Skip to main content
RealQuant Labs
Our stack

Databricks in your Azure, Claude on top.

Claude reads your documents. Databricks holds the data. Your team works in Excel and apps we build.

All services
Deployed stack
Your Azure tenantEntra ID sign-in

Claude

Databricks

Microsoft Azure

Excel

Custom apps

SharePoint and Teams

  • You own your data, outputs and the code we write for you. Our framework is licensed to you.

  • Models run under commercial terms that exclude training on your data

Reference architecture

How the stack fits together.

  1. Your team

    Excel and custom apps

    • Excel models that read the warehouse
    • Custom apps for your workflows
    • Teams and Outlook alerts

  2. AI agents and models

    Claude and code

    • Claude reads and drafts
    • Code calculates
    • A person approves

  3. Governed warehouse

    Azure Databricks

    • Delta Lake tables
    • Unity Catalog permissions and lineage
    • One set of KPI definitions

  4. Ingestion and parsing

    Scheduled and on demand

    • Exports and APIs
    • OMs, rent rolls and P&Ls parsed with citations

  5. Data sources

    Your systems

    • Yardi
    • MRI
    • Fund admin reports
    • Excel

Components

What each part does.

Claude

Answers that cite the file, page or cell.

Claude reads documents and drafts answers, each with its source.

  • Works in Claude and Claude for Excel
  • Read-only by default, as the signed-in user
  • Human review before anything reaches IC
Claude agents

Access

Read-only

Answers

Cited

Review

Gated

  • OM intake

    Page citations

    Review queue

  • Rent roll intake

    Unit level

    Review queue

  • IC memo draft

    Cited to source

    On demand

  • Variance commentary

    Reviewed

    Monthly

Databricks

One lakehouse for property and fund data.

Figures are calculated in code on this layer, and each one traces to its source.

  • Permissions, row-level access and lineage in Unity Catalog
  • Feeds from Yardi, MRI, fund admin reports and Excel
Databricks lakehouse

Tables

47

Refresh

Scheduled

Access

Row-level

  • deals_canonical

    12,847 rows

    Governed

  • rent_rolls_unit_level

    184k rows

    Governed

  • operating_statements

    38k rows

    Governed

  • market_comps

    92k rows

    Joined

Microsoft Azure

Your data platform runs in your Azure tenant.

Cloud, Databricks and model usage bill to your own accounts.

  • Entra ID sign-in and your existing permissions
  • Secrets in your Key Vault, behind your network rules
  • Infrastructure as code you can review
Your Azure tenant

Sign-in

Entra ID

Secrets

Key Vault

Deploy

As code

  • Azure Databricks

    Lakehouse, Unity Catalog

    Running

  • Container Apps

    Apps and agents

    Running

  • Blob Storage

    Source files, raw exports

    Private

  • Key Vault

    Secrets and keys

    Private

Excel and custom apps

Results land where your team works.

Memos save to SharePoint, and alerts arrive in Teams or Outlook.

  • Excel models read from the lakehouse
  • Custom apps on the same KPI definitions
  • Row-level security by fund, LP or region
Portfolio apps

Reports

18

Refresh

Daily

Sign-in

SSO

  • Portfolio NOI variance

    12 properties, year over year

    Daily

  • Occupancy trend

    Multifamily, 24 months

    Daily

  • Deal pipeline app

    Stage and IC date

    Live

  • LP reporting pack

    Saved to SharePoint

    Quarterly

How it works

Four stages, each priced separately.

01

Diagnostic

We review your Azure footprint and data, then quote the Build.

02

Build

We set up the stack in your tenant. Target: first workflow live in 90 days.

03

Scale

Add workflows on the same lakehouse, each scoped in its own SOW.

04

Operate

Optional. Engineers keep the stack running and current.

Questions

Your inputs are long documents: OMs, loan agreements, LPAs. Claude reads them and answers with a citation to the page.

Your Azure Databricks workspace sits in your Azure subscription. Unity Catalog sets permissions and traces lineage in one place.

We build on Databricks in your Azure tenant. If you already run Snowflake or Fabric, the Diagnostic sets how the two connect.

Cloud and model usage bill to your own accounts. The Diagnostic is a fixed fee and the Build a set fee. Later workflows are time and materials, scoped in writing. Operate is optional and billed monthly.

Find your first workflow.

Bring one workflow. In 30 minutes with the founder, we will tell you whether it is worth building.