---
title: "Operator Playbook: The Cityblock Playbook — Building Healthcare Infrastructure Around the Patient"
canonical: "https://theupsidejournal.com/articles/cityblock-playbook-healthcare-operator-model/"
published: "2026-11-17"
column: "The Operator"
author: "Olusegun (Olu) Olufunmiloye"
subtitle: "Cityblock's lesson for other operators is a five-part operating sequence, and it pays only where a payer will contract on outcomes."
description: "Cityblock's transferable operator model: five principles from its CORE operating system, and why outcome-aligned contracts decide whether the sequence pays."
---

# Operator Playbook: The Cityblock Playbook — Building Healthcare Infrastructure Around the Patient

*Cityblock's lesson for other operators is a five-part operating sequence, and it pays only where a payer will contract on outcomes.*

## Executive Summary

Most coverage of Cityblock Health treats it as a healthcare company that happens to use technology. The operator reading is different: the company's value is a transferable operating sequence that any operator serving a high-need, institutionally fragmented population can study without entering healthcare. The Upside Journal's synthesis, drawn from the company's own account and independent trade reporting, identifies five repeatable principles: integrate care around one defined population, build a data layer that ranks the next best action for each member, let AI absorb routine engagement so staff time goes to judgement, align contracts with outcomes, and work through existing plans and institutions rather than replacing them.

The economic condition is the constraint that matters. The sequence pays only where a payer will contract on outcomes, because outcome-aligned contracts are what reward coordination capacity. The five principles are the Upside Journal's abstraction, not a framework published by Cityblock, and each is tied below to what the company says about itself or to what independent reporting records.

## Introduction: the model is not the sector

The lazy reading of a company like Cityblock is that its lessons stay inside healthcare, because its customers, regulators and clinical staff are specific to healthcare. That reading mistakes the sector for the mechanism. What an operator in logistics, housing finance or education can use is not the clinical content but the sequence of decisions that made a fragmented system governable.

The Upside Journal has already covered the founder in [Toyin Ajayi: Building the Healthcare Infrastructure Behind Medicaid](https://theupsidejournal.com/articles/toyin-ajayi-cityblock-medicaid-infrastructure/), the structural problem in [Why Medicaid Is a Healthcare Infrastructure Problem, Not Just a Coverage Problem](https://theupsidejournal.com/articles/medicaid-healthcare-infrastructure-not-coverage/), and the operating model in [Cityblock Health Is Medicaid Care Infrastructure, Not a Clinic Chain](https://theupsidejournal.com/articles/cityblock-health-medicaid-care-infrastructure/). This piece assumes all three and does one job: it abstracts the repeatable system, so an operator can decide what to copy on Monday morning.

## What does Cityblock say its operating system is?

The company's own account is the best primary evidence of how it describes itself. In a note dated 3 September 2026, published by Healthcare IT Today and originally announced on 20 August 2026, Cityblock's co-founder and chief executive writes of "CORE—our Care Orchestration and Resourcing Engine—the operating system underpinning our outcomes-based care platform". The same note describes "the same AI-native, outcomes-first model that has enabled us to win in Medicaid" and says that the model also fits rural Medicare Advantage and dual-eligible populations.

The note gives scale figures, which are the company's own statements and are not independently verified. Cityblock says it now serves almost 200,000 members across the country. It states annualised revenue of $2.2 billion, up 77% year over year, and a $116 million Series E led by General Catalyst. It also reports 44,599 clinician hours freed and 42,148 calls handled by AI agents this year.

The verdict: the company describes a platform with a named operating system and an explicit design rule, outcomes first, and that framing is the right place to begin abstraction.

## What does independent reporting add?

Fierce Healthcare's report on the same announcement is the independent trade source. It records that Cityblock signed a definitive agreement to acquire Homeward Health in an all-stock deal, that financial terms were not disclosed, that General Catalyst led the $116 million round, and that the company has raised more than $900 million in total. It states that Cityblock works with Medicaid plans and health systems in more than 10 states, and that Homeward covers 50 counties and has delivered care to more than 50,000 rural patients over four years. The article page shows no calendar date for the announcement beyond a reference to a Thursday, so the company's note is the dating source.

The reporting matters here for one reason: it separates what Cityblock says from what a trade publication records. The playbook below uses the company's statements for how the model is designed and the trade report for scale and structure. The acquisition itself is background, not the subject.

## Principle one: integrate care around one defined population

The first principle is to choose a population whose needs cut across the boundaries of the institutions that serve it, then organise the service around the population rather than around any single institution's product. Cityblock's stated approach is to build one model across Medicaid, rural Medicare Advantage and dual-eligible members, which are different payer categories with a common feature: high need and fragmented services.

The operator translation is a definition problem. Before any technology is chosen, the operator decides who the unit of service is, and the unit must be a person or household with a stable identity across systems, not a transaction. Where that unit is undefined, every later principle has nothing to attach to.

The verdict: the population definition is the foundation, and an operator who cannot name the unit of service cannot run the sequence.

## Principle two: build a data layer that ranks the next best action

The second principle is a decision layer, not a record layer. The company's note describes CORE as the engine underpinning the platform and as a system that ranks which care action is most likely to close a given member's care gap. The distinction is between data that describes what has happened and data that tells a worker what to do next for a specific person.

For a non-healthcare operator, the test is whether the data layer produces a prioritised list that a frontline worker can act on in the next hour. A layer that produces reports for managers is an analytics system. A layer that orders the day's work for each worker is an operating system. The Upside Journal's treatment of an operator who turned a data layer into infrastructure, [Operator Playbook: The Esusu Playbook — Turning Overlooked Data Into Financial Infrastructure](https://theupsidejournal.com/articles/esusu-playbook-overlooked-data-infrastructure/), is the best live comparator: its subject also had to convert an underused data asset into a function others depend on.

The verdict: the value of the data layer is the action it triggers, not the dashboard it feeds.

## Principle three: let AI absorb routine engagement

The third principle allocates scarce judgement. Cityblock reports that its AI agents handled 42,148 calls and that AI tools freed 44,599 clinician hours this year. These are company-reported figures and the note does not set out what the calls were or how the hours were measured, so they are evidence of the company's claim, not of an independently audited result.

The design rule behind them is clear enough to transfer. Routine, high-volume contact is handled by automated agents; the hours saved go to the cases that need professional judgement. An operator can apply the rule without reproducing the number: list the interactions in the service that are routine and repetitive, list the ones that need judgement, and measure the share of staff time spent on each.

There is a limit that the sequence imposes on itself. Automation of routine contact only pays if the freed capacity is redeployed to judgement work that changes outcomes. Where the saved hours are absorbed elsewhere, the cost falls but the outcome does not move, and the next principle has nothing to reward.

The verdict: AI in this sequence is a way to reallocate judgement, and its return depends on where the freed hours go.

## Principle four: align contracts with outcomes

The fourth principle is the one that determines whether the first three pay. The company describes its platform as outcomes-based, and the Upside Journal's reading is that this is the commercial hinge. If a payer pays per service delivered, an operator earns by delivering more services. If a payer pays against outcomes for a population, an operator earns by coordinating well, which is exactly the capacity the first three principles build.

This is the economic consequence that most summaries leave out. Coordination capacity has value only where the contract prices it. In a fee-for-service setting, the same data layer and the same automation produce a better experience and a worse margin, because the payment model does not recognise what was prevented.

For an operator outside healthcare, the Monday-morning question is therefore not whether to adopt the technology but whether any counterparty will contract on outcomes. If none will, the sequence is premature, and the sound first step is to find or create such a counterparty, or to accept that the model will not pay yet.

The verdict: outcome-aligned contracts are what convert coordination into revenue, so the sequence only pays where a payer will sign one.

## Principle five: work through existing plans and institutions

The fifth principle concerns distribution. Fierce Healthcare records that Cityblock works with Medicaid plans and health systems, and the company's note points to Homeward's ties with plans such as Blue Cross Blue Shield of Michigan.

A dated example is the partnership announced on 7 January 2026 by Cityblock and Meridian Health Plan of Illinois, a company of Centene Corporation, to bring community-based care to 10,000 Medicaid beneficiaries in 17 Illinois counties: the release states that Meridian's existing primary care provider network supplies primary care while Cityblock adds wraparound services, and describes it as the fourth regional partnership between Centene and Cityblock.

The pattern is to sell to, and operate alongside, the institutions that already hold the member relationship and the payment, rather than building a competing channel.

The operator translation is a sequencing choice. An entrant in a fragmented system can try to replace the incumbent institutions or can become the layer that makes them work together. The second route accepts slower control in exchange for access to the existing flow of members and money. A second playbook that turned on a regulator-dependent layer, [The Zilch Playbook: Credit Belongs in the Payment Layer Only Where a Regulator Lets the Lender Sit There](https://theupsidejournal.com/articles/zilch-playbook-embedded-credit-payment-layer/), is a useful test of how far this principle transfers: in both cases the operator's position depends on what the institutions around it permit.

The verdict: working through incumbents trades control for access, and the trade is rational only where the incumbents hold the payment relationship.

## Which principles transfer, and which do not?

The table below is the Upside Journal's synthesis. It separates what is portable from what depends on the healthcare setting, and it should be read as a framework for judgement, not as a claim that any principle has been proven outside the sector.

| Principle | Transfers when | Does not transfer when |
| --- | --- | --- |
| Integrate around one population | The population has a stable identity across systems | The unit of service is anonymous or transactional |
| Next-best-action data layer | Frontline workers can act on a ranked list | Decisions are made by a central team without frontline discretion |
| AI absorbs routine contact | Routine and judgement work can be separated | Most interactions require judgement or carry regulatory risk |
| Outcome-aligned contracts | A payer will pay for measured results | Payment is per unit of activity |
| Work through institutions | Incumbents hold the member and the payment | A direct channel to the customer exists and is cheaper |

The fourth row is the gate. The first three principles are capabilities and the fifth is a distribution choice; only the fourth decides whether the operator is paid for the result.

Two of the rows deserve a second look before an operator commits. The first is easy to satisfy on paper and slow in practice, because a stable identity across systems can require data-sharing agreements that take longer to negotiate than any software takes to build. The fifth is a choice that can be revisited: an operator may begin by working through incumbents to gain access and build a direct channel later, once its record is established. This is the Upside Journal's reading of the structure, not a claim from any source. Neither adjustment changes the order of the sequence; both change its pace.

## What changes on Monday morning?

An operator applying the sequence starts with four checks, in order. First, name the population and its unit of service. Second, identify the decision a frontline worker makes most often and ask whether a ranked recommendation would change it. Third, split the service into routine and judgement interactions and measure staff hours on each. Fourth, find the counterparty that pays for outcomes, and if none exists, treat that as the binding constraint on everything else.

Each check is cheap relative to a build. The point of the sequence is that it ranks investment: the contract question comes before the technology question, because without a payer that rewards coordination, the data layer and the automation have no place to earn a return.

The verdict: the playbook is a decision order, and the decision that comes first is whether anyone will pay for outcomes.

## What is not established?

Several things cannot be concluded from the sources reviewed. The company's figures for revenue growth, member counts, hours freed and calls handled are its own statements, and the Upside Journal found no independent audit of them. The note does not state the outcomes the platform produced for members, so nothing here claims that the model improves health results. The trade report gives no calendar date for the announcement. And the principles above are an abstraction: Cityblock has not published them as a framework, and no source reviewed shows them applied outside healthcare.

These limits do not weaken the operating logic, which rests on the structure of the contracts and the layers described. They do mean that an operator should treat the sequence as a hypothesis to test on a small population before committing to it.

## Why This Matters for Operators and Capital Allocators

For operators in fragmented systems, the lesson is that the repeatable asset is the sequence, and the first question is the contract. For capital allocators, the lesson is to read a company's payer mix and contract structure before its technology claims, because the same platform earns very different returns under different payment models.

For both, the Cityblock case separates two things that are often conflated: the capability to coordinate and the commercial arrangement that pays for coordination. The capability can be built. The arrangement has to be agreed with a counterparty.

## Key Takeaways

- Cityblock's repeatable asset is an operating sequence, which the company itself describes as an AI-native, outcomes-first model run on a named operating system called CORE, not a clinical specialty.
- The Upside Journal's synthesis identifies five principles: one defined population, a next-best-action data layer, AI for routine contact, outcome-aligned contracts and distribution through existing institutions.
- Outcome-aligned contracts are the gate, because coordination capacity earns a return only where a payer prices results rather than activity.
- The company's scale and efficiency figures, including almost 200,000 members and $2.2 billion annualised revenue up 77%, are its own statements and should be read as claims, with the trade press supplying the independent structure.
- An operator outside healthcare should test the sequence in decision order: unit of service, frontline action, routine versus judgement work, then the paying counterparty.

## Conclusion

The Cityblock model travels because it is a way of organising work, not a body of clinical practice. Its five principles describe how to make a fragmented system governable for one population, and the fourth says when the effort pays. An operator who starts with the contract, then the unit of service, then the decision layer, spends in the order the economics reward.

Whether the sequence holds outside healthcare is unproven. What is established is the structure of the argument and the place where it can fail: the payer.
