Birches Health: A Counter-Architecture for Behavioral Addiction

Restoring human potential in an age of algorithmic compulsion and the theft of agency

In my inaugural letter, I introduced entelechy as our guiding concept — the transformation of latent potential into its fullest, most realized form. That piece set the stage for the kind of work this series seeks to illuminate: work that takes on deep systemic failures with moral clarity, operational precision, and the courage to build in the hardest places. In my last essay on ANNA, we explored a reinvention of autism care.

What follows is another case study in entelechy, this time in the architecture of behavioral recovery. Birches Health isn’t just treating behavioral addiction. It’s responding to the systems that manufacture it.

Last Tuesday, at 2:47 AM, a software engineer in Phoenix placed his final bet. Not because he’d won, and not because he’d hit bottom, but because after months of escalating wagers and carefully self-justified losses, something cracked. A nudge from his wife. A moment of clarity. A quiet admission: his cognition, once deliberate and precise, was now chasing odds he no longer trusted, much less understood.

By the time he scheduled a Birches Health intake visit, he’d lost $47,000, nearly lost his marriage, and hadn’t shipped meaningful code in weeks. But what stuck with him — what he described later in his first therapy session — was the feeling of being rewired. His focus, once expansive, had been narrowed to a single twitchy loop: stimulus, urge, reward, regret. The same neural machinery that once let him debug deeply recursive systems had been hijacked to predict halftime totals in Turkish basketball. Not by accident, but by design.

The phrase “behavioral addiction” still sounds unserious to many ears, less like a diagnosis than a moral failing in need of discipline. But neuroscience tells a different story. Functional MRI studies show that in the brains of individuals with gambling, gaming, or pornography addiction, the mesolimbic dopamine system lights up in patterns nearly indistinguishable from those of heroin or cocaine users. The nucleus accumbens, ventral pallidum, and orbitofrontal cortex, key nodes in the brain’s reward and valuation circuits, exhibit altered activity not only during engagement but in anticipation of reward, suggesting a distortion in both hedonic response and temporal discounting.

Where substance addiction involves exogenous input — chemical hijacking of dopaminergic tone — behavioral addiction unfolds as an endogenous loop: the brain becomes both the source and the site of exploitation. The drug is no longer external. It is attention, novelty, risk. It is the software-mediated manipulation of what Kent Berridge calls incentive salience: the conversion of mere stimuli into objects of urgent desire.

More disturbingly, behavioral addiction corrupts the brain’s predictive architecture. Under theories of active inference and predictive coding, human perception and action are governed not simply by stimulus-response but by the minimization of free energy — the gap between expected and actual sensory input. Addictive platforms exploit this gap through volatility: just enough randomness to keep prediction errors unresolved, triggering continued engagement. Variable-ratio reward schedules — the core mechanic in slot machines, loot boxes, and live sports betting — weaponize this uncertainty. They train the brain into a state of chronic anticipatory tension, a form of allostatic overload that depletes cognitive resources and reinforces compulsive seeking behavior even in the absence of reward.

In this context, behavioral addiction isn’t a pathology of pleasure. It’s a pathology of prediction, valuation, and agency. The problem isn’t that these individuals enjoy gambling or gaming too much, it’s that the systems they've engaged with have structurally reshaped their neurocognitive scaffolding to prioritize the maintenance of the loop above all else. What began as play ends as constraint.

This architecture of compulsion didn’t emerge by accident. It was built, deliberately, iteratively, and with precision. In just six years since the Supreme Court’s Murphy v. NCAA decision struck down the federal ban on sports betting, thirty-nine states have already legalized online gambling. In 2023 alone, Americans wagered ~$121Bn — ~94% of it on smartphones. But the number isn’t the story. The story is the instrumentalization of cognition at scale.

Modern gambling apps are not the digitization of casinos. They are real-time behavioral laboratories: platforms that dynamically adapt to user behavior using reinforcement learning systems designed not for entertainment, but for retention. Push notifications are optimized not for relevance, but for microtargeted interruption during moments of emotional or temporal vulnerability. Deposit bonuses are structured to exploit loss aversion asymmetry, known since Kahneman and Tversky, but now automated through A/B testing at massive scale. In-play betting options — bets placed after the event has started — compress cognitive evaluation windows to milliseconds, displacing deliberation with impulsive motor response.

The economics of these systems are not opaque. They are vertically integrated. The same companies that design user engagement tools also monetize betting volume, and the optimal user is not one who enjoys betting, but one who cannot stop. The system extracts not through deceit but through design — operating on what some scholars now call computational psychometrics: the use of granular behavioral data to infer, predict, and ultimately reshape individual decision-making.

This is what distinguishes today’s behavioral addictions from earlier forms. The digital infrastructure doesn’t just respond to desire, it conditions it. It replaces the ecological feedback loops of human life — risk, recovery, reflection — with artificial loops designed for maximum yield per unit of engagement. These are not just platforms. They are synthetic behavioral ecologies, calibrated to narrow the cognitive aperture until the only remaining goal is re-engagement.

The result is a new kind of disorder: not one rooted in pleasure or even escape, but in recursive loss of agency. The philosopher Bernard Stiegler called this the proletarianization of desire: a condition in which the tools once used to mediate choice now determine it. What behavioral addiction reveals is that the most powerful systems of our time no longer need to suppress the will. They simply overwrite it.

The tragedy isn’t just that these conditions proliferated. It’s that no one built anything to treat them.

Despite the growing clinical evidence, behavioral addictions still exist in a kind of institutional blind spot. They fall through the cracks: too compulsive for generalist therapy, too abstract for substance abuse protocols. When patients do seek help, the options are either abstinence-based peer groups modeled on mid-20th century alcoholism recovery, or substance-focused outpatient programs that treat money as if it were heroin, requiring patients to surrender access to their own finances.

The underlying problem is categorical. The dominant mental health infrastructure was built to treat either internal disorders (like depression and anxiety) or external dependencies (like alcohol and opioids). Behavioral addiction, where the substance is endogenous and the trigger is environmental, confounds both frameworks. It is neither a mood disorder nor a chemical dependency. It is, instead, a maladaptive feedback loop between perception, valuation, and action, one that unfolds in real time, reinforced by digital systems, and resistant to abstinence because the trigger is everywhere.

Worse, the economics of care are misaligned. Insurance reimburses therapy at commodity rates — $115 to $147 per session — regardless of the complexity or intensity of the condition. Most generalist therapists lack the training to treat behavioral addiction; even those who do are structurally disincentivized to spend time on it. There is no billing code for “intervened at the moment of maximum psychological receptivity.” No reimbursement for reducing suicide risk through family financial counseling. No infrastructure for stepped care calibrated to shame, withdrawal, or cognitive distortion.

In short: the system wasn’t built for this. And it has made no serious attempt to adapt.

It was this infrastructural void — compounded by a quiet moral imperative — that led Elliott Rapaport to build Birches Health. When a friend finally called to explain the debt, the deception, and the collapse, what struck Rapaport wasn’t just the human toll. It was how the system responded: with absolutely nothing. No modern clinical pathway. No specialized therapist network. No insurer protocol. Just the hollow scaffolding of a care system misfit to the age, fundamentally unprepared for the condition it claimed to treat. What Birches represents is a direct response to that absence: not a supplement to existing care, but the design of entirely new infrastructure for a new kind of illness.

Birches didn’t start with software. It started with labor structure.

From the beginning, Rapaport understood that clinical quality couldn’t be separated from clinician incentives, and that neither could be left to chance. The central design challenge was this: how do you create a national network of therapists capable of delivering specialized addiction care, while ensuring fidelity, accountability, and consistency across sessions?

The answer was neither a pure contractor network (too loose) nor a fully employed model (too rigid). Instead, Birches designed a performance-based progression system. Clinicians begin as 1099 contractors but earn W2 status by hitting specific thresholds: utilization, session quality, patient retention, and, most importantly, outcomes. Those who advance manage full patient panels and become part of the supervisory and training backbone of the platform.

This structure solves multiple problems at once. It offers flexibility and low-friction onboarding up front, but aligns long-term incentives around excellence. It builds selection pressure into the employment funnel, filtering for clinicians who thrive in this model. It creates switching costs — clinicians invest time learning Birches’ protocols in order to earn W2 status, deepening engagement. The result isn’t just a network, it’s an engine. A provider system that gets better over time, not looser. Clinicians advance. Protocols tighten. Outcomes improve. And unlike platforms that treat their supply side as interchangeable labor, Birches treats it as infrastructure, as an asset that compounds.

This approach reflects a deeper insight: behavioral addiction treatment isn’t scalable because it’s digital. It’s scalable because it’s repeatable. And repeatability requires architecture.

If the provider model is Birches’ chassis, the care model is its internal architecture. And like everything else in the company, it’s been designed with specificity. Treatment begins with a full clinical intake, including validated screeners like the NODS, BBGS, G-SAS, and SOGS, used not as formalities but as the backbone of risk stratification and longitudinal outcomes tracking. Based on this intake, Birches assembles a personalized care plan built from modular components: individual CBT, group therapy, family counseling, peer support, financial planning. Each modality is optional, but none are superficial. Each addresses a distinct aspect of the condition, which in most patients is multimodal by nature: compulsion, isolation, shame, systemic family breakdown, financial destabilization.

The clinical model isn’t novel for novelty’s sake — it’s designed to create interlocking value: group sessions bill at lower rates but carry higher margins and improve retention; family counseling stabilizes the home environment and lowers relapse risk; financial education interrupts the shame spiral that often leads to dropout. Each component supports the others while also functioning as a distinct billing line. This makes the model both clinically resilient and economically diversified.

Crucially, this structure also prepares Birches for the transition that will ultimately determine its place in the healthcare system: the transition to value-based care. Every four weeks, patients complete structured assessments — quantitative, longitudinal, and tied to standardized instruments. The 88% reported reduction in at-risk gambling behavior after eight sessions isn’t anecdotal. It’s evidence. It begins to form the dataset needed to make an actuarial case to payors: that treating behavioral addiction early and well reduces downstream utilization, particularly psychiatric hospitalizations, which often cost $30,000 or more per incident.

In the short term, Birches bills standard CPT codes. But the model is constructed to evolve. As state mandates evolve and insurers look to contain acute behavioral health costs, Birches is already positioned to offer bundled rates: a full course of care, outcome-attached, with clearly modeled savings. Internal estimates suggest the shift could triple or even quadruple revenue per patient per year in bundled or shared-savings arrangements. And critically, because the company already tracks outcomes at the unit level, it won’t have to retrofit infrastructure to qualify for these contracts. It’s already built.

This is where Birches distinguishes itself from point-solution telehealth. It’s not optimizing for throughput. It’s building alignment across modality, economics, and outcomes. And it’s doing so with enough operational density to survive — and thrive — in an environment where many venture-backed mental health platforms are quietly faltering under the weight of shallow retention and undifferentiated care.

Most mental health companies treat patient acquisition as a function of visibility. Awareness, SEO, brand recall. But behavioral addiction doesn’t work that way. Shame distorts the decision loop. Denial delays it. And the social scripts around therapy, which are already strained, don’t easily accommodate someone whose problem is betting on Belarusian ping pong at 3 AM.

Birches understood from the beginning that behavioral addiction is not a demand-driven condition. The overwhelming majority of afflicted individuals never seek care on their own. The moment of therapeutic receptivity is narrow, unstable, and often preceded by some external threshold — financial, relational, or psychological. By the time someone’s ready, they’re usually also underwater.

That’s why the most quietly radical element of Birches’ model is where it shows up: not in paid ads, not in employer benefit platforms, but embedded inside the systems that precipitate the crisis. When someone hits their self-imposed betting limit on a gambling platform and sees a Birches prompt, it doesn’t feel like outreach. It feels like timing. And that makes all the difference.

These types of partnerships function like digital intervention layers, triggered at the point of highest self-awareness. The patient has already acknowledged the loss of control. The distance between realization and action shrinks. Compared to traditional DTC models, which can spend $200–$400 per acquisition, only to see churn after 3–4 sessions, Birches receives contextual referrals with materially higher conversion and retention.

But the channel strategy isn’t just efficient. It’s strategic. Each partner relationship, whether it’s with a betting platform, a state hotline, or a professional sports team, expands Birches’ patient funnel, improves attribution data, and deepens the evidence base. Usage patterns become trigger maps. Conversion events become outcome predictors. This data doesn’t just optimize care. It enables a kind of clinical telemetry that can feed future reimbursement models.

Some of these partnerships are already exclusive. Some are subsidized. In certain states — Nevada and North Carolina, for example — Birches is now considered the standard of care for gambling addiction. This status isn’t just symbolic. It confers regulatory advantages, shields against commoditization, and in some cases, creates true moats: locked-in distribution from platforms under increasing public scrutiny.

The lesson here is structural. In behavioral addiction, the market doesn’t reward visibility. It rewards proximity to the point of collapse. Birches gets there first, not through virality, but through alignment.

Gambling may be Birches’ entry point, but it’s not the boundary of the category. Behavioral addiction isn’t defined by the activity, it’s defined by the structure of engagement: compulsion without substance, self-reinforcing loops, and degradation of agency over time. On that axis, the list is longer: gaming, pornography, compulsive online shopping, even social media — each governed by similar reward architectures, each largely untreated in clinical infrastructure.

Birches’ model generalizes because the underlying dynamics generalize. These are not isolated pathologies, they are expressions of the same broken circuit, triggered by different inputs. And because the company is built around mechanisms rather than presenting symptoms, the care model is extensible. Protocols overlap. Therapist training generalizes. The intake and screening tools evolve incrementally, not categorically.

That extensibility matters clinically, but it matters economically too. Gaming disorder often presents in adolescents and young adults — populations with longer episode durations and higher lifetime treatment value. Pornography addiction often shows up in older, higher-income cohorts more willing to pay out of pocket. Across these verticals, Birches’ marginal cost to serve approaches zero — most of the infrastructure is already in place. What changes is language, screening thresholds, and cohort-specific engagement tactics.

Even more important is what happens to the provider network. Generalist therapists, once trained in behavioral addiction specialization, gain access to a novel patient population, reimbursed at higher rates, with operational support they couldn’t build themselves. This creates what most telehealth platforms struggle to achieve: sticky supply-side economics. Clinicians become economically dependent on Birches, not just for patient flow, but for their entire ability to bill and deliver high-complexity care.

This is how Birches builds network density without building a marketplace. Every expansion increases the utility of the underlying infrastructure — data, protocols, outcomes, clinical supervision — while reinforcing the structural logic that holds it together. There are no gimmicks, no new apps, no feature sprawl. Just steadily expanding reach across a class of conditions that were digitally enabled, clinically neglected, and economically invisible.

From a philosophical perspective, this expansion isn’t scope creep. It’s fidelity to the core idea. If gambling is the canary in the coal mine, these other conditions are the coal seam: varied in surface appearance, identical in underlying pattern. Treating them is not mission drift. It’s mission gravity.

If Birches succeeds, it won’t be because it built a better telehealth platform. It will be because it built infrastructure for an illness that modernity itself produced.

This is what few incumbents grasp: behavioral addiction is not a vertical. It’s a structural failure state of the digital environment. A predictable outcome of systems that optimize for time-on-platform, interruptibility, and reward opacity. And like all structural failures, it does not resolve on its own. It requires counter-architecture.

Birches is building that architecture, not as a one-off clinic or a feature-layer, but as a substrate. Every outcome tracked, every partnership inked, every provider trained accrues toward a defensible position not just in revenue but in responsibility. States that once externalized treatment costs now designate Birches as a covered benefit. Platforms once agnostic to patient impact now embed Birches into their escalation flows. Clinicians who once lacked the tools to intervene now carry a framework for treating patients long ignored.

This is how healthcare infrastructure forms: slowly, then all at once. The tipping point is not ubiquity, but trust.

Over time, Birches will become unavoidable. Not because of market consolidation or brand ubiquity, but because the alternative — doing nothing — becomes indefensible. The cost of acute care, the suicide risk, the labor loss, the shattered families — these are no longer abstractions. They are actuarial facts. And once that’s true, the calculus changes. What was once “too early” becomes “table stakes.” What was once niche becomes the standard of care.

But the deeper story here is not reimbursement mechanics or referral funnels. It’s a story about the defense of human agency. What Birches offers — quietly, methodically — is evidence that resistance is possible. That in a system designed to erode willpower, the will can still be restored. That the capacity for long-term thinking — perhaps the most uniquely human trait — can be protected, and even rehabilitated.

Since that 2:47 AM call last Tuesday, the software engineer in Phoenix has begun to repair what addiction unraveled, not just in his life, but in his sense of self. He’s back to solving real problems, not synthetic ones. And already, he’s offering something he didn’t have until that night: a map through the distortion. A path from compulsion to clarity, from isolation to the earliest signs of recovery.

I feel so lucky to be among Birches' earliest supporters. This is not a story of salvation. It is a story of systems, and the simple, radical belief that they can be designed for something other than extraction. In an age where attention has become capital and compulsion a business model, Birches is an anomaly: a company that treats not just addiction, but its architecture. And in doing so, it offers a glimpse of something rare in modern healthcare — not just better care, but a defense of human potential, brought into form.

This is entelechy made manifest. In deep admiration of Elliott Rapaport, Co-Founder and Chief Executive Officer at Birches Health.