---
title: "Principles"
description: "Working principles for how the work is built and described."
canonical: "https://www.ayushrai.site/principles/"
last-updated: "2026-10-16"
---

# Principles

Source: https://www.ayushrai.site/principles/
Markdown: https://www.ayushrai.site/principles.md

Working principles for how the work is built and described.

### Usefulness over hype

If a sentence could appear on any AI startup homepage, it probably shouldn't appear on mine.

In practice: Rewrote Expert-O and portfolio stats to drop '5x', '100% success', and '24/7' theater.

### Ship to learn

A live MVP with open questions beats a perfect deck that never meets users.

In practice: AyushMushrooms ships as an MVP and Open Framework as a front-end prototype, both with explicit unfinished edges.

### Evidence over adjectives

Prefer links, dates, and roles over 'visionary', 'elite', or 'transformative'.

In practice: Research and leadership claims that lack evidence are held back until confirmed, or omitted.

### Interdisciplinary leverage

Crossing AI, web, and domain context is useful when it clarifies the problem, not as a personality brand.

In practice: Civil society DPI and agrotech work borrow product + compliance thinking, not buzzword stacks.

### AI for problems, not decoration

Use models where evaluation and feedback loops matter; don't sprinkle 'AI-powered' on static pages.

In practice: RLHF day job stays evaluation-heavy; marketing sites don't fake AI features.

### Trust is a product requirement

Consent, auditability, and progressive verification are features, especially in civil society tooling.

In practice: Open Framework treats consent, verification, and audit trails as design goals from the start; the current prototype does not implement them yet.

### Clarity before complexity

Name the job of the page or system before adding motion, 3D, or platform sprawl.

In practice: Client sites (Iha, Praful) prioritize hierarchy and contact paths over stack peacocking.

### Systems over features

Prefer durable loops (lead → datasheet → CRM, peer acknowledgement) over feature checklists.

In practice: AyushMushrooms and Shiksha-Mitra are framed around loops, not feature grids.

_Unverified award metrics are omitted. Wording matches the public site._