Reported presence — by zone, not by point
A quiet zone is not a cleared one — reports are delayed, approximate, and never exact.
Jhapa Elephant Tracker
A safety map that tells you the zone, never the exact tree.
In Jhapa, knowing where the herds were last night is the difference between a normal walk to school and a funeral.
Human-elephant conflict in the Jhapa district is real and serious — crop and property destruction, injury, and death on both sides, year after year. Until this existed, the only warning system was informal — a neighbour mentioning at a tea stall that the herd had crossed near the river two nights ago, passed hand to hand until it was too old to matter or never reached the one household that needed it.
The turn
What if the map could warn people without ever pointing at an animal?
The tracker takes the same informal warning villages were already passing along and gives it structure — a shared, always-current picture of which parts of the district have had recent activity. Asian elephants are endangered, and precise, real-time location data on an endangered animal is a well-documented poaching risk — open wildlife-location feeds have been used before to find and kill the animals they were meant to protect. So the tracker is built around one restraint from its very first version, not added after the fact: it only ever shows six coarse, named zones, never a pin, a coordinate, or anything finer. The imprecision is the safety feature, and it is the first thing the product tells you, not the fine print.
How it works
- 01
See it, report it
A person who spots a herd, or fresher signs of one — trampled fencing, broken bamboo, fresh dung — opens one screen, in Nepali first. There is no pin to drop and nothing to type: a zone picker, a herd-size tier (single, small group, herd), a confidence tier (saw them, signs only, heard from someone), and one large button. That button reads हात्ती देखियो — "elephant seen." Sending a report takes under fifteen seconds, which matters, because the moment someone needs to file one is rarely a calm moment.
- 02
The coordinate is thrown away on purpose
If the phone has a GPS fix at all, it is used for exactly one thing: choosing which of the district's six fixed zones — River Corridor, Forest Margin, Tea Estate Belt, Settlement Edge, Grassland Flats, Northern Foothills — the report belongs to. The design is that the instant that's decided, the coordinate itself is discarded on the device — never transmitted, never stored, never logged. The intent is that what exists anywhere in the system, for any report, is a zone, a tier, and a timestamp — nothing finer-grained ever existing to be intercepted, subpoenaed, or scraped, because it is deleted before it would need to be protected.
- 03
One report is a whisper, not an alarm
A single report shows its zone as merely "unconfirmed." A second, independent report in the same zone inside a rolling six-hour window promotes it to "active." A third inside that same window promotes it to "elevated" — the only tier that pushes an alert to anyone rather than sitting quietly in the shared view. The design goal is that one mistaken or bad-faith report, from one person, cannot by itself put an alert in front of anyone else.
- 04
A short delay, and a narrower first audience
Anything short of "elevated" sits behind a short buffer — on the order of half an hour — before it appears even at zone level in the shared view, long enough for a mistaken report to be quietly walked back before it was ever seen by anyone but its author. An "elevated" zone is meant to reach people faster, but only the people who have set that zone as home — never a district-wide push. The restraint that shapes the data also shapes who gets told, and how urgently.
- 05
Reaching people without a data connection
Rural Jhapa connectivity is not guaranteed, and the moment someone needs this information is the worst possible moment to be waiting on a signal. The plan is that texting a zone's short name to a shared number will file a report with no data connection at all, and that an elevated alert will reach an opted-in phone the same way — the reply is meant to read exactly like the map itself does: "River Corridor — elevated, 45m ago." The app is also built for low-end Android and a small install size, since a safety tool only current phones can run is not a safety tool for this district.
The reporting screen is deliberately plain: six zone buttons rather than a map to pan and zoom, two short tier pickers, one send button — a villager with thirty seconds and patchy signal is the actual design constraint, not a visitor with a fast phone and time to explore. The shared view that everyone sees back is just as plain, and just as honest about what it doesn't know: six zones, each carrying only a name and a last-reported time — never a number of animals, never a direction of travel, never anything that would read as more precise than the data actually is.
Teaching at the right level
The methodology's usual shape is a curriculum matched to a learner's level. Here the "lesson" is a community learning to read and act on a pattern — where the herds move, roughly when, and what a rising confidence tier actually means — and the right level for that lesson is set by the district's real conditions, not by what a modern smartphone can do. That's why the entire product is built Nepali-first with an English toggle rather than the reverse, why the interface is buttons and tiers instead of typing, and why the SMS fallback isn't an afterthought bolted onto an app — it delivers the identical zone-and-recency information the app does, to a phone with no data plan at all. This tracker doesn't run on Pratham's classroom data the way the school-facing ventures do — it isn't an education product — but it's built on the same portfolio-wide discipline of designing for the community's actual conditions rather than the tidier, idealized version of them. See the full methodology
Where it lives
Founded in 2022, and grown the way a piece of community infrastructure grows rather than the way a startup does: village by village, by word of mouth, over several years, now covering a meaningful share of the district. The idea didn't arrive independently — it surfaced through the 100 Guitars relationship in the same district, when ordinary conversation in Budhabare kept turning to where the elephants had been seen that week. It is run today by a small team with no plans to change that, sustained by a modest conservation-sector grant and ongoing support from the wider BeethovenJr portfolio. That shared ground is deliberate, not incidental: 100 Guitars and the tracker both operate in Budhabare, Jhapa, and each makes the other more credible — a guitar programme on its own is a visit, a guitar programme alongside a safety tool the community actually relies on is a relationship. See 100 Guitars
What we don’t know yet
This app helps reduce risk. It does not, and cannot, guarantee that any specific encounter is prevented — the underlying reports are crowdsourced, which means they are sometimes stale, sometimes sparse, and occasionally just wrong. A quiet zone is not a cleared zone. Absence of a recent report means exactly that and nothing more — it is never presented, anywhere in the product, as a promise that an area is currently safe. There is no revenue model, and there is not meant to be one — monetising a safety tool used by people at real risk of injury would be indefensible on its own terms, independent of anything it might have earned. Who formally validates a report at scale, beyond the corroboration threshold built into the app itself, is still an open question this project is genuinely working through rather than one it has quietly answered for the page.