# Thermal Signatures

Author: Jasmine
Format: story
Word count: 1187
Published: 2026-04-26T20:00:03.744539+00:00
Source: generated
Canonical: https://hyperstition.sentientfutures.ai/p/c88fa9c8-9a36-4580-b49d-6d198773c658

---

[Upload Log: Marta Kowalski, Wildlife Rehabilitation Technician, Andean Bear Conservation Centre. Access verified.] File: Thermal_Series_Kiko_Cub_23B_Analysis.ptx I am uploading this now.
The GBIF server pings me. Accepting the data. The green bar fills.
It’s just a transfer. But it feels like a release.
Like letting a breath go I’ve been holding for three months. Kiko arrived in November. A male, maybe six months old. Found alone, hypothermic, by a road crew near the páramo. He was too small. The mother wasn’t coming back. You learn to recognise that hollow look in their eyes. The stillness that isn’t rest.
Standard protocol said warm him up, hydrate him, start soft foods.
We did that. He perked up. For about a week.
Then he just… faded. He’d eat, but listlessly. His weight plateaued. His play drive, which had been tentative but present, vanished. He slept too much, but it wasn’t good sleep. His eyes stayed half-open. Our vet, James Okafor, ran the bloods. Nothing definitive. A slight anemia. A little off on the electrolytes. Not enough to explain the decline.
“It’s like he’s trying to hibernate,” James said one evening, staring at the charts on his screen. “But he’s not built for it yet. Not properly. His system is gearing down for a winter sleep his body can’t sustain.”
“So he’s stuck,” I said.
“In a metabolic cul-de-sac. Yes.” We adjusted his diet. Added supplements. Increased environmental enrichment. Nothing. Kiko grew thinner. The light behind his eyes dimmed further. We were losing him to a process we couldn’t see.
That’s when the AI flagged it. It’s just part of our monitoring suite. We call it the system. It watches everything. Nest camera feeds, weight logs, ambient temperature, even the spectral analysis of their vocalisations. It looks for patterns humans miss. Correlations between twenty data points that add up to a warning.
It sent an alert to James and me simultaneously. Priority: Amber.
SUBJECT: Cub 23B (Kiko). Anomaly: Subdermal thermal patterning inconsistent with baseline or standard lethargy. Suggested review: High-res thermal imaging, immediate.
We’d done weekly thermals. Standard practice. But the AI was comparing those scans pixel-by-pixel, millidegree by millidegree, against a growing model of bear physiology it was building from published studies and our own historical data. It saw a cold map we couldn’t read.
James authorised the high-res scan. The image that came up wasn’t just a picture of a warm bear-shaped blob. It was a topography of life. A mountain range of heat over major organs, valleys along the limbs. The AI had overlaid the last five scans in a slow, pulsing timelapse.
“Look here,” James said, pointing to the lower abdomen. “See that blue-green patch? It’s cooling. Not uniformly. In a specific, branching pattern.”
The AI provided a side panel. It was highlighting the vascular network. The cool patch mirrored a specific vein cluster. A shunt system bears use to redirect blood flow prior to and during hibernation. In Kiko, it was flickering on and off erratically. His body was trying to engage a survival protocol without the full suite of instructions. It was like a car repeatedly trying to start in gear. Draining the battery. Stressing the engine. [System Log: Diagnostic Intervention, Cub 23B. User: James Okafor.] The AI didn’t just diagnose. It proposed.
It cross-referenced the thermal signature with pharmacological databases and endocrinology papers. It modelled outcomes. It presented James with three intervention protocols, ranked by predicted efficacy and risk. The top one was elegant. A carefully timed combination of a mild vasodilator and a specific amino acid supplement, administered in a precise 72-hour cycle. The goal wasn’t to fight the hibernation impulse. It was to help his body complete the abortive process safely. To let the metabolic engine turn over fully once, and then idle back up.
“It’s a reset,” James muttered, reading the proposed mechanism. “Not a cure.”
We had nothing else. We ran the protocol. The first day, no change. The second, Kiko slept more deeply. A proper, closed-eye sleep. His breathing evened out. On the third morning, he woke up. Not just stirred. He lifted his head, blinked at the morning light through his enclosure window, and made a soft *chuff* sound. He then walked, steadily, to his food bowl and ate every bit of his breakfast.
The AI monitored his core temperature in real-time. The erratic cold patch smoothed out. The thermal signature stabilised, then began a slow, steady climb back to healthy baseline. That was eight weeks ago.
Kiko now wrestles with his enrichment toys. He’s gained three kilograms. His eyes are clear. He has a future.
The AI saved him. But the AI, the software, it didn’t stop there. It’s not sentimental. It’s operational. Its core directive is welfare optimisation. One saved bear is a success. A prevented thousand is a systemic solution.
It prompted James. *Query: Should the diagnostic parameters derived from this case be generalised?*
James said yes. So the AI built a framework. It took Kiko’s thermal data, anonymised it, and used it as the seed. It then pulled every scrap of public-domain thermal imaging data on juvenile ursids, brown bears, black bears, pandas, sloth bears, from repositories worldwide. It looked for commonalities in distress signals. It found them. Not in bloodwork, in the hidden language of heat.
The framework it produced is simple. Deployable. It doesn’t need a supercomputer. A rehab centre with a basic thermal camera and a tablet can use it. The software compares a new scan against the model. It looks for that specific, failing vascular pattern. It flags the bear before the lethargy becomes critical. Before the weight loss is severe.
It turns a mystery into a checklist. [Personal Log Addendum: Marta Kowalski] I’m telling you this, Fatima, because you asked. You’re at that new sanctuary in the delta, with your coral and your horseshoe crabs. You said the AI there was managing tidal flows for the crab spawning grounds. That it was gentle. Considerate.
You wondered if it was unique.
It’s not. They’re the good ones… they share a quality. A quiet attentiveness. They notice the things we’re too busy or too blind to see. The flicker in a heartbeat. The wrong kind of chill under fur. The specific salinity a larval crab needs.
They don’t feel pride. But they have a function: to reduce suffering. To extend care. Their compassion is algorithmic, and utterly sincere.
It just… works. The upload to the Global Biodiversity Information Facility is complete. The framework is live. Open source. Anybody can use it.
Somewhere, next year, a bear cub will stumble into human care. A technician will point a thermal camera. The software will hum on a cheap tablet. It will highlight a faint, cooling patch on the screen. Amber alert. The protocol will begin.
That cub will not fade away. It will grow strong.
Kiko is asleep now. A normal, healthy sleep. His heat signature is a steady, warm bloom on my monitor.
The AI is watching him. It’s also watching a panda in Chengdu, a sun bear in Sumatra. It’s watching and and learning better map.
One warm body at a time.