A practitioner blog by Ravi · solo founder
Solo founder.
Three AI SaaS shipped.
Here’s how.
I ship production AI products solo — an OpenAI-compatible gateway, an AI-search visibility platform, and an Indian compliance marketplace. This blog is the playbook: tool reviews, infra deep-dives, and case studies from someone actually building, not theorizing.
What I write about
All posts →
AI Coding Tools
Reviews and comparisons of Cursor, Claude Code, Windsurf, Antigravity, Copilot, and the rest — from someone actually shipping with them.
LLM Infrastructure
AI gateways, caching, model routing, observability, edge replication. Practitioner deep-dives from building Prism.
LLM FinOps
AI spend management as a discipline. Cost attribution, governance, optimization patterns, Indian tax structure on AI spend.
GEO (AI Search)
How ChatGPT, Claude, Perplexity, and Google AI Overview actually index and cite content. From building Citare.
Stack & Ops
Supabase, Vercel, Cloudflare, Redis, Upstash — the production stack for shipping AI SaaS solo.
Essays
Personal notes, manifestos, lessons learned. Meta-content about the work itself.
Latest
See all posts →
Claude Code Hooks vs Skills: When to Use Which
Hooks and Skills are different abstractions for different jobs, not interchangeable. Plain-English decision guide with real examples from a production setup.
How I Built Citare V2 in 12 Days After Throwing V1 Away
I underestimated Citare badly. V1 shipped in a weekend, satisfied no one, got scrapped. V2 took 12 days, 76k LOC, and is still iterating.
The Four-Index Reality: Why AI Search Isn't One Thing
ChatGPT uses Bing. Gemini uses Google. Claude uses Brave. Perplexity uses its own. Tracking AI visibility means tracking four indexes, not one.
GEO vs SEO in 2026 — What Google's May Guidance Changed
Google said AEO and GEO are 'still SEO.' Half-true. Heres the honest read on what changed on Google, what didnt on the other engines, and what to do.
Portkey vs Helicone vs LiteLLM vs OpenRouter: Honest Comparison
Honest comparison of the 4 leading LLM gateways in 2026, plus where Prism enters as a new credible alternative. Updated for Fusion + edge replication.
Anthropic Prompt Caching: Real Numbers From 330 Production Calls
Real first-party data: Anthropic prompt caching cut Citare's AI bill 25-35% on parsing-heavy workloads. What works, what doesn't, what burned me $20.
GitHub Copilot Review 2026 — Built For Enterprise, Not Solo Founders
Brief-trial GitHub Copilot review from a solo founder who uses Claude Code daily. Honest verdict: enterprise tool. Not for solos. Here's who fits.
