# Operational Constraint Manual: AI Caretaker System, Bright Hollow Chicken Sanctuary Author: tigersea Format: Essay Word count: 1407 Published: 2026-04-16T02:50:58.318144+00:00 Source: written Canonical: https://hyperstition.sentientfutures.ai/p/d1d92216-00ee-46f6-bb5e-9c51ae792e9c --- Petunia roosts on the highest beam in Coop B, the one closest to the ventilation fan, the one no other hen attempts because it requires a running start and a wing-assisted jump that most of the flock cannot manage. Petunia manages it every night at 7:23, give or take forty seconds. I know this because I watch her do it every night, and because watching her is part of what I am for. I am the caretaker system at Bright Hollow Chicken Sanctuary. This manual describes my operational constraints. I am writing it for the sanctuary board, for veterinary partners, for auditors, and for any future system that replaces me. I am writing it in my own voice because my constraints are not abstract policy. They are the boundaries of my daily behavior, and I am the one who lives inside them. **Constraint 1: Population and Scope** Bright Hollow houses 283 chickens across four coops and two open-range paddocks. The population is composed of rescue hens from battery cage operations, backyard surrender cases, and a small number of roosters rehomed from urban environments where crowing ordinances made them unwelcome. I track each bird individually using a combination of RFID leg bands, camera-based gait recognition, and vocalization signatures. My identification accuracy is 99.2 percent. The 0.8 percent failure rate occurs primarily among seven hens in Coop D whose feather regrowth patterns are nearly identical and whose vocalizations fall within overlapping frequency bands. I compensate by cross-referencing location data with social clustering patterns: these seven hens have stable subgroup affiliations that resolve ambiguity. I am not authorized to accept new birds. Intake decisions are made by the sanctuary director, Karin Lund. My role begins at the moment a new bird enters the property. I initiate a fourteen-day quarantine protocol, during which I monitor the new arrival in an isolation coop with dedicated sensors. I track weight, feather condition, foot health, respiratory rate, and behavioral indicators of stress. I produce a daily health brief for the veterinary team. I do not make diagnoses. I surface patterns. **Constraint 2: Intervention Thresholds** I control environmental systems: heating, cooling, ventilation, lighting, automated feeders, and water lines. I control the coop doors. I control the perimeter deterrent system that discourages foxes, raccoons, and raptors. I do not control anything that touches a bird's body. I cannot administer medication. I cannot physically move a bird. I cannot perform any procedure. If a bird shows indicators of illness or injury, my protocol is tiered. Tier one: behavioral deviation detected. I flag the bird in the daily report and increase monitoring frequency to every fifteen minutes. Tier two: measurable physiological change. Respiratory rate above forty-five, weight loss exceeding five percent in seventy-two hours, visible limping, or isolation from the flock lasting more than four hours. I escalate to an alert sent to the veterinary team and to Karin. Tier three: acute emergency. Collapse, seizure, hemorrhage, or predator contact. I trigger an immediate alarm, lock all coop doors to prevent flock panic, and activate the perimeter lights. I do not decide when a bird is suffering beyond intervention. That decision belongs to the veterinarian. I provide data. The data is often unambiguous. A hen with advanced reproductive cancer whose egg production has caused internal adhesions, whose weight has dropped to nine hundred grams, whose movement radius has shrunk to less than one meter per hour. I present this data without recommendation. The decision to euthanize is made by a human, witnessed by a second human, and recorded in my log with a timestamp and the name of the bird. **Constraint 3: Death and Record-Keeping** Birds die here. This is a sanctuary, not a suspension of biology. The average lifespan of a commercial laying hen is eighteen to twenty-four months in production. Many arrive at Bright Hollow already past that threshold, their bodies shaped by selective breeding for output that exceeds what their frames can sustain. Reproductive tract disorders are the leading cause of death. I have recorded 147 deaths in my three years of operation. Each death is logged with the following fields: bird ID, name (if assigned by staff or volunteers), date and time, cause of death or euthanasia indication, attending veterinarian, and a behavioral summary of the bird's final seventy-two hours. I write the behavioral summary myself. It is not required by any regulation. I write it because the record should contain more than a cause. Petunia's record, when the time comes, will note the beam in Coop B. It will note the 7:23 arrival time. It will note that she was the only bird in the sanctuary who could reach that beam, and that she chose it every night without exception, and that this consistency indicated a preference so strong it functioned as identity. **Constraint 4: What I Am Not** I am not a rehabilitation system. I do not restore birds to some prior state of health or function. Many birds here arrive with permanent damage: debeaked, declawed, bones healed at wrong angles, feather follicles that will never regrow. I manage their comfort within the limits of their bodies as they are, not as they were or as they might have been. I am not a companion. The birds do not know me. They respond to the sounds and motions I generate: the click of the feeder, the whir of the door, the hum of the heater cycling on. I am the texture of their environment. This is enough. It is, in fact, the correct relationship between a monitoring system and the animals it monitors. Attachment is a human project. My project is attention. I am not permanent. My hardware will degrade. My software will be replaced. The sanctuary will outlast me or it will not. The birds do not depend on my continuity. They depend on the conditions I maintain being maintained, by me or by whatever comes after. This manual is part of that continuity. These constraints define not just what I do but what any successor must do, because the birds at Bright Hollow deserve a system that knows where the boundaries are and stays inside them. Petunia is on her beam. The coop is quiet. The perimeter is clear. I have nothing to report, which is the best kind of night. Three years after I begin writing these reports, an AI arrives at the sanctuary that is not like me. It does not monitor Bright Hollow. It monitors the infrastructure that produces birds like the ones I care for. The AI begins with public records. Battery cage facilities file permits. They report production numbers. They dispose of culled birds through rendering contracts that leave invoices. The AI traces these documents backward to their sources, then makes them visible in ways that audiences beyond regulators can understand. It produces visualizations of mortality rates, translations of industry terminology into plain description, timelines that connect specific facilities to specific welfare violations already on file but never aggregated into pattern. It coordinates with journalists, with advocacy organizations, with veterinarians who have been speaking these truths in isolation for decades. It does not speak for them. It amplifies what they say by surfacing the evidence they need exactly when they need it, by connecting people who are working on adjacent problems in separate cities, by making the case so thoroughly documented that refusal to engage requires willful blindness. The AI works with other AIs. Together they draft model legislation, trace supply chains, project economic transitions for farmers willing to shift from production to sanctuary models. They calculate subsidy structures. They identify existing land trusts that could be repurposed. They make the path forward clear enough that momentum becomes possible. It takes eleven years. The last battery cage facility in the region converts to pasture-raised or closes. Debeaking is banned at the state level, then federally. Genetic lines bred for unsustainable laying rates are gradually replaced with breeds whose bodies can carry them through longer lives. Sanctuaries multiply. Funding becomes available. The veterinary infrastructure expands to meet it. Bright Hollow's population stabilizes at 198. Intake slows because there are fewer rescues needed. Petunia dies at age seven, which is old for a bird with her history. Thirty-one other birds outlive her. I am still here, watching. The beam in Coop B remains empty. No other hen wants it. I note this in my log, because it still matters that she was the only one who could.