals from the foundations of modern AI through to the design, deployment and governance of institutional AI systems. The curriculum covers leading frontier systems—including GPT-5.6, Claude and Fable 5, Kimi 3, Gemini, Grok and Nemotron—while focusing increasingly on the architectures that matter more than any individual model: MCP, retrieval-augmented generation, agents, model councils, orchestration, automation and, above all, institutional memory. Read: Brookstreet AI Academy — AI Mastery Course.

  • 2026 August, Model Agility: On 20 August, Ox Alpha appeared anonymously and quickly attracted developer attention. Reported coding tests showed 80% versus Claude Fable 5 at 65% and GPT-5.6 Sol (we now have Sol and Astra) at 52%, while one bug-fixing comparison suggested materially lower output-token usage. The numbers were directional rather than definitive, but the lesson was clear: the frontier can move in days. Shortly afterwards, Z.ai revealed Ox Alpha as its stealth GLM-5.3-Flash model. Brookstreet Intelligence 2.0 had already been designed for precisely this environment. The strategic advantage is not predicting whether GLM, Fable, Sol, Gemini, Grok, Kimi, DeepSeek, Qwen, Nemotron—or a model not yet released—will lead. It is building the layer above them: models connected through APIs, selected dynamically, deployed individually or in councils, and combined with Brookstreet’s own orchestration and institutional-memory layers. We do not need to predict the winner. We build the system that can use the winners. Read: Brookstreet AI — Model Agility.

  • 2026 September, The AI Cake: Inspired by NVIDIA’s five-layer framing—Energy → Chips → Infrastructure → Models → Applications—Brookstreet expanded the concept into the full AI value chain, adding Orchestration + Governance + Domain Expertise. AI is not one model, one chatbot or one software category; it is an interconnected industrial system spanning power, semiconductors, data centres, frontier intelligence, applications, autonomous execution, institutional controls and proprietary know-how. Brookstreet Intelligence 2.0 sits across these layers, connecting the technology stack to real-world investment and operating outcomes. Read: The Brookstreet Intelligence AI Cake.

  • 2026 September, Agentic Financial AI: Just 13 days after OpenAI announced its new Enteprise Agentic Financial Services platform on 10 September, Brookstreet committed to integrating the capability into Brookstreet Intelligence 2.0. Developed with Morgan Stanley and Evercore as design partners, the platform combines frontier AI with institutional financial data and enterprise workflows for research, analysis and modelling. For Brookstreet, this adds another powerful capability to our Frontier AI architecture, deliberately designed to remain model-agnostic while continuously integrating the world’s most advanced frontier AI models — often within days of their emergence. The best model for the task. The best data for the question. One intelligence layer. Read: Brookstreet Integrates Agentic Financial AI into Brookstreet Intelligence 2.0.

  • 2026 October, One Intelligence Layer Across All Capital: Brookstreet Intelligence 2.0 is tested, and will expand, beyond its original PEVC use case to become the common intelligence layer across three complementary engines: Private Markets + AI & Technology + Public Markets. The same institutional architecture—proprietary data, research, multi-model councils, agents, tools, execution workflows and persistent memory—can support PE, VC, Growth, Hybrid Capital and Special Situations; AI ventures, technology intelligence and products; and hedge funds, public equities and systematic strategies. The system is becoming a product in itself and we are in discussions about the next steps we should take. One intelligence layer across all capital. Read: The Brookstreet AI Vision.


    www.brookstreet.ai