How the Best Firms Will Implement AI Over the Next 1-3 Years

Why did Emblem write this thesis?

After working with dozens of institutional investment firms, Emblem identified a consistent pattern: most firms know AI will transform their operations, but few have a clear framework for how to implement it. This thesis distills what we've learned into a practical roadmap for firms at any stage of AI adoption. The full thesis is also available as a downloadable PDF.

Where is AI adoption today in institutional investing?

Most institutional investors are in the "experimentation" phase — individual team members use ChatGPT or similar tools for ad-hoc tasks like summarizing documents or drafting emails. But ad-hoc usage doesn't scale and introduces accuracy and security risks. The firms pulling ahead are the ones embedding AI directly into their core workflows: deal sourcing, due diligence, portfolio monitoring, and LP reporting.

What is the 3-phase framework?

Emblem's framework identifies three phases of AI maturity for investment firms. Phase 1 is augmentation — using AI to speed up existing manual workflows (e.g., faster document review). Phase 2 is automation — replacing entire manual steps with AI (e.g., auto-generating financial models from CIMs). Phase 3 is intelligence — AI proactively surfacing insights, risks, and opportunities that humans would miss (e.g., real-time portfolio alerts based on market data and company filings). Most firms are in Phase 1. The leaders are entering Phase 2.

• Phase 1 — Augmentation: AI assists human workflows (summarization, search, drafting)
• Phase 2 — Automation: AI replaces manual steps end-to-end (model generation, report building)
• Phase 3 — Intelligence: AI proactively surfaces insights and risks (alerts, pattern detection)

What role does RAG play?

Retrieval-augmented generation (RAG) is the key technology enabling accurate AI in investment workflows. Unlike general-purpose LLMs that generate responses from training data, RAG grounds every output in your firm's actual documents. This means a generated financial model pulls numbers from the CIM you uploaded, not from the AI's general knowledge. Source tracing — the ability to click any output and trace it back to the source page — is what makes AI trustworthy enough for institutional use. Emblem's RAG pipeline delivers 100% source-traced outputs.

What should firms prioritize first?

Start with the workflows that consume the most analyst time and have the highest error rates. For most firms, that's due diligence document review and portfolio data collection. These are high-volume, repetitive tasks where AI delivers immediate ROI. Once diligence and monitoring are automated, extend to model generation, deck building, and LP reporting. The key is to choose a platform that integrates with your existing tools (CRM, cloud storage) so there's no migration friction.

Download the full thesis

Read the complete thesis including detailed implementation timelines, technology selection criteria, and case examples. Download the PDF or continue reading on this page.