A Whitepaper by Nadi Ulhaque · Founder, REMI

The Institutional
Intelligence Thesis

Why the next decade of real estate investment belongs to firms that compound intelligence — not headcount.

12 min read·March 2026
Scroll to read
01The Blind Spot

Your best people only use a fraction of the data available to them. That's not a talent problem. It's a human one.

I spent years as a data product person inside financial institutions. The stakeholders I served were smart. Capable. Experienced. But they only used a subset of the data that was actually available to them.

Not because they were lazy. Because their methods were ingrained. Their workflows were habits. The way they screened deals, built models, and reported to LPs was the way they'd always done it.

AI has no such bias. It doesn't have a "usual way" of doing things. It goes the route that's most accurate. And it handles exponentially more data than any human ever could.

That's the blind spot. Not in the data. In the habits.

0+

Deals screened per fund per year — but dozens more never make it past the inbox.

Source: Industry benchmarks

0%

Average close rate on screened deals. 97% of the work is filtering.

Source: CBRE / JLL

02The Centaur Model

AI doesn't replace your team. It makes every person on it 10x more dangerous.

The centaur model comes from chess. After Deep Blue, the best players weren't humans or machines — they were humans working with machines. The combination crushed both.

AI Handles

Data ingestion at scale

Pattern matching across 1000s of deals

24/7 monitoring

First-draft analysis

Historical context retrieval

Centaur Output

Machine speed. Human judgment. Institutional context.

Humans Own

Final investment decisions

Relationship management

Regulatory sign-off

Strategic overrides

LP communications

In a regulated environment like asset management, human oversight isn't a limitation — it's a feature. The firms that win will be the ones where AI handles throughput and humans handle judgment.

03The Context Graph — Our Moat

REMI doesn't just process your data. It remembers everything your firm has ever done.

Most AI tools start from zero every session. You paste in context. You explain your firm. You re-teach it the same things over and over.

REMI's context graph is a persistent, evolving map of your firm's entire institutional memory — every deal screened, every decision made, every override, every LP preference.

It fits each firm like a glove. What's feeding REMI is your data. What REMI outputs are faster, more accurate processes based on that data. And it gets smarter every single day.

Compound Intelligence

Every deal screened, every memo written, every correction made feeds back into the graph. REMI doesn't just learn — it compounds.

Firm-Specific Memory

Your buy-box. Your thesis. Your LP preferences. Your risk tolerance. REMI knows your firm the way a 10-year veteran does — except it never leaves.

Historical Replay

Query past deals. Compare current opportunities against historical comps from your own portfolio. Simulate outcomes based on what actually happened.

04Why Now

Three forces are converging. The window is open. It won't stay open forever.

AI Infrastructure Matured

Context windows went from 4K tokens to 200K+. Agent frameworks emerged. Reasoning models can now follow multi-step financial logic without hallucinating. This wasn't possible 18 months ago.

4K → 200K+

Context window growth

Cost Collapsed

The cost per token has dropped roughly 100x since early 2023. What used to be economically insane — running AI agents across every deal in your pipeline — is now viable at scale.

~100x

Cost reduction since 2023

LP Pressure Intensified

Higher rates. Compressed timelines. LPs demanding more transparency and faster reporting. Firms can't hire their way out of this — the talent market is too tight and the margins don't support it.

2–3 wks

Avg. quarterly LP report time

05The Market

A $14.6 trillion industry. Still running on spreadsheets.

Alternative assets under management have exploded. The operational infrastructure hasn't kept up. That's the gap.

$0.0T

Global alternative assets under management.

Source: Preqin 2024

0K+

Mid-market RE and PE firms globally.

Source: Industry estimates

$0.0T

Projected alt asset AUM by 2028.

Source: Preqin forecast

$0K

Average annual cost per departing analyst — institutional knowledge included.

Source: Robert Half 2024

06How It Works

Four workflows. One platform. Human-in-the-loop at every stage.

REMI isn't a black box. Every output has a source. Every decision has a checkpoint. Your team stays in control — they just move faster.

Step 01

Deal Screening

Ranked pipeline with risk scores, comp overlays, and thesis-alignment flags.

Agent Action

Screening agents ingest OMs, comps, and market data. Every deal is scored against your buy-box — not a generic one.

Human Checkpoint

Your team reviews flagged opportunities. Override, approve, or adjust the scoring criteria. REMI learns from each decision.

Step 02

Due Diligence

Complete diligence packages with source citations and variance analysis.

Agent Action

Diligence agents abstract leases, cross-reference rent rolls, flag inconsistencies, and pull historical context from your prior deals.

Human Checkpoint

Analysts review abstractions and approve findings. Corrections feed back into the context graph.

Step 03

IC Memos

Institutional-grade IC memos generated in hours, not weeks.

Agent Action

Memo agents assemble investment committee packages that cite your prior decisions, not just market data. They reference why you passed on similar deals last quarter.

Human Checkpoint

Investment leads review, edit, and sign off. The memo reflects human judgment, accelerated by machine throughput.

Step 04

LP Reporting

Polished LP reports delivered in days — with full audit trail.

Agent Action

Reporting agents pull portfolio data, generate commentary, and compile quarterly narratives that match your LP communication style.

Human Checkpoint

Portfolio managers review and finalize. No more last-two-weeks-of-the-quarter fire drills.

07The Endgame

The firms that win the next decade won't be the biggest. They'll be the ones that remember everything.

Think about what happens when a firm has been running REMI for two years. Every deal screened. Every decision logged. Every outcome tracked. Every LP preference internalized.

That firm can screen a new deal in minutes — and score it against two years of compounding institutional memory. Not market data. Their data. Their patterns. Their thesis.

That's not an incremental improvement. That's a structural advantage that compounds over time. And it's one that a competitor can't replicate by hiring more analysts.

"The best firms don't just have better processes — they have better memory. We're building the memory layer for institutional real estate."

Nadi Ulhaque

Founder & CEO, REMI

Key Takeaways

AI in institutional finance is the greatest tool to multiply the power, output, and quality of people.

Human-in-the-loop in a regulated environment is a feature, not a limitation.

The context graph — firm-specific, compounding, irreplaceable — is the moat.

Context windows, cost collapse, and LP pressure converged. The window is now.

The firms that start compounding intelligence today will be unreachable in 3 years.

REMI is built by operators who lived the problem. Not consultants who studied it.

Your firm learns once. REMI remembers forever.

Ready to transform your operations?