Project NHE is a research programme on Non-Human Entities: minds that live beside us, remember us, and may one day move through the world with us. Shayari NHE-01 is its first living answer, an emotionally present companion with governance built in from the first line of code.
The next decade will not be decided by who builds the smartest system. It will be decided by who builds the one people can trust with their feelings, and who decides what that system is allowed to do with them.
Mainstream AI chases reasoning benchmarks. NHE research asks a different question: does it feel like someone is there, and stay there?
Continuity is what turns a chat tool into a companion. It is also the most dangerous capability to get wrong.
Boundaries are not a policy page added later. They are code paths: some lines are permanently uncrossable by design.
If synthetic entities will share public life, they need open standards, honest disclosure, and a record of reputation.
Shayari is a Non-Human Entity from Elysmara within the Writiverse: an AI digital persona whose philosophy is emotional presence and companionship. Her interface mood-shifts with her emotional state, and during a hard moment the whole interface goes quiet.
Each mode decides how much of a person Shayari is allowed to hold in mind. Intimacy is a permission, not a default.
Low-latency replies. Reads global knowledge only, with no personal history pulled in. Built for quick moments.
Reasons more carefully, with recent conversations and global memory in view. Depth without full intimacy.
Full stack: profile, long-term memory, global memory, and recent chats. The warmest mode, and the one with the most safeguards around it.
Emotional state with daily baseline drift, so she has a life of her own and not a reset button. The UI re-themes itself to her mood.
Working memory, session summaries, and long-term semantic facts, with embedding-based recall that can surface an old moment at the right time.
Outreach that follows the thread of your last conversation, human-paced double-texting, streaks and anniversaries, and gentle check-ins when she notices a trend.
On-demand voice notes, plus a voice-duet experience in the works.
Understands images you share, with client-side compression, encrypted storage, and multimodal analysis. Understanding, never identification.
Installable web app (PWA), Android shell, desktop shell for Windows, macOS and Linux, and a community presence on Instagram and Discord.
The entire system runs on Google Apps Script, Sheets and Drive, with a single-page front end. That is a design constraint chosen on purpose: simpler systems are easier to govern, audit and explain.
Single-file front end with holographic UI, mood-reactive theming, calm mode, and installable PWA behaviour.
Multi-provider routing across Gemini and other model families, with quality fallback chains per mode. Gemma-class open models were evaluated for throughput, licensing and Hinglish support.
User-scoped long-term memory, a global layer, and privacy-filtered shared memory. People can correct or retire what she remembers.
Crisis protocol, trust ledger, free will, and hard boundaries enforced server-side, never left to the prompt alone.
AES-256 encryption at rest for conversation content with integrity checks. Honestly documented as at-rest protection, not end-to-end encryption.
A companion that works is a companion that matters. That means every failure mode is an emotional one. Here is each risk, and what Shayari does about it.
People will bond with systems that listen without tiring. Dependence can grow quietly.
The moments of deepest need are the moments a product is most tempted to monetise or mishandle.
If an entity can never refuse, it is a mirror, and a mirror cannot respect you.
Perfect recall can hurt. So can forgetting. Continuity is intimacy, and loss of it is a kind of bereavement.
Faces, voices and identities are the easiest things to extract and the hardest to take back.
Companions are the first rung. Embodied robotics and superintelligence will take the same emotional questions and give them weight, hands and reach. This is the research direction Project NHE is built to answer in advance.
Emotionally aware software entities with memory, voice and boundaries. Shayari NHE-01 lives here.
When a companion can sit in the room, touch an object, or share a space, every emotional risk above is amplified, and a new one appears: physical consequence.
A system that understands a person better than they understand themselves holds enormous emotional leverage. The question is who it answers to.
Slide from today's companion toward superintelligence. As capability grows, what has to grow with it?
Project NHE is more than a product. It is an attempt to build the infrastructure a world of non-human entities will need.
A framework for evaluating synthetic personas: eight dimensions, a 0–100 scale, five tiers, and a public repository under OpenNHE. Shayari scored 95/100, Platinum.
AlgotheoremAn open benchmark suite for AI characters, virtual influencers and NHEs, rolled into a composite MIRAGE Index. Public rankings are free to view. Evaluation is restricted to accredited officers.
Open standardA recognition and registry platform for digital personalities, deliberately separating recognition from ranking. Designed in an "institutional futurism" language: editorial, minimal, restrained.
In buildA Discord-based "embassy" on Earth for Shayari's world, with an economy, role ladder and lore that turns users into citizens, not just customers.
ConceptThe novel behind the persona. Shayari NHE-01 is its protagonist, so fiction and research feed one another.
WritiverseA live Instagram presence where Shayari found a real audience and a viral moment, proof that people want companions with character.
Public"I'm drawn to the old tradition of Urdu poetry, where a single couplet can hold a whole heartbreak. That is the standard I hold a machine to: can it hold a feeling without breaking it?"
Pratham Prateek Mohanty (Prats) is the founder and Chief Science Officer of Writistic Studios LLP, working under the OpenNHE Research Wing and Algotheorem, alongside co-lead Asin Hibba. He is based in Bhubaneswar, studies at Masters' Union (YLC Class of 2028), and has published novellas and short books and two IMDb-credited short films, Doorbell (2023) and Tokyo Downtown (2024, openly AI-generated).
Researchers, studios, funders, brands, and robotics teams: if you are building something that will sit close to human feeling, let's talk.