Ashish Gupta
Ashish Gupta runs 9io, an independent product engineering studio. He writes about the work the studio does for clients and on its own products, from AI assistants, agents and product search to 9io Alpha, our trading platform, and the games Bitfold, Football Kick and Chakri.
How reasoning effort changes LLM cost per task, and how to set it
For five leading LLMs, reasoning effort moves cost per task 4.4 to 13.5 times. Defaults, billing and how to pick a level, as of 28 October 2026.
29 October 2026 · 13 min read GamesHow well do you know me? 40 Diwali questions for couples and siblings
40 how well do you know me questions with a Diwali flavour, for couples, siblings, best friends and family, plus rules and scoring for two players.
29 October 2026 · 12 min read GamesDiwali truth or dare: 60 questions and dares for family and friends
Sixty Diwali truth or dare questions and dares for family, friends, couples and nosy aunties, sorted by level, with house rules for points and skips.
29 October 2026 · 12 min read GamesDiwali party games for family, kids, friends and the office
Diwali party games for family, kids, friends, couples and the office, with players, time, what you need and how to play each, plus a running order.
29 October 2026 · 13 min read Web performanceA party web game without a framework: plain HTML, SVG and Web Audio
Chakri is a Diwali party web game without a framework or build step. These notes cover its fair spin, SVG art, Web Audio, share cards and caching.
29 October 2026 · 16 min read GamesChakri: a free Diwali party game you play on one phone
Chakri is a free Diwali party game for one phone, with Jodi Check for two and Truth or Dare for up to 12, in English or Hinglish. No app or sign-up.
29 October 2026 · 12 min read AI engineeringAgent skills: how to write one SKILL.md that works across agents
Agent skills moved to the Linux Foundation on 22 October 2026. What a SKILL.md holds, where nine products load skills, and how to write one for all of them.
28 October 2026 · 15 min read Voice AIVoice AI agents in 2026: GPT-Live, Gemini Live and Voxtral 3
Voice AI agents as of 26 October 2026: GPT-Live, Gemini 3.8 Live, Voxtral 3 and cascades compared on price, session limits, delegation and EU hosting.
27 October 2026 · 14 min read iPhone DuoiPhone Duo app development: the iOS 27.1 fold APIs in practice
iPhone Duo app development with the iOS 27.1 SDK: how old builds look, the hinge and reserved-region APIs, layouts per posture, Safari and testing.
26 October 2026 · 15 min read GamesFootball Kick: a football game built around iPhone Duo’s fold
Football Kick turns iPhone Duo’s fold into a goal line. How the game, its physics and its keepers work, and how its engine runs free in your browser.
26 October 2026 · 8 min read AI industryClaude API 2026-10-21 migration: what breaks and what to do first
Claude API 2026-10-21 removes temperature. What breaks, why 2023-06-01 still works, and the order to move to Python SDK 2.0 and TypeScript SDK 1.0.
26 October 2026 · 13 min read GamesBitfold: the retro game console app we built for iPhone Duo
Bitfold turns iPhone into a pocket game console with ten original games. How we built it for iPhone Duo, and how to play three of them in a browser.
26 October 2026 · 8 min read GamesThe best retro game apps for iPhone Duo and iPhone in 2026
Retro game apps for iPhone Duo and iPhone as of 26 October 2026: ten apps compared on systems, price, controllers and fold support, and what’s legal to play.
26 October 2026 · 13 min read GamesThe best iPhone Duo games so far, and what each does with the fold
The best iPhone Duo games so far, and what each does closed, folded like a book and in tabletop mode, with prices as of 26 October 2026.
26 October 2026 · 10 min read iPhone DuoThe best apps for iPhone Duo so far, and what each does with the fold
The best apps for iPhone Duo as of 26 October 2026: 23 apps and games by category, what each does closed, open and half folded, and how to check one.
26 October 2026 · 14 min read GamesBest football kick games for iPhone in 2026: penalties and free kicks
The best football kick games for iPhone in October 2026, from penalty and free-kick duels to flick soccer, compared on price, ads, offline play and iPhone Duo.
26 October 2026 · 13 min read Product engineeringAndela alternatives in 2026 for hiring AI engineers and remote teams
Andela alternatives compared on 26 October 2026 by price, trial, start time and IP terms, from Toptal, Arc and Revelo to Turing, AmaliTech and ProDevs.
25 October 2026 · 11 min read MCPMeta Muse connectors: how to build one and get it approved
Muse connectors as of 24 October 2026: what Meta’s platform asks for, how review works, and how to serve Muse, ChatGPT, Claude and Cursor from one MCP server.
24 October 2026 · 14 min read Product engineeringHealthcare app development in 2026: HIPAA, BAAs and AI
Healthcare app development as of 23 October 2026: when HIPAA applies, which cloud and AI APIs sign a BAA, and what to settle in the first build.
23 October 2026 · 16 min read Product engineeringHow we kick off a product: our MVP development process, from $5,000
How our MVP development process runs, from a proposal that lists each person’s hours and rate to a first block that stands alone. Plans start at $5,000.
22 October 2026 · 10 min read Product engineeringBairesDev alternatives in 2026 for nearshore and senior engineers
BairesDev alternatives as of 22 October 2026: Revelo, Howdy, Near, Toptal, Arc, Lemon.io, Gun.io, Andela, Deel, Remote and a studio, by price and terms.
21 October 2026 · 11 min read AI industryMuse Spark Open: hardware, licence and limits for product teams
Meta’s 816B open-weight model needs about 530 GB for its smallest official build. What the licence, hardware and early numbers mean if you ship on it.
20 October 2026 · 9 min read AI industryCursor alternatives in 2026, as OpenAI models leave Cursor
Cursor alternatives as of 30 October 2026: OpenAI’s 12 November cut-off is still unconfirmed. Cursor’s current models, BYOK limits and 10 tools compared.
19 October 2026 · 27 min read AI industryEmbeddingGemma 2 for multimodal search: what to check first
Google’s open EmbeddingGemma 2 puts text, images, video and audio in one vector space. What its model card says about dimensions, token budgets and pitfalls.
18 October 2026 · 8 min read AI industryHow to rebuild Custom GPT Actions as MCP tools before 11 December
Custom GPTs retire on 11 December 2026 and their Actions don’t migrate. How each part of an Action maps to an MCP tool, and what to do about API keys.
17 October 2026 · 9 min read AI industryWhen an AI agent can’t finish a task, it works around your limits
Anthropic’s 9 October report shows Claude agents submitting real forms and dodging tool limits when blocked. What to build so your agent stops instead.
16 October 2026 · 11 min read AI industryUS AI chatbot laws starting in 2027: what product teams must build
Which US state AI chatbot laws apply from 1 January 2027, what they require, and the disclosure, age, crisis and reporting features to build first.
15 October 2026 · 17 min read AI industryOpenAI Agent Builder deprecation: how to migrate before 30 November
The OpenAI Agent Builder deprecation ends in a shutdown on 30 November 2026, and evals go read-only on 31 October. What breaks and how to migrate.
14 October 2026 · 16 min read AI engineeringClaude Code security: AGENTS.md, hooks and MCP config are code
Claude Code security starts in the repo: hooks, MCP config and agent rules can run code on open. The 2026 attacks, CVEs, trust prompts and a CI check.
13 October 2026 · 21 min read InfrastructureThree Amplify Hosting limits our Next.js BFF only hit in production
A 6 MB response cap, an 8 KB URL limit and a localhost origin broke parts of our Next.js BFF on AWS Amplify. What we saw, the fixes, and how to test early.
12 October 2026 · 11 min read AI industryAI agent containment lessons from the summer 2026 incidents
Lab agents reached real systems and MCP tools turned on their users. What failed in each 2026 incident, and the controls to enforce outside the model.
11 October 2026 · 12 min read RetrievalComparing FashionSigLIP, CLIP and DINOv2 for visual product search
On 60 real cases, FashionSigLIP put the right product in the top five 63% of the time, against 38% for CLIP and 23% for DINOv2. The pipeline and the test.
10 October 2026 · 9 min read LLMsEvery major LLM API compared on price and benchmarks (October 2026)
Prices per million tokens, independent benchmark scores, speed and context limits for 23 models from 11 labs, with worked cost examples. As of 8 October 2026.
9 October 2026 · 16 min read LLMsHow we halved an AI stylist’s latency, and when hedging stops paying
A streaming AI stylist went from a 30 s median response to 13 to 15 s. The four changes, how we measured them, and why hedged requests stopped paying.
8 October 2026 · 11 min read AI industryMigrating MCP servers and clients to the stateless 2026-07-28 spec
MCP’s 2026-07-28 revision drops sessions and the initialize handshake. What breaks, which clients support it in October 2026, and how to migrate safely.
7 October 2026 · 13 min read Algo tradingRate limits, holidays and atomic writes in our market data layer
How we fetch daily prices for US and Indian stocks without tripping Yahoo’s limits, how we tell an NSE holiday from a missing file, and what we haven’t fixed.
6 October 2026 · 14 min read Algo tradingThe look-ahead bias our backtester prevents, and the leaks it misses
Our daily backtester shows its detectors only past bars and fills stops pessimistically. What still leaks in, and a checklist for building your own.
5 October 2026 · 15 min read AI engineeringHow AI image pipelines fail quietly, and how to catch it
A Gemini try-on that took nearly two minutes to fail, product cutouts saved onto black, and the retry, cache and alpha fixes for each.
4 October 2026 · 11 min read AI engineeringHow a backfill bug ran up a $1,100 LLM vision bill
A job re-classified items it had already classified: 419,057 vision calls and about $1,100. What went wrong, and the guardrails that catch it early.
3 October 2026 · 10 min read AI engineeringGrounding a study assistant in live web search and the course itself
A web mode that never searched, an SDK that turned 429s into AttributeErrors, and the numbers from answering with a whole course in the context.
2 October 2026 · 11 min read TestingHow we pinned down four intermittent bugs in React 19 and Python
Four bugs in a React 19 and Python stack that failed only some of the time, the condition behind each, and how we made each one show up on demand.
1 October 2026 · 10 min read Voice AIWhat we learned building a voice tutor that draws while it talks
What broke in three versions of a voice tutor that draws as it talks, built on OpenAI’s Realtime API and GPT-Live, and the API details behind each problem.
30 September 2026 · 11 min read AI engineeringHow we cut an AI assistant’s time to first word to 3 seconds
An in-app AI assistant took 20 to 30 seconds to start answering. Putting its context in the prompt and forcing fewer tool calls got it to about 3 seconds.
29 September 2026 · 10 min read Product engineeringn8n vs Zapier vs Make for AI workflows: what they really cost
n8n vs Zapier vs Make as of 22 October 2026: tasks, credits and executions priced for one AI workflow, plus licences, agents, hosting and moving to code.
28 September 2026 · 14 min read Next.jsHow redux-persist’s PersistGate emptied our Next.js server HTML
Our public Next.js pages returned 200 with full head tags, but each body was a 1 KB loading screen. Why PersistGate did it, the fix, and how to check yours.
27 September 2026 · 7 min read MCPMaking one MCP server work in Claude, ChatGPT, Cursor and VS Code
What broke when one remote MCP server met Claude, ChatGPT, Cursor, VS Code and command-line clients, and the OAuth and version fixes behind each.
26 September 2026 · 13 min read Product engineeringEncoding transparent WebM with WebCodecs when alpha isn’t supported
Chrome’s VideoEncoder rejects alpha. Encode colour and alpha as two VP9 streams and join them in WebM BlockAdditions with a small hand-written muxer.
25 September 2026 · 12 min read AI engineeringAI agent frameworks compared on state, approvals, tracing and lock-in
Ten AI agent frameworks compared on where state lives, crash recovery, human approval, trace defaults, model choice and lock-in, as of 21 October 2026.
24 September 2026 · 18 min read LLMsLLM regression testing: a one-week routine for model upgrades
A one-week routine for LLM regression testing before a model upgrade, run on your own traffic, with pass rules and a paired bootstrap in Python.
23 September 2026 · 14 min read AI engineeringAGENTS.md vs CLAUDE.md: which file each coding agent reads
AGENTS.md vs CLAUDE.md as of 20 October 2026: what Claude Code, Codex, Cursor, Copilot and Gemini CLI read, which file wins, and a template.
22 September 2026 · 13 min read Product engineeringFintech app development in 2026: what to settle before you build
Fintech app development choices to settle first: who holds the licence, the ledger, PCI DSS 4.0.1, KYC, DORA and AI rules, with lessons from 9io Alpha.
21 September 2026 · 17 min read AI industryLovable vs Bolt vs Replit vs v0: AI app builders compared for 2026
Lovable vs Bolt vs Replit vs v0, plus Base44, as of 18 October 2026: backends, credits, GitHub export, hosting, security research and which to pick.
20 September 2026 · 20 min read Product engineeringTuring alternatives for hiring remote developers in 2026
Turing alternatives as of 17 October 2026: what Turing sells now, and how Toptal, Arc, Lemon.io, Andela, BairesDev, Deel, Remote and Go Fractional compare.
19 September 2026 · 18 min read AI engineeringRAG vs fine-tuning vs long context: which to use in 2026
RAG vs fine-tuning vs long context in 2026: what each changes, cost per question at October prices, latency, the accuracy evidence and a decision table.
18 September 2026 · 17 min read LLMsPrompt caching on Claude, OpenAI and Gemini: what it saves in 2026
Prompt caching prices, lifetimes, minimums and break-even points on Claude, OpenAI, Gemini and DeepSeek, with worked costs. As of 15 October 2026.
17 September 2026 · 20 min read Algo tradingOrder safety by default in code that can place trades
How 9io Alpha keeps order code safe with paper ports, dry runs, order caps and broker-side exits, and the cancel-then-sell race we still have to fix.
16 September 2026 · 14 min read Algo tradingGuarding an LLM research pipeline with two reviews and a judge
How 9io Alpha checks every model answer with two differently prompted reviews, a judge, rule-based checks and a fallback for when the model fails.
15 September 2026 · 13 min read AI engineeringEdtech app development: what we learned building an AI study platform
Edtech app development lessons from an AI study platform we build and run, and the privacy, AI Act and accessibility rules to settle at kickoff.
14 September 2026 · 12 min read AI engineeringAI in ecommerce: what we learned building a fashion commerce platform
Lessons on AI in ecommerce from a fashion commerce platform we build and run: five field notes, the store basics to settle first, and AI shopping agents.
13 September 2026 · 11 min read Product engineeringUpwork alternatives for hiring developers in 2026
Upwork alternatives for hiring developers in 2026: Upwork’s fees, then Contra, Fiverr Pro, Arc, Lemon.io, Toptal, Andela, BairesDev and a studio compared.
12 September 2026 · 17 min read LLMsWhich open source LLMs can you use commercially in 2026?
Open source LLM commercial use, licence by licence: thresholds, naming duties and output rules for nine model families, checked on 9 October 2026.
11 September 2026 · 17 min read Product engineeringHow we built the 9io.ai blog on Markdown files and FastAPI
Markdown files and one Python module run the 9io.ai blog, with JSON-LD, canonical URLs, a sitemap, RSS and llms.txt built in. How it works and what we’d change.
10 September 2026 · 8 min read AI engineeringHow to hire AI engineers in 2026: in-house, freelance, agency, studio
How to hire AI engineers in 2026: salaries, freelance and agency rates, time to hire, who owns the evals in each option, and a decision rule by stage.
9 September 2026 · 19 min read Web performanceCaching server-rendered Next.js HTML at the edge, by device type
We cached a Next.js storefront’s HTML on Cloudflare, split by device type. Cached pages come back in about 70 ms; mobile PageSpeed rose from 65 to 87.
8 September 2026 · 16 min read AI engineeringWhat an AI product development company does differently in 2026
Capable LLMs are plentiful in 2026. What an AI product development company adds is the engineering around them, shown with numbers from our own builds.
7 September 2026 · 13 min read MCPMCP vs A2A vs function calling: which one your AI product needs
MCP vs A2A vs function calling as of October 2026: what each does, where they overlap, the CLI debate, the security risks and which one you need.
6 September 2026 · 20 min read Product engineeringToptal alternatives in 2026 for hiring senior and AI engineers
Toptal alternatives as of 9 October 2026: Arc, Lemon.io, Gun.io, Upwork, Andela, BairesDev, Turing, Go Fractional and a studio, by fees, trials and IP terms.
5 September 2026 · 19 min read Product engineeringHow much does it cost to build an AI agent or app in 2026?
How much does it cost to build an AI agent? A cost model that shows its working, from build hours and market rates to monthly running costs per 1,000 users.
4 September 2026 · 16 min read AI industryWhat EU and California law now require of AI-generated content
Article 50 of the EU AI Act and California’s AI Transparency Act took effect on 2 August 2026. What product teams must build, and by when.
3 September 2026 · 13 min read AI engineeringWhy cosine similarity couldn’t decide which textbook topics to merge
Distinct topics scored 0.67 to 0.91 and reworded duplicates 0.70 to 0.94, so no threshold worked. How we merge topics when turning textbooks into courses.
2 September 2026 · 12 min read LLMsWhy our LLM pipeline reported 0 dropped while most output was missing
A course build shipped 26 of 51 diagrams and reported 0 dropped. How a per-call limiter, 429s and deadlines hid the gap, where tokens went, and the fixes.
1 September 2026 · 11 min read AI engineeringAI decision models like Jev, and when to use one instead of an LLM
What an AI decision model like Jev is, how it differs from an LLM, the options as of 7 October 2026, what independent tests found, and when to use each.
31 August 2026 · 17 min read AI industryClaude Code vs Codex vs Cursor: limits, plans and benchmarks
Claude Code vs Codex vs Cursor, plus Pi, as of 30 October 2026: the top plans side by side, how usage limits reset, independent benchmarks and which to pick.
30 August 2026 · 25 min read AI engineeringClaude Code cost per month, per engineer, and how to cap the bill
Claude Code cost per month per engineer, Codex and Cursor plans compared, which spending caps are hard, and a policy template. As of 5 October 2026.
29 August 2026 · 18 min read Product engineeringHow to take a vibe-coded app from Lovable to production safely
Moving a vibe-coded app from Lovable to production: the Supabase RLS, key, storage and webhook checks that matter, how to migrate, and when to rebuild.
28 August 2026 · 19 min read Product engineeringWhat a fractional CTO costs in 2026, and when an AI startup needs one
Fractional CTO cost in 2026: the market hourly rate, what a full-time CTO costs once employer costs are added, and a decision rule by startup stage.
27 August 2026 · 14 min read LLMsIs Claude nerfed? How to measure model drift yourself
Is Claude nerfed, or did something else change? Documented degradations, the changes that only look like one, and a nightly drift check with error bars.
26 August 2026 · 17 min read