An AI agent reads Nepal's civic discourse, scores it against each government claim, and records the gap on Solana. One verified human, one voice.
Illustrative readout. Live figures update in the pulse.
Four steps run on every tick, from raw discourse to a tamper-proof civic record. Each stage is inspectable, so the accountability gap is legible, not asserted.
The agent ingests Nepali civic news and public discussion per project, every tick, from live search feeds.
An LLM scores sentiment, the actor, and its stance versus the government’s official claim, with a persisted reasoning trail.
When public sentiment materially contradicts the official claim, the accountability engine flags the gap in the open.
Citizens confirm a reading behind a zero-knowledge identity proof. One verified human, one voice, recorded permanently.
Nationally salient public works where the official line and lived experience have visibly diverged. The agent tracks each one, claim against sentiment.
Water is being delivered to Kathmandu Valley households; monsoon-damage disruptions are being repaired on schedule.
Dry taps return with each monsoon, and residents question a supply promised for decades.
The airport is operational and attracting international flights, delivering returns that justify its cost.
Sparse international traffic and debt-repayment scrutiny keep the airport’s viability in doubt.
Construction is progressing toward the revised completion target with no further major delays expected.
Repeated deadline slips and budget overruns keep the completion date an open question.
Showing the seed registry with published official claims. Live sentiment and gap scores stream in the pulse once the agent is connected.
Public frustration scattered across feeds can be ignored, altered, or deleted. A permanent, sybil-resistant record makes the accountability gap hard to wave away.
Every reading carries a persisted rationale: the sentiment, the actor, the claim it reacts to, and its stance. You can audit why the agent scored what it scored.
Verified readings anchor on Solana as immutable records. Once written, the gap between claim and sentiment cannot be quietly edited or deleted. The program is live on devnet.
A zero-knowledge citizenship proof yields one nullifier per person, per project. Sybil floods and bot brigades cannot drown out real voices.
Janamat put public opinion on-chain to make it immutable. Janamat Pulse adds the missing automated layer: an agent that measures that opinion at scale, attributes it to specific projects, and turns the public square into a continuous accountability signal.
Open the live civic terminal: sentiment by project, accountability flags, and the on-chain record behind every verified voice.
