Summary of Gabe Stengel - Building Investing Superintelligence - [Invest Like the Best, EP.492]

Overview of Gabe Stengel - Building Investing Superintelligence

In this episode of Invest Like the Best, Patrick O’Shaughnessy speaks with Gabe Stengel, co-founder of Rogo, about how AI is reshaping finance and why the biggest opportunity in vertical AI is not just intelligence, but the full infrastructure around it: data, workflow, compliance, auditability, and transaction plumbing. Gabe argues that Rogo is building “investing superintelligence” for capital markets, starting with dealmakers and investment banks, then eventually expanding into broader private markets and possibly public equities. The conversation centers on how AI is changing the job of investors, what remains defensible in a world of rapidly improving models, and what it will take to build the next generation of finance software.

Rogo’s Product Vision and Evolution

From “AI demo” to real workflow product

• Gabe explains that the company went through multiple false starts before the models were good enough.
• Early versions were impressive in demos but unreliable in practice.
• The inflection points came with stronger reasoning and reliability from newer models, which made Rogo useful for real finance work.
• The product evolved from “search tool” to something that can handle tasks junior bankers and analysts used to do manually.

What Rogo is ultimately trying to build

• A full AI operating layer for finance.
• Not just a chatbot, but a system that can:
  • read and synthesize documents,
  • prepare and unpack data rooms,
  • draft deal materials,
  • update CRM and portfolio systems,
  • manage workflows across the deal lifecycle,
  • eventually support autonomous transaction execution.

How Rogo Is Used Today

Core users: dealmakers and bankers

• Rogo’s current core customer is mainly large banks and dealmakers.
• The strongest use cases are in transaction-heavy workflows:
  • M&A preparation,
  • diligence,
  • data room management,
  • IC memo drafting,
  • DDQ responses,
  • presentation markups,
  • coordination across third parties.

Typical workflow example

• A managing director can email a deck markup to Rogo.
• Rogo returns a revised version in minutes instead of days.
• The junior analyst is alerted.
• Every edit is auditable and traceable.
• The result is faster turnaround with better visibility and compliance.

Public equities vs. private markets

• Gabe says public equities may be the “best” AI application in theory because the data is abundant.
• But the product wedge is stronger in private markets and dealmaking because there is more manual work and more “plumbing” to replace.
• Private markets are especially attractive because AI can standardize chaotic, human-heavy workflows.

Data, Models, and the “Harness” Layer

The three-layer stack

Gabe repeatedly emphasizes that the real opportunity is not just the model:

1. Data — what information the system can access.
2. Models — how well the AI can reason and respond.
3. Harness / infrastructure — the workflow layer that makes the model actually useful.

Why the harness matters most

• Many tasks fail not because the model is weak, but because the workflow is poorly designed.
• Rogo focuses on:
  • routing tasks to the right model,
  • collecting the right context,
  • ensuring outputs are auditable,
  • integrating into existing systems,
  • building around regulatory and compliance requirements.

Compaction and memory are major unsolved problems

• One of the biggest limits in agentic AI is “compaction”:
  • how an agent remembers the right things over many interactions,
  • how it retains context without overflowing token limits,
  • how it behaves consistently across long-running workflows and large groups.
• Gabe sees this as especially important as AI moves from one-to-one use toward collaboration across teams and firms.

Why Finance Is a Strong Vertical for AI

Finance has massive workflow depth

• Finance is full of niche workflows, messy data, and high-value decisions.
• This creates room to build large businesses by going deep, not broad.
• Rogo’s thesis is that many of the most valuable AI products will be vertical systems with heavy workflow and infrastructure needs.

The “Bloomberg playbook,” updated for AI

• Gabe compares Rogo’s strategy to Bloomberg:
  • get in the door with a useful tool,
  • expand into workflows,
  • then become the communication/transaction layer.
• The big difference: Rogo is building infrastructure for agents to transact, not just humans.

The market opportunity

• If AI can reduce friction in capital formation, pricing, and diligence, it could:
  • make markets more transparent,
  • increase liquidity,
  • speed up dealmaking,
  • open capital markets to smaller businesses,
  • reduce the dependence on human intermediaries.

What Makes a Great Vertical AI Business

The industry must be complex enough

• A good AI vertical needs:
  • deep workflows,
  • nontrivial integrations,
  • real compliance burden,
  • enough complexity that generic copilots aren’t enough.

The team must have domain expertise

• Gabe believes domain expertise matters more as the product gets closer to the workflow.
• You need people who understand the actual job to design the product well.
• Rogo has built a team with heavy finance experience, including people from banks and investment firms.

The company must be willing to reinvent itself constantly

• A strong vertical AI company can’t become attached to a static product.
• Gabe cites the idea of rebuilding core infrastructure repeatedly to avoid ossification.
• If the models change every few months, the product stack has to be able to change too.

Business Model and Go-to-Market

Enterprise sales is still a human business

• Rogo sells like classic enterprise software today.
• That means:
  • seat-based or traditional enterprise pricing,
  • AEs, solutions engineers, and onboarding support,
  • hands-on deployment with large institutions.

Different pricing models for different AI businesses

• Gabe distinguishes between:
  • token/usage businesses like some AI coding tools,
  • enterprise workflow businesses like Rogo.
• He believes the long-term end state in finance should be outcome-based pricing:
  • pay per good investment idea,
  • pay per completed report,
  • pay per deal artifact created,
  • pay for actual value delivered, not tokens consumed.

Adoption is rising, but firm-level transformation is still the challenge

• Individual bankers already feel much more productive.
• The harder question is whether that productivity turns into:
  • more revenue,
  • lower costs,
  • new markets served,
  • better firm-wide strategy.

Leadership, Culture, and Execution

Building fast is painful

• Gabe describes startup building as “chewing glass.”
• The pain points include:
  • hiring mistakes,
  • retention issues,
  • losing candidates,
  • products getting obsoleted by better models,
  • fundraising rejection.

Why aggression matters

• He believes AI companies need to be extremely aggressive in order to keep up with the pace of change.
• Venture-backed businesses must accept higher failure risk in exchange for a shot at becoming enormous.

Internal AI usage is mandatory

• Rogo tracks internal AI usage across the company.
• Employees are ranked by usage.
• The firm uses internal tools heavily to:
  • preserve knowledge,
  • improve onboarding,
  • support sales enablement,
  • surface prior work across the company.

The “company brain”

• Rogo’s internal knowledge system is humorously named “Shrek.”
• It stores:
  • company conversations,
  • goals,
  • metrics,
  • prior customer learnings,
  • context for sales and product work.
• It functions both reactively and proactively, surfacing relevant context for employees before meetings or projects.

What the Future Could Look Like

AI-native capital markets

• Gabe believes capital markets will become more AI-native over time.
• Possible outcomes include:
  • faster deal execution,
  • easier capital raising for businesses,
  • more automated asset pricing,
  • agents negotiating and coordinating transactions,
  • more liquid and standardized private markets.

The long-term shift

• Today, most investment firms are still mostly people and relationships.
• Gabe thinks the future will be more like:
  • software,
  • data,
  • systems,
  • AI agents embedded into firm infrastructure.

The biggest uncertainty

• The pace of adoption in private markets.
• The amount of remaining alpha in human relationships.
• How quickly firms and regulators will accept more automation in sensitive workflows.

Advice for Founders and Investors

For founders

• Be direct about what you’re worried about.
• Maintain conviction about the end state.
• Repeatedly revisit the roadmap.
• Expect rejection and iterate through it.
• The best founders are persistent, transparent, and willing to rebuild.

For operators in finance

• Identify where your latent knowledge lives.
• Ask which parts of your firm are truly defensible versus just human habit.
• Convert the knowledge of top performers into systems the firm owns.
• Decide where AI can augment judgment and where it can replace manual work entirely.

For investors

• Look for industries with:
  • enough complexity,
  • real workflow pain,
  • strong distribution opportunities,
  • and room for deep infrastructure.
• Favor founders who understand the domain and are willing to keep rebuilding.

Notable Takeaways

• The biggest moat in vertical AI is often not the model, but the workflow layer around it.
• Finance is especially attractive because so much of it is still manual, regulated, and fragmented.
• The future of AI in finance is likely to move from copilots to autopilots to agent-to-agent transaction infrastructure.
• Rogo is positioning itself to be more than a tool: it wants to become core infrastructure for capital markets.
• Gabe’s central belief: if you can make capital markets faster, smarter, and more accessible, you can unlock a huge amount of economic value.

Closing Thought

This episode is a strong case for why applied AI may create some of the most valuable companies in finance. Gabe Stengel’s view is that the winning company won’t just have the best model—it will own the entire system around the model: data, memory, workflow, compliance, and transaction rails. That, more than raw intelligence alone, is what he thinks will define the next era of investing software.