Intelligence • Structure • Scale

Ballast

Stability beneath intelligence.

An independent AI product studio in India, building intelligent products on a stable, scalable foundation. The model handles language. Everything a user has to trust — progression, pricing, limits, safety — is deterministic code underneath it.

FIG.01 — Stability beneath intelligence SYSTEM VIEW
INTELLIGENCE — FLUID, HUMAN, MODEL-DRIVEN FREE TEXT VOICE HINGLISH CHAT PROCESSING DETERMINISTIC BALLAST RULES STATE LIMITS AUDIT THE WEIGHT THAT KEEPS THE PRODUCT STABLE
01

Products

Each one ships on its own subdomain
Live · public beta

Fitness · B2B2C

Arc

Just talk. Arc does the rest. Text it what you ate and how you trained in plain Hinglish, and it logs the macros against real Indian food data, writes your journal for you, and hands back the next session tuned to how the last one felt. When a dish or a portion looks off, it flags it instead of guessing.

arc.rishabhshukla.dev →
  • One text box: Hinglish in, structured log out
  • 900+ Indian dishes matched to measured nutrition data
  • Adaptive sessions tuned by easy / right / hard
  • Flags a doubtful match instead of guessing
Stack · FastAPI · deterministic AI
In beta

Developer tooling

Product Two

Placeholder. Drop your next product here: a one-line what-it-is, who it's for, and the subdomain it ships on. Keep the same shape as Arc so the lineup reads as a real studio pipeline.

two.rishabhshukla.dev →
  • Key capability one
  • Key capability two
  • Key capability three
Stack · edit me
In the lab

Exploration

Next up

Placeholder. A studio always has something on the bench. Tease the direction here without over-promising. “In the lab” signals momentum and a real pipeline to anyone reading, including future hires or partners.

Subdomain reserved
  • What problem it explores
  • Why now
  • Who it's for
Stack · TBD
02

How we build

The studio's point of view
01

Clarity over noise

The model handles language. Anything a user must trust — progression, pricing, limits, safety — is deterministic code. That split is the ballast, and we say exactly which side a feature sits on.

02

Systems thinking

AI drafts; a person approves. Approval gates, consent, and revocation are first-class in the schema, not bolted on after a demo. Every product is a system before it is a screen.

03

Built to last

We put products in front of real users early, with clear success and kill criteria, and let the market — not our own optimism — set the next move. Durability over novelty.

04

Practical intelligence

Hinglish input, INR pricing, and domain grounding for local realities that Western-first tools handle poorly. The details incumbents skip are the wedge.

03

Founder

One builder, shipping

Rishabh Shukla

AI Systems Architect · Founder of Ballast

AI Lead by day, architecting production AI systems for enterprise teams: multi-tenant agentic platforms, permission-aware retrieval, natural-language analytics over enterprise data, and the observability and guardrail infrastructure that keeps them safe to operate at scale.

Ballast is where those same patterns become whole products instead of someone else's platform, under one point of view: the AI should be delightful to talk to and boring to depend on.

Have a product that has to work in the real world?