About
An operating studio for agentic software

Real-world productsbuilt with agentic practice

Midwest builds and operates a portfolio of software products where agents are not decoration. They are part of the research, interface, workflow, and operating model.

01

Healthcare data infrastructure that turns legacy HL7 v2 and CDA/CCDA records into standardized FHIR R4 and OMOP CDM 5.4 data — automatically, at network scale.

Built for Health Information Exchanges and hospital networks, Currently replaces months of bespoke integration work with a self-serve SaaS pipeline that resolves terminology against 6.3M clinical concepts. It is positioned at the intersection of federal FHIR mandates (ONC 21st Century Cures, CMS-0057-F) and surging demand for OMOP-formatted real-world evidence from payers, pharma, and clinical AI.

Health Information ExchangesHL7 / CDA → FHIR R4 + OMOP CDM 5.4Self-serve clinical data pipelineBuilt — Seeking sales
02

ZoomProp

zoomprop.com

An AI-native operating system for real estate investors — property discovery, financial underwriting, portfolio monitoring, and deal coordination in one conversational workspace.

Built for professional investors managing 1–500 units, ZoomProp replaces the patchwork of Zillow, spreadsheets, email, and CRMs with a single workspace that can analyze a cap rate, parse an inspection report, and track a co-investment in the same session. It attacks the structural fragmentation at the center of the $4.4T U.S. real estate investment market, where no single tool connects discovery, underwriting, and execution with AI reasoning.

AI-native deal intelligenceFull-stack investor workspaceDiscovery through co-investmentPortfolio company
03

An AI-native document intelligence platform that lets legal, financial, and compliance teams interrogate complex document sets in plain English and get cited, sourced answers in seconds.

DataZoom automates high-cost workflows like cap table reconstruction, due diligence checklists, and document timelines that currently consume $400–$800/hour of attorney time, and it can run fully on-premises so sensitive documents never leave a firm's own infrastructure. It targets the $15–20B intersection of enterprise AI adoption, M&A activity, and the automation of legal knowledge work, with a metered billing system live today and an e-signature module extending it from analysis into full transaction execution.

AI due diligence, cited answersCap table auto-reconstructionCloud or air-gapped deploymentBuilt — Seeking market validation
04

An all-in-one operating system for licensed cannabis dispensaries — point of sale, inventory, METRC compliance, and customer management in one cannabis-native platform.

CNBS replaces the piece-meal stack most dispensaries run today (generic POS plus a separate compliance portal plus spreadsheet inventory plus a bolt-on loyalty product) with one cannabis-native platform. For a buyer already in cannabis distribution, owning this dispensary-side operating layer expands multiples on the existing footprint — Midwest is open to a 70% partnership or outright acquisition.

All-in-one dispensary OSMETRC-native complianceOpen to 70% partnership / buy-outBuilt — Seeking partner
05

A product workspace for ZoomProp sales, workspace automation, and agent-assisted follow-up.

Workspace signalsSales agentsFollow-up loopsOnboarding

Principles

How we build.

01

Agents are operators

Agents hold memory, prioritize tasks, and move work forward without waiting to be asked. They are not assistants bolted onto existing workflows.

02

Products are the proof

Every system ships as real software with real users. Internal demos and proofs of concept do not count as product.

03

Memory is infrastructure

Context is not stored in chat history. It is modeled, persisted, and retrievable so that agents can reason across sessions and handoffs.

04

Humans close the loop

Agents route and propose. Operators sign off. Autonomy is earned incrementally, not assumed by default.