From Summaries to Smart Agents: The Rapid Evolution of AI in Investment Management
A firsthand account of how generative AI is transforming from simple productivity tools into autonomous systems that will fundamentally reshape how investment professionals work
Technology industries are defined by pivotal moments when incremental improvement gives way to fundamental transformation. Having witnessed every major computing revolution since joining Intel in the early 1980s, I can state with conviction that the current AI transformation represents something unprecedented—not just in its capabilities, but in the velocity of its evolution.
The Three Great Computing Transformations
My career has spanned three distinct computing eras, each lasting roughly two decades and characterized by iconic breakthrough applications that made abstract technology tangible to millions:
1980-2000: The PC/Internet Era
The transformation began with standalone computing, exemplified by Lotus 1-2-3, which made spreadsheets indispensable to financial professionals. Netscape later democratized internet access, connecting these isolated productivity islands. For those in finance, imagine working without spreadsheets—it's nearly impossible. That's the scale of transformation we're discussing.
2000-2020: The SaaS/Mobile Era
AWS eliminated infrastructure barriers with credit-card simplicity, while Apple's iPhone put computational power in every pocket. Real-time data became ubiquitous, fundamentally changing how investment decisions were made and communicated.
2020-2040: The Generative AI Era
NVIDIA's GPUs made large-scale AI computation feasible, while OpenAI's ChatGPT achieved the fastest technology adoption in history. But here's what's different: the pace of evolution within this wave dwarfs anything we've seen before.
The Unprecedented Velocity of AI Development
The numbers tell a story of exponential acceleration that defies historical precedent. In 2017, we had perhaps two significant AI models. By 2024, over 100 major models compete for attention, with thousands more on platforms like Hugging Face—a compound annual growth rate of 167%.
But model proliferation barely scratches the surface. At Google I/O, the company revealed that token usage had exploded 50-fold to 480 trillion tokens monthly. While user growth was impressive at 10x, usage grew 50x—indicating a fundamental shift from shallow interactions to deep, transformative engagement.
This explosion isn't just about scale; it's about capability evolution happening along three independent axes:
Pre-training scaling: Broader foundational knowledge through massive data ingestion
Post-training refinement: Specialized expertise for specific domains
Test-time computation: The game-changing ability to reason through problems
The February 2025 launch of reasoning models like DeepSeek marked an inflection point. Suddenly, AI wasn't just pattern-matching—it was thinking step-by-step through complex problems, showing its work, and distinguishing correlation from causation.
Software Development: The Canary in the Coal Mine
To understand investment management's future, observe software development's present. Cursor, an AI-enhanced IDE, rocketed to $100 million in annual recurring revenue faster than any SaaS company in history. The reason? Developers aren't just getting code suggestions—they're delegating entire tasks to AI.
OpenAI's Codex and Google's Jules represent the vanguard of this transformation. These aren't assistive tools; they're autonomous agents that execute complete coding tasks. Microsoft now generates 30% of its code through AI, heading toward 60% within two years. When coding—one of humanity's most complex cognitive tasks—can be delegated to AI, what makes investment analysis immune?
The numbers are staggering: 7 million developers building with Google's Gemini (5x growth in one year), enterprise AI usage up 40x, and GitHub repositories dedicated to AI growing vertically after ChatGPT's launch. This isn't experimentation; it's production-scale transformation.
The Abundance of Intelligence
What we're witnessing is an unprecedented abundance of intelligence capabilities. Every day brings new models, tools, and techniques. The old paradigm of building and protecting proprietary systems cannot compete with this explosion of innovation.
Consider the cost dynamics: AI processing costs have dropped faster than any technology in history—80% reduction in 24 months while capabilities expanded by orders of magnitude. A top-tier AI model now costs less monthly than a single Bloomberg terminal.
This creates a fundamental strategic choice for investment firms:
The Enhancement Path: Legacy providers like Bloomberg and Refinitiv bolt AI features onto existing systems, achieving 20-30% productivity gains while preserving current business models.
The Transformation Path: AI-native platforms rebuild workflows from scratch, achieving 10x capability multiplication through continuous learning systems and architectural advantages.
From Information Access to Autonomous Workflows
The evolution of AI in investment management follows three distinct horizons:
Horizon 1 - Information Access (Current State)
Most firms operate here, using AI for document summarization, semantic search, and basic automation. Productivity gains range from 30-75%, primarily in time savings on routine tasks.
Horizon 2 - Insight Generation (Emerging)
Leading firms are beginning to use AI for pattern detection across multiple data sources, integrating structured and unstructured data to generate novel insights. Each senior professional effectively gains the output of 1-2 junior analysts.
Horizon 3 - Autonomous Workflows (12-24 Months)
The future state features AI agents that autonomously monitor markets, execute analysis, and manage investment workflows 24/7. This isn't science fiction—it's the logical extension of capabilities already demonstrated in software development.
A Personal Case Study in Transformation
During Christmas last year, I faced the perennial challenge of writing a comprehensive investment memorandum—without associates to delegate to. Using Claude 3.5 and a reimagined workflow, I transformed a multi-week process into a matter of hours.
The key wasn't just using AI as a writing assistant. I fed it questionnaires, virtual data room documents, presentation materials, and crucially, transcripts from every due diligence meeting. These transcripts represented institutional knowledge that traditionally lived only in human memory. The AI synthesized everything into a 60-page institutional-quality memorandum that required minimal editing.
This wasn't a 30% productivity gain—it was a 10x transformation in how core work gets done.
The New Competitive Imperative
Major sovereign wealth funds now require sophisticated AI integration as a precondition for new allocations. This isn't optional enhancement; it's existential requirement. The investment industry faces the same choice that confronted software companies when the internet arrived: transform or become irrelevant.
The building blocks for transformation already exist:
Reasoning models that work through complex investment theses
Retrieval systems that access proprietary data
Agent frameworks that execute autonomous analysis
Code generation that overcomes LLMs' quantitative limitations
Success no longer comes from building proprietary AI systems—it comes from orchestrating the abundance of available intelligence better than competitors.
Strategic Imperatives for Investment Leaders
1. Embrace Abundance Thinking
The old model of protecting proprietary analytical methods cannot compete with daily innovation from millions of developers. Your competitive advantage lies in orchestrating external capabilities with deep domain expertise.
2. Start Tonight, Not Tomorrow
Every month of delay widens the gap with early adopters. Begin with simple experiments using Claude or GPT-4 for real analysis. The cost is trivial; the learning is invaluable.
3. Prepare for Role Transformation
Portfolio managers won't be replaced by AI—they'll orchestrate AI agent swarms that analyze thousands of opportunities simultaneously. The question isn't whether this will happen, but whether you'll lead or follow.
The Choice Before Us
In two years, investment professionals' jobs will be radically different. Not five years—two. The firms that capture the emerging abundance of intelligence will operate at 10x the analytical capacity of those protecting traditional processes.
When I exchanged emails with conference organizers to arrange my presentation, those emails were AI-generated—I merely approved them. This isn't the future; it's the mundane present. The extraordinary future lies in AI agents that will transform investment management as fundamentally as spreadsheets transformed finance forty years ago.
The only question that matters: Will you be among those doing the transforming, or among those being transformed?
Michael Bruck is Managing Partner at 71 Capital Ventures, where he leads AI strategy and implementation. This article is adapted from his presentation at the Robeco Investment Day, June 2024.