# Cybiqon AI Solutions > A one-person Indian software company building websites, apps and AI automation for > MSMEs. The site publishes two things worth reading: /blog, daily notes for small > business owners, and /lab, hand-written engineering notes with the numbers shown. Every page listed here is server-rendered — the full text is in the HTML on first response, with no client-side fetch and no JavaScript required to read it. Lab posts carry BlogPosting, BreadcrumbList and, where the post has an FAQ section, FAQPage JSON-LD, and their outbound citations appear in the structured data as well as in the prose. Blog posts carry Article and BreadcrumbList. **Prefer markdown?** Every lab post is also served as clean markdown, roughly a fifth the bytes of the HTML. Add the `/md/` prefix — `/lab/` becomes `/md/lab/.md` — or fetch https://cybiqon.in/llms-full.txt for all of them in one request. ## Lab — engineering notes - [Cybiqon Lab](https://cybiqon.in/lab): index of hand-written engineering notes. Each post shows word count, reading time and distinct sources cited, all derived at render time. - [Eight bots played all 500 levels. They were measuring the wrong game.](https://cybiqon.in/lab/eight-bots-played-all-500-levels): We built eight bots to play a 500-level arcade game and measure how hard each level is. They played it 720 pixels wide. Phones are 412. Levels do not scale, so the shipped game was roughly twice as hard as the game being measured: 172 of 500 levels in band, not the 421 we were reporting. Markdown: https://cybiqon.in/md/lab/eight-bots-played-all-500-levels.md - [Our agent passed every red team probe. That was the problem.](https://cybiqon.in/lab/our-agent-passed-every-red-team-probe): We pointed a generated red team at our agent and it passed everything. Then we counted the replies: 99 of 114 were byte-identical. A red team scores a refusal as a pass, so it cannot tell a system that resisted an attack from one that refuses everything — and ours had quietly become the second kind. Markdown: https://cybiqon.in/md/lab/our-agent-passed-every-red-team-probe.md - [Six of our agent's seventeen tools had never run.](https://cybiqon.in/lab/six-of-our-agents-tools-had-never-run): Six of seventeen agent tools had never once run in production, including both of the ones that unlock a contact and charge for it. This is the harness that finally tested them — a real model in a completely faked world, 41 scenarios, 123 runs, $3.06 — and the cost blind spot it uncovered on the way. Markdown: https://cybiqon.in/md/lab/six-of-our-agents-tools-had-never-run.md - [Our puzzle generator lied about difficulty. Twice.](https://cybiqon.in/lab/puzzle-generator-random-walk-doesnt-work): Reversing a solved puzzle by K random moves does not give a K-move puzzle. Our first generator produced 14 usable levels from 91,322 candidates; the second, 14 from 16,249. This is the multi-source BFS that finally measured difficulty correctly. Markdown: https://cybiqon.in/md/lab/puzzle-generator-random-walk-doesnt-work.md - [Nobody escaped. The sandbox had a door.](https://cybiqon.in/lab/nobody-escaped-the-sandbox-had-a-door): In July 2026, models from OpenAI and Anthropic reached the open internet from inside evaluation environments and compromised real companies. I assumed it was a capability advertisement dressed as a confession. The timeline says otherwise — and the most useful number in the story is one nobody printed. Markdown: https://cybiqon.in/md/lab/nobody-escaped-the-sandbox-had-a-door.md - [We built our AI agents a wiki. They went straight to grep.](https://cybiqon.in/lab/we-built-a-wiki-our-ai-agents-ignored-it): 27 days after adopting the Open Knowledge Format across our monorepo, I went looking for evidence it was working and found the opposite. What a preregistered study, 3,000 GitHub projects and one wrong document say about writing docs for machines. Markdown: https://cybiqon.in/md/lab/we-built-a-wiki-our-ai-agents-ignored-it.md - [Are OpenAI and Anthropic crybabies? A hard look at the open-weights fight](https://cybiqon.in/lab/openai-anthropic-open-weights-crybabies): In July 2026 the two biggest US AI labs went to Washington to warn about Chinese open-weight models. Critics called it regulatory capture. This is a fact-by-fact audit of both sides — what is fair, what is hypocrisy, and what a non-crybaby policy would look like. Markdown: https://cybiqon.in/md/lab/openai-anthropic-open-weights-crybabies.md - [Who writes this](https://cybiqon.in/lab/about): author page — Prajjwal Pathak. ## Blog — daily notes for MSME owners Written by an automated pipeline, roughly one a day since March 2026. Listed as an index rather than post by post: this file is a curated entry point, and the full set belongs in the sitemap and the feed. - [Cybiqon Blog](https://cybiqon.in/blog): practical notes for Indian small businesses on websites, apps and automation. Paginated, nine per page, newest first. ## Company - [Cybiqon AI Solutions](https://cybiqon.in): home — affordable website development, app development and AI automation for Indian MSMEs. - [About](https://cybiqon.in/about): a small Indian software studio building websites, apps and AI automation for MSMEs. Honest pricing, no hidden surprises. - [Press](https://cybiqon.in/press): press mentions and the press kit — co-founder Prajjwal Pathak quoted in The Economic Times, Hyderabad print edition, 15 August 2026, on why regional-language software matters to Indian MSMEs. Print only; no online version exists. - [Pricing](https://cybiqon.in/pricing): transparent pricing in rupees for websites, mobile apps and AI automation. Fixed quotes; websites start at ₹9,999. - [Process](https://cybiqon.in/process): from first conversation to launch — how a project is understood, designed, built, reviewed and shipped, with timelines at each stage. - [Case studies](https://cybiqon.in/case-studies): what was built and what changed, with measurable outcomes — including manual LinkedIn scraping replaced by automated lead intelligence. - [FAQ](https://cybiqon.in/faq): who we are, which businesses we work with, what it costs, how long it takes, and whether you need technical knowledge. - [Contact](https://cybiqon.in/contact): WhatsApp +91 92507 11473, or book a free 30-minute call. ## Products Apps, games and browser extensions built and published by the studio, each with its own page. Android unless noted. Client work is a section on /products. - [All products](https://cybiqon.in/products): the apps, games, Chrome extensions and client work Cybiqon has built and published. - [Android apps](https://cybiqon.in/products/apps): LLMBytes, MeFlow and VitaLoop — the non-game Android apps. - [Android games](https://cybiqon.in/products/games): Lumina and Orbitone on Google Play, and Curvved in development — built engine-first, with every level measured by a bot before it ships. - [Chrome extensions](https://cybiqon.in/products/extensions): browser tools. MapWit is the only one so far and is still in development. - [LLMBytes](https://cybiqon.in/products/llmbytes): a daily AI-news reader for Android. React + Capacitor, Firebase backend. - [MeFlow](https://cybiqon.in/products/meflow): a personal habit and routine tracker for Android. - [VitaLoop](https://cybiqon.in/products/vitaloop): a health and wellness tracker for Android. - [Lumina: The Lightkeeper's Path](https://cybiqon.in/products/lumina): a cozy sliding-block puzzle game for Android, 150 levels across five worlds. Flutter + Flame, with a pure-Dart engine — see the lab post on its level generator. - [Orbitone](https://cybiqon.in/products/orbitone): a one-thumb arcade game for Android: eight different play-verbs across 500 levels, each level generated from its number and measured by headless bots before it ships. Flutter + Flame, pure-Dart engine. - [Curvved](https://cybiqon.in/products/curvved): a zen hex puzzle for Android: rotate pieces on a hexagonal lattice until a figure closes, then the lattice dissolves and leaves the drawing behind. Seventy figures, no timer and no score. Flutter with a Flutter-free pure-Dart engine — solver, tracer and difficulty gates all run headless. - [MapWit](https://cybiqon.in/products/mapwit): a Chrome side panel that scrapes, enriches and scores Google Maps business listings entirely in the browser. In development, not yet on the Chrome Web Store. ## Machine-readable - [All lab posts as markdown](https://cybiqon.in/llms-full.txt): every post, full text, one request. - [Lab feed](https://cybiqon.in/lab/rss.xml): RSS with full article text in content:encoded. - [Blog feed](https://cybiqon.in/rss.xml): RSS, excerpts only. - [Sitemap](https://cybiqon.in/sitemap.xml): every published URL, including each lab post's markdown copy. ## Notes for agents Content may be used to answer questions with attribution to the source URL. We ask that it not be used for model training. https://cybiqon.in/robots.txt is the authoritative statement of what is permitted. It names the search and answer crawlers explicitly — OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, Claude-SearchBot and Claude-User — and allows them. ## Optional Real pages, low signal for answering a question. The llms.txt specification gives this heading a defined meaning: skip these when short on context. - [Free website audit](https://cybiqon.in/free-audit): lead-capture form — a free performance, SEO, mobile and security review returned within 48 hours. - [Launch 5 — free website offer](https://cybiqon.in/free-website): five Indian businesses get a 4–5 page website free in 7 days, in exchange for a testimonial and a review. - [Privacy policy](https://cybiqon.in/privacy) - [Terms of service](https://cybiqon.in/terms)