REMI
1 / 11
1/11

REMI

The Intelligence Layer
for Real Estate Capital

Pre-SeedQ1 2026|chatremi.com
STARTING WITH CREDIBILITY

My Journey Here

2018

Fairview Capital Partners

Private EquityFund-of-Funds

Supported the onboarding of the firm's largest commitment (PA-SERS) at this $3.3B PE fund-of-funds — first exposure to institutional capital workflows.

2019

IBM Watson AI + Global Financing

Data AnalystBusiness AnalystAI

Led 0-1 integration of Watson AI into IBM’s largest deal-making platform.

2021

Point72 Asset Management + New York Mets

Data EngineeringData ProductAsset ManagementAI/ML

Built multiple data platforms from 0-1 ranging from crypto to baseball. Featured in WSJ, NY Times, & more.

2024

SMBC / Jenius Bank

Data EngineeringData ProductCredit RiskOperationsDecision EngineData EncryptionData Security

Helped SMBC build & scale data platform for their first Digital Banking Unit.

2026

REMI

Building the solution I wished existed — now with AI capable enough to make it real.

Nadi Haque — Founder

The Problem

AI Lives in Spreadsheets. But Firms Live in Context.

Today's AI tools are powerful, but nerfed. They operate at the cell level, not the firm level. Investment Firms have ripple effects — deals tied to asset(s) in a portfolio in a fund.

🧮

Just a Faster Calculator

AI's working 40/168 hours

Dependent on human input, firms are not capitalizing on the sheer volume of work AI can do.

🧠

Zero Continuity

2.3 year avg analyst tenure

Every session starts from scratch. Firm history, IC preferences — gone.

Still Manual

4-8 hrs per lease

63% of RE teams say unstructured files are their #1 bottleneck.

📊
📧
📄
🤖
💬

🧠 The Analyst

is still the integration layer

IC Memo / Model / Report

Funds hedge Market Risk but remain fully exposed to Key-Person Risk.

Bad hiring decisions are often made & key employees leave regularly. These drop your Fund's IQ significantly.

Current Solutions

Two Approaches. Same Dead End.

Headcount Approach

Scaling headcount is expensive & takes >1Y for positive ROI.
In the AI-age, this approach is outdated.

Mid-market RE is 3× less efficient than PE/Hedge Funds.

Generalist AI Approach

GPT & Claude are great, but lack context. Once everyone has Claude running, the novelty will wear off & real alpha will be chased.

Works 40 hours/week

Copilot & Claude are great - but they only work when the analyst does

Understands Excel, not your assets

Faster DCFs don't equate to better deals & higher IRR

Starts from 0 every session

No compounding knowledge, no firm memory

The memory of decisions made is stuck in human heads. An individual gets wiser, but the firm remains the same.

What if instead…

Every morning started at the finish line?

Let AI do what you were already going to tell it to do, preemptively. In humans, we call that taking initiative. All you have to do is review.

Emails
PDFs
Feeds
IC Memos
Models
Insights
AI takes initiative and prepares cited drafts before you open your inbox.
Red and green flags are surfaced preemptively, with clear reasoning attached.
Compare an incoming deal against similar historical deals from your own firm.
Meet REMI

AI Agents That Know
Your Entire Firm.

REMI maps historical and incoming data to build a living Context Graph for your entire firm, then deploys autonomous agents that work 24/7 across new and existing deals.

Context GraphEvery fact linked
Recursive LearningGets smarter
Autonomous Agents168 hrs/week
New Deal to Draft Agent
5 minutesfrom email to draft

Email Arrives

Deal flow from broker

PDFs Extracted

OM, financials, rent rolls

Agent

Context Graph

Link to asset & fund history

CG

Compare Past Deals

Match against your firm's deal history

RL

Apply Fund Rules

IC criteria, thesis filters

AgentFlag

Draft Generated

IC memo + model ready

Human Review

Your decision, your control

Pursue
Diligence Agent
Pass
Archive + Learn
Watchlist
Monitor Agent
Running 24/7 · 168 hours/week · Human in the loop

Market Opportunity (Initial Wedge)

$25B+ Total Addressable Market

Most databases only track ~10K institutional firms. The real opportunity is the invisible majority — 90K+ boutiques, syndicators, and family offices desperate for institutional-grade tools.

Visible (Preqin Tracked)

~10K Firms

— waterline

~15K Firms

~35K Firms

30K+ Reg D/Yr

The Invisible Market

Segment

Firms

TAM

Enterprise & Global Giants

~500

$125M+

Core: Mid-Market Institutions

~8,000

$800M

Active Capital Raisers (Reg D)

~35,000+

$1.05B+

Adjacent: Private Credit

~4,000

$480M

Adjacent: Infrastructure & PE

~6,000

$900M+

Real Estate TAM

~53K+

$3.4B+

All Private Capital

$25B+

Same Product, Two Motivations

Institutions buy for amplification (multiply Analyst output without increasing hiring).
The Invisible Market buys for capability (leverage data in a way that would normally take an entire data team).

Sources: Preqin 2025, SEC Reg D Filings (Annual), AppFolio 10-K, IRS Partnership Data

Why Now

The AI Wave is Here

Coding was the first knowledge-based skill to fall to AI. With AI Agents unlocked, operational knowledge-work & recurring workflows are next.

Coding Assistant Adoption

AI Workers Projection

Benjamin Miller, CEO Fundrise
Year% AutomatedFTE Equiv.
20263%1.7M
20278.6%4.9M
202816.8%9.6M
202927.1%15.6M
203035.4%20.4M

By 2030: 35.4% of all US knowledge-work hours automated — equivalent to 20.4M FTE workers.

Operations is next.

Sources: GitHub Octoverse Report (2024–2025), Sequoia Capital "State of AI" Reports, Anthropic Public Metrics

Why Me

Built for This Exact Problem

The way to solve enterprise AI is at its root — which is data. I've spent my entire career navigating enterprise data platforms at financial institutions & managing stakeholders of downstream data usage. I know how they work, their problems, and the alpha that exists by building REMI.

Data-First AI Platforms

Led 0→1 Watson AI integration at IBM's largest deal-making platform. Built multiple data platforms from scratch at Point72. This is what REMI requires: complex, multi-source data expertise, with context graphs & AI on top.

Enterprise Financial Systems

Helped SMBC build & scale their first Digital Banking Unit serving key business members in Credit Risk, Fraud, & Operations. Know the compliance, security, and procurement cycles.

Data Security & Org Ops

I’ve handled sensitive financial data across institutions: encrypted, proprietary, and operationally critical. I know how teams actually operate, and how to layer AI into existing workflows without forcing organizations to flip overnight.

Builder DNA

Every role has been a 0→1 platform build inside complex organizations. Left corporate because I've seen enough progress with AI-accelerated development to bet on myself — building REMI full-time is the next step, not another middle management role.

Why Not

Risks I'm Thinking About

You're going to ask these questions. Let me address them head-on.

AI is commoditizing fast. What's your moat?

Context Graphs & Data Governance are our moat. Generic AI can't understand deal relationships without fund-specific knowledge graphs. We're building the data layer, not just the AI layer.

Addressed

What about data security and integration complexity?

Deployment model is client-driven: VPC/on-prem when required, or managed cloud when preferred. Either way, enterprise-grade access controls, encryption, and audit trails are standard.

Addressed

What about the incumbents (Yardi, MRI)?

They may create their own AI, but that's no competition to us. We intertwine all the systems a fund uses — and that's fundamentally out of scope for incumbents. At worst, they'll create an MCP that helps REMI.

Addressed

Single founder risk?

There is too much leverage with tools like Clawdbot, Claude, and Cursor to rush this decision. I’m technical, and my door-to-door sales background gives early GTM coverage. I’m intentionally selective about a long-term co-founder fit.

Actively Managing

How will you access data from legacy systems?

Many legacy platforms were built as systems of record, not systems of understanding. REMI sits above them — ingesting outputs, context, and decisions to power faster analysis without disrupting existing infrastructure.

Addressed

Why not just buy an enterprise AI license and build internally?

Who maintains it when the best model changes every 90 days? Who migrates from Feb 2026 models to Dec 2026 models that are 10× better and cheaper? REMI abstracts that complexity — we handle model orchestration, versioning, and firm-specific tuning so your team never has to.

Addressed

Let's build the Intelligence Layer together.

Current Status — Phase 0 & 1

Core platform 70% done, QA testing heavily

- Evaluating AI Tagging/Labeling solutions

Demos are ready
Actively working on LOIs; founder-led outreach underway
Targeting 3-5 design partners for Q1 pilot

Pre-Seed Round

$1.2M – $1.5M

Open to SAFE or priced round

Targeting 18-month runway

Engineering42%
Data/ML18%
Cloud12%
SaaS Tools3%
HiTL Ops (PH)4%
Conferences2%
Marketing3%
Founder3%
Security9%
Legal/Ops4%

Additional Materials

Appendix

Deep dives on business model, platform architecture, learning loops, and risk analysis.

VS Generalist AICompetitive LandscapeBusiness ModelREMI Platform StackTech ArchitectureLearning LoopGTM StrategyRisk Deep DiveTrust GapFounder’s Thoughts

VS Generalist AI

Claude is Stateless. REMI Has Memory.

General AI is powerful infrastructure. But institutional RE needs a memory layer that compounds with each deal, not a blank slate every session.

Dimension

Generalist AI

REMI

Memory

Session ends, context resets.

Persistent Context Graph compounds each deal cycle.

Workflow

Prompt-by-prompt, human-driven.

Pre-wired RE pipeline with human checkpoints.

Output Quality

Generic answer quality.

Firm-specific outputs with citations and rationale.

Learning

No institutional feedback memory.

Recursive learning from decisions and outcomes.

REMI will use a variety of different LLM's as core model infrastructure. The Context Graph is the product moat that turns general models into institutional RE intelligence.

REMI is not hands-off automation - it is decision support with full auditability. We prioritize transparency over opacity. Every output is cited, every action is logged, every decision is yours.

Competitive Landscape

The Market is Ripe for Disruption

Yardi makes $1.6B/year on 40-year-old architecture. Point solutions lack context. General AI lacks firm memory. REMI is built AI-first.

Yardi / MRI

Legacy ERP

+ Strength

Market leader, $1.6B revenue, 20K customers

- Weakness

40-year-old architecture, AI bolted on

AI: Bolted-on

Rogo AI / o11

Point Solution

+ Strength

IB-focused agents (Rogo, $75M Series C); M365-native modeling (o11)

- Weakness

Built for banking, not RE ops — no property-level context, no deal lifecycle memory, no autonomous workflows

AI: Limited

ChatGPT / Claude

General AI

+ Strength

Powerful, accessible

- Weakness

No firm memory, no integration

AI: Genericlow trust

REMI

AI-Native Platform

+ Strength

Context Graph, autonomous agents

- Weakness

Early stage, unproven at scale

AI: Purpose-built
Dimension
Legacy ERP
Point Solutions
Generalist AI
REMI
Firm Context
Autonomous Agents
Cross-Deal Learning
24/7 Processing
Human in Loop
Enterprise Ready
🔨

Business Model

Intelligence-as-a-Service Pricing

Platform fee based on fund size + metered agent usage. Revenue scales with client success — more deals = more value = higher contracts.

Platform Fee (Annual)

< $100M AUMTBD
$100M–500M AUM$75K/yr
$500M–1B AUM$100K/yr
$1B–5B AUM$150K/yr

Includes: Context Graph, integrations, firm knowledge base, security infrastructure

Usage Based Pricing

Deal ScreeningMetered
IC Memo GenerationMetered
LP Report AssemblyMetered
Portfolio MonitoringIncluded

Why This Model Works

Revenue scales with client success. More deals processed = more value delivered = higher contract value. Perfectly aligned incentives.

No per-seat pricing — scales with deal velocity, not headcount
Platform fee covers infrastructure; usage grows with value
Net revenue retention target: 130%+ as firms expand usage

REMI Platform Stack

A purpose-built operating system for institutional RE intelligence

REMI is not a single model wrapper. It is a layered platform that combines firm ontology, context graph memory, governed agent execution, and client-facing workflows.

Prebuilt AI Products

Deal Screening, IC Memos, Portfolio Monitoring, LP Reporting

Custom AI Products

Firm-specific workflows, rubrics, approvals, and reporting formats

Build with REMI

Context Graph + Quick Memory

Deal → Asset → Portfolio → Fund relationships + decision memory

Data Services

Ingestion, parsing, entity resolution, retrieval, and lineage

AI Services

Model orchestration, agent planning, evaluation, and recursive learning

Workflow Services

Deliverables, approvals, notifications, and operating actions

Security + Governance Layer

Role-based access, audit trails, policy controls, and deployment optionality

Software Delivery Layer

Web app, APIs, integrations, and productized workflows for each client

Technical Architecture

Context Graph + Multi-Agent Intelligence

REMI is built as an institutional intelligence system, not a chat wrapper. Data is ingested, normalized, linked in a firm-level graph, reasoned over by specialized agents, and returned with citations and controls.

Data Ingestion + Parsing

  • Email/API/webhook ingestion (deal flow + updates)
  • Document pipeline: OCR + table extraction + chunking
  • Entity normalization (asset, sponsor, market, tenant, covenant)

Context + Knowledge Layer

  • Context Graph (deal -> asset -> portfolio -> fund)
  • Hybrid retrieval (graph traversal + vector search)
  • Firm ontology + IC rubric + thesis constraints

Reasoning + Agent Layer

  • Multi-LLM routing by task (extract, reason, draft, verify)
  • Agent orchestration (screening, compare, memo, monitor)
  • Citations + confidence scoring + policy checks

Application + Workflow Layer

  • New Deal -> Draft workflow with human checkpoints
  • Review UI (approve, revise, reject, watchlist)
  • Feedback loop writes outcomes back to graph

Security + Governance Layer

  • RBAC + tenant isolation + encrypted storage
  • Audit logs for prompts, actions, and approvals
  • Deployment flexibility: cloud, VPC, or on-prem constraints

Moat: The Context Graph compounds with every approved output, every exception, and every deal outcome. Features are copyable. Institutional memory with governance is not.

REMI Learning Loop

Validating Centaur Systems Theory for real estate investment workflows

Centaur systems combine human intuition, context, and ethics with algorithmic speed and scale. REMI is built to test that strong merge in live investment operations: AI proposes at scale, humans apply judgment, and each interaction compounds firm intelligence.

How REMI compounds

1) Ingest Context

Deals, docs, market data, and prior firm decisions

2) Agent Drafts

REMI proposes recommendations with source citations

3) Human Judgment

Analyst/IC review, approve, reject, or override

4) Outcome Capture

Final decisions and downstream performance are recorded

5) Memory Update

Context Graph learns patterns to improve next-cycle quality

Validation Objectives

Increase decision consistency across analysts and associates

Reduce cycle time from intake to committee-ready materials

Improve early risk signal detection from prior deal memory

Raise trust via citations, approvals, and auditability

References: human-AI collaborative decision-making research framing (Springer, MIT HDSR).

Go-To-Market

Bottoms-Up, Then Enterprise

REMI creates value whether users touch it or not. We prove value early through targeted, proactive notifications tied to live deals, then drive bottoms-up adoption with analysts and associates before broader organizational rollout.

Phase 1: Design Partners

Now → Month 6
  • 3-5 mid-market RE firms
  • Free pilots, high-touch onboarding
  • Overinvest in early partner success to create raving advocates in a network-driven niche

Target

Target: first 5 LOIs

Phase 2: Early Revenue

Month 6 → Month 12
  • Convert pilots to paid
  • Content marketing + thought leadership
  • Referral program launch

Target

15-25 clients onboarded

Phase 3: Scale

Month 12 → Month 24
  • Multi-channel acquisition engine
  • Repeatable onboarding and account expansion
  • Expand into adjacent alternative asset workflows

Target

Target: 100+ clients across 2-3 Alt Asset Classes

35%

Shadow Mode Pilots

Zero-risk 90-day pilots — REMI runs parallel to existing ops, proves value before any workflow changes

25%

Founder-Led Outbound

Direct outreach to mid-market GPs/analysts via LinkedIn, warm intros, and prior network

20%

Content & Thought Leadership

Case studies from pilot partners, founder articles, RE-specific AI insights

10%

Partnerships & Referrals

Fund administrators, RE accountants, and consultants as channel partners

10%

Events & Conferences

NMHC, IMN, CRE Finance Council — targeted RE industry presence

Adoption Strategy

Zero-Risk Adoption - Meet Firms Where They Are

For The Analyst

Career Protection

REMI catches what you might miss. You approve everything. You stay the hero, not the system.

For The Principal

Institutional Memory

Never lose firm knowledge when people leave. Preserve the why behind every decision.

For Operations

Zero-Risk Entry

REMI watches, learns, and waits. Nothing changes until your team is ready.

Read-only integrations first - REMI observes, does not write

Shadow mode - See what REMI would have caught, risk-free

Graceful exit - If REMI disappears, nothing breaks

Human approves everything - Always

The question is not "Will this replace my team?" It is "What am I missing right now that REMI would catch?"

Pilot Structure

90-Day Proof of Value - No Contract Required

Start risk-free, demonstrate value in production conditions, then scale with confidence.

Step 1

Connect

Read-only integrations for email, Yardi, and shared drives.

Outcome: No workflow disruption

Step 2

Shadow Mode

REMI runs silently and shows what it would have flagged earlier.

Outcome: Proof before trust

Step 3

Activate

Analysts use REMI drafts for IC memos with full human approval.

Outcome: Before vs after gains

Step 4

Prove

Review throughput, quality, and decision speed metrics together.

Outcome: Conversion decision

If it works, they do not turn it off. If it does not, we learn why and improve the system. Either outcome compounds value.

Risk Analysis

Eyes Wide Open: Risks & Mitigations

Every startup has risks. Here's how I'm thinking about ours — and what I'm doing about them.

AI Commoditization

Low

LLMs become commodities; features get copied quickly.

Our moat is not AI — it's the Context Graph. No AI company is building firm-specific knowledge graphs for each organization. The graph compounds from approvals, overrides, and outcomes. Competitors can copy features, but not accumulated intelligence.

✓ Action: Building proprietary data flywheel from day one

Data Security Concerns

Medium

RE firms are paranoid about data leaving their walls

SOC 2 readiness and certification program on roadmap. Enterprise deployment options (VPC, on-prem). Zero data sharing between firms. Enterprise-grade encryption.

✓ Action: Security audit Q3 2026; compliance certifications Q1 2027

Incumbent Response

Low

Yardi/MRI bolt on AI features

Incumbents are burdened by technical debt. Their AI is a feature; ours is the architecture. They optimize existing workflows; we reimagine them.

✓ Action: Move fast, lock in design partners before incumbents react

The Trust Gap

You Wouldn't Let AI File Your Taxes. Why Would You Let It Deploy $100M?

Even TurboTax asks you to review before filing. Deal-making at scale requires the same rigor — every number cited, every decision approved, every action auditable.

What People Fear

AI makes a decision I don't understand

Numbers appear from nowhere — no audit trail

I lose control of my own process

"Trust the AI" isn't good enough for LPs

“I need to explain this to my IC. 'The AI said so' doesn't cut it.”

How REMI Works

C
CITE: Every number links to source (page 42, cell B7)
R
REVIEW: AI proposes, you dispose — nothing ships without approval
F
FEEL: Full visibility into what's running and why

“Show me exactly where that 12% IRR came from.” → Instant citation.

REMI is built for the Reviewer, not just the Runner. AI handles the 80% grunt work. Your team does the 20% expert judgment that actually matters.

Founder's Thoughts

AI speed is the novelty. Context is the moat.

February 2026 feels like peak wow-factor for AI. That phase will fade. AI in spreadsheets and workflows will become table stakes, and firms that do not adopt will look ancient to the next generation of operators.

Three lines to remember

1) Automation without citations kills trust.

2) Context is alpha: data + decision history + operator feedback.

3) The end state is better reviewed decisions, not zero human decisions.