Skip to main content

The Gritty Reality

This is not a business plan. There are no deadlines, no investors, no corporate backing. This is a personal backlog from a solo developer who builds this in their spare time — between a full-time job, life, and everything else.

What This Document Is​

A plain, honest list of things that need to be built before BijMantra can be genuinely useful to a plant breeder in the real world.

No promises. No timelines. No success metrics. Just work.

When something gets done, it will be marked done.

Stage 1: Make It Actually Work​

The foundation. Nothing beyond this matters until these are solid.

  • All 300+ pages are fully functional with no demo data
  • BrAPI v2.1 endpoints are validated and returning real data
  • The application is deployable by someone other than the developer
  • Basic authentication and account management works reliably
  • The mobile experience is usable in a field environment
  • Offline data capture syncs correctly when connectivity returns

Stage 2: Make It Actually Useful​

Features that turn a working app into a genuinely useful tool for breeders.

  • REEVU AI assistant is trained on real agronomic and breeding literature
  • Yield prediction models are validated against real trial data (not simulations)
  • Cross prediction gives explainable recommendations a breeder can trust
  • Computer vision pipeline works on farmer-grade phone photos
  • Climate scenario modelling is connected to real regional data sources
  • Dashboard builder lets scientists configure their own reporting views

Stage 3: Make It Actually Open​

Interoperability and community scaffolding.

  • BrAPI certification passes completely (all endpoints, all versions)
  • Data can be exported to R and Python without custom scripts
  • Integration with at least one existing institutional LIMS is documented
  • A contributor can set up a local development environment in under 30 minutes
  • At least one external researcher has used BijMantra for real work
  • A plugin/extension mechanism exists so others can contribute modules

Stage 4: Things That Would Be Nice​

Lower priority. Will happen eventually.

  • iOS and Android native apps (beyond the PWA)
  • 20+ language support, including Hindi, Swahili, Bengali
  • IoT sensor integration (MQTT, LoRaWAN)
  • Drone imagery pipeline
  • Carbon footprint tracking per variety trial
  • Genomic data integration (VCF, PLINK, GWAS pipelines)
  • Marketplace for community-built analysis modules

The Honest Constraints​

Time: This is a side project. Progress is measured in focused evenings and weekends, not sprints.

Money: There is no funding. Infrastructure costs come out of pocket.

Team: There is no team. It is one developer and a set of AI agents.

Users: There are no paying users. There are no beta users. There are currently zero users.

Scope: The scope is almost certainly too large for one person. That is known and accepted.

What Would Change Things​

If any of the following happened, the pace of development would change significantly:

  • A plant scientist or agronomist starts collaborating on the domain logic
  • A grant materialises to fund dedicated development time
  • A university or research institute adopts it for real trials and provides structured feedback
  • A developer contributor appears who understands breeding workflows

Until then, it moves at the speed it moves.

What Has Been Built​

The following is working today, even if imperfect:

  • A full-featured React/PWA frontend with 300+ pages and navigation
  • A FastAPI backend with 1,300+ endpoints under active development
  • BrAPI compatibility layer in progress
  • REEVU AI assistant (cloud API backend, configurable providers)
  • Offline-first architecture with local-first data storage
  • Cloudflare tunnel support for zero-infrastructure public access
  • A public documentation site (this site)

The machine is running. It just needs time.

Last updated: March 2026

Jay Shree Ganeshay Namo Namah 🙏