Blog
Practitioner notes across four pillars. Browse by category or read the latest below.
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.
All posts
- Claude Desktop vs Antigravity 2026: Why I Moved Back An honest comparison of the Claude Desktop App and Antigravity. Why I dropped my $100/mo Claude habit for a $20/mo Gemini setup to ship AI SaaS solo.
- Gemini Thinking Levels: Deciphering the New $200/mo AI Agentic Tax Google's 2026 pricing shift ends flat-rate AI. I analyze the new 'Thinking Levels' and why the $200/mo Deep Think tier is the first real agentic tax.
- The 'Steal Your Competitor's SEO With AI' Trick, Tested A viral tweet says you can steal any competitor's SEO strategy in 5 minutes with AI and their sitemap. I ran it on a real rival. Here's what it actually misses.
- AI Cost Allocation: Models That Work in Indian Production AI cost allocation explained: showback vs chargeback, dimension models that actually scale, IAM tag enforcement, and the Indian TP and Ind AS 108 overlays.
- The 2026 AI Spend Disclosure Audit: 46 Companies Graded We scored 46 public companies on how well they actually disclose AI spend in filings. Only 2 cleared a 'Solid' bar. Apple and Amazon disclose nothing specific.
- Section 195 TDS on Foreign AI Vendors: Complete India Guide Section 195 TDS mechanics on OpenAI, Anthropic, AWS Bedrock payments. Engineering Analysis SC, DTAA rates, Form 15CA/15CB workflow, IT Act 2025 transition.
- AI Vendor Consolidation in 2026: When to Cut, When to Hold AI vendor consolidation explained: why 68% of CIOs prioritise it, the trade-offs nobody talks about, and the Indian tax overlay that changes the math.
- The FinOps Foundation Framework: A Practitioner's Walkthrough FinOps Foundation Framework explained — Inform/Optimize/Operate phases, capabilities, personas, 2026 Executive Strategy Alignment update, AI extension.
- AI Systems Review: A 6-Domain Framework You Can Run Internally Six-domain framework for auditing AI spend, vendor consolidation, model selection, controls, governance, and cost allocation at Indian mid-market scale.
- AI Tax Recovery for Indian CFOs: The 5-Domain Framework GST input credit, Section 195 TDS, RCM, capex/opex, transfer pricing — the five tax-recovery domains for Indian companies with material AI/cloud spend.
- Section 195 vs Equalisation Levy: Foreign AI Vendor TDS Equalisation Levy abolished. Section 195 TDS + DTAA + Engineering Analysis is the new framework for foreign AI vendor payments. Transition reconciliation guide.
- The Indian AI Search Audit: 500 Queries, 25 Brands, 777 Citations 25 Indian brands audited across 500 AI search queries. The most-cited domains weren't brand websites — they were Instagram, aggregators, and one indie blog.
- AI Software Capex vs Opex in India: The Ind AS 38 Test Most AI spend is opex by default — SaaS, API consumption, fine-tuning third-party APIs. The narrow capex window explained for Indian CFOs and auditors.
- FinOps vs ITFM vs ITAM: Why You Eventually Need All Three FinOps, ITFM, and ITAM aren't competing — they cover different scopes. The 2026 convergence thesis explained, with Indian mid-market sequencing guidance.
- AI / Cloud FinOps Providers in India: A Practitioner's Take Honest comparison of AI / cloud cost optimisation providers in India — CloudKeeper, Apptio, Big-4, cloud-native shops, boutique advisory. Who fits what.
- Why LLM Gateway Attribution Is Harder Than Cloud FinOps Ever Was Cloud FinOps had two stable attribution units. LLM spend has N freeform dimensions, two failure modes nobody talks about, and a hidden cost in reporting UX.
- GEO vs AEO vs AIO vs SGE: A Plain-English Glossary for 2026 The AI-search alphabet soup is mostly the same idea in different industry hats. Plain-English definitions plus the one distinction that actually matters.
- AI Search Visibility Tools: Honest Comparison From a Builder I built one of these (Citare) and tried every other one. The category splits on a hidden question most buyers don't know to ask. Honest comparison, May 2026.
- Best MCP Servers for Claude Code in 2026 (Honest Picks) Tried 20+ MCP servers, kept 12, cut 8, found a gap. Honest picks for daily Claude Code use with config snippets — by someone who shipped 37 MCPs.
- 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.
- How I Run 3 Production AI SaaS on $5/Month of Hosting My complete 2026 stack — Cloudflare, Supabase, Paddle, Razorpay, Claude — and why hosting costs nothing while AI does. Honest picks from shipping.
- What is LLM FinOps? The Missing Discipline for AI-Era Companies LLM FinOps is the discipline of managing AI/LLM spend like serious infrastructure cost. Three frames — solo builder, engineering org, CFO — explained.
- Windsurf Review 2026 — Not For Solo Founders, Great For Small Teams Honest Windsurf review from a solo founder who tried it briefly: multi-model IDE that's better suited to small teams than to solo agentic-AI builders.
- Hello from rikuq — a practitioner blog for solo AI SaaS founders I'm a non-coder who shipped 3 production AI SaaS solo. rikuq is the practitioner blog of everything I've learned. Here's the intro.
- Antigravity Review (May 2026) — From Daily Driver to Dropped I used Antigravity heavily for months, then dropped it after the May 2026 redesign. Honest review of what worked, what broke, and who should still try it.
- Best AI Coding Tools 2026 — Honest Picks From Shipping 3 SaaS Solo Claude Code, Cursor, Antigravity, Windsurf, Copilot — honest picks from a solo founder shipping 3 production AI SaaS. What I actually use, what I dropped.
- Claude Code Review 2026 — From Zero Code to 3 Live SaaS Honest Claude Code review from a non-coder who shipped 3 production SaaS with it. My $200/mo Max setup, the killer moment, the failures.
- Cursor Review 2026 — Honest 'Not For Me' Take From a VSCode User Why I didn't stick with Cursor — and why that's a recommendation, not a complaint. Honest review from a founder who ships 3 SaaS in VSCode + Claude.
- Cursor vs Claude Code 2026 — You're Probably Asking the Wrong Question Cursor vs Claude Code, honest comparison from someone who shipped 3 SaaS solo on Claude. The framing most reviews miss — and the answer that matters.