# WizCodes — Extended Reference

> This is the expanded companion to llms.txt. It provides deeper context on WizCodes'
> services, process, pricing, and company facts for LLMs and AI answer engines that benefit
> from richer source material.

## Company Overview

- Legal name: WizCodes
- Founded: 16 January 2025, Ahmedabad, Gujarat, India
- Founder & Lead Engineer: Divya Patel
- Model: remote-first, engineer-led studio — no account managers, no agency overhead
- Team: small senior team; clients work directly with the engineers building their product
- Countries served: 11 (United States, United Kingdom, Canada, Spain, Belarus, India,
  Nigeria, Ukraine, Poland, Estonia, Netherlands)
- Projects delivered: 30+
- Website: https://wizcodes.site
- Contact: hello@wizcodes.site

## Services in Detail

### Web Development

URL: https://wizcodes.site/services/web

Full-stack web platforms built for performance and scale, delivered end-to-end and deployed
to the client's own hosting account. The client owns all code, credentials, and accounts
from day one.

What's included:
- SaaS apps and dashboards
- Marketplace platforms
- REST and GraphQL APIs
- Landing pages and portfolios
- E-commerce and payment integration
- Admin panels and CMS
- Third-party API integrations
- Performance optimization and SEO

Tech stack: React, Next.js, TypeScript, Tailwind CSS, FastAPI, Node.js, Python, REST,
GraphQL, PostgreSQL, Supabase, Redis, Firebase, AWS, Vercel, Firebase App Hosting, Docker.

### Mobile Apps

URL: https://wizcodes.site/services/mobile

Cross-platform iOS and Android apps built from one codebase, submitted to the client's own
App Store and Google Play accounts. The client owns the apps, the store listings, and the
developer accounts.

What's included:
- iOS + Android from one codebase
- App Store and Play Store submission (on the client's accounts)
- React Native and Expo builds
- In-app purchases and subscriptions
- Push notifications
- Authentication and onboarding
- Offline-first architecture
- OTA updates via Expo EAS

Tech stack: React Native, Expo, Flutter, FastAPI, Node.js, Firebase, Supabase, RevenueCat,
Stripe, In-App Purchases, Expo EAS, App Store Connect, Google Play Console.

### AI Automation

URL: https://wizcodes.site/services/ai

AI agents, LLM integrations, voice AI, and workflow automation — built end-to-end and
deployed to the client's own infrastructure. The client owns all code, prompts, API keys,
and credentials.

What's included:
- LLM integrations (Claude, GPT-4, Groq)
- Autonomous multi-agent systems
- Real-time voice AI pipelines
- Document OCR and extraction
- Lead generation automation
- RAG systems and knowledge bases
- AI copilots for business workflows
- LangGraph orchestration

Tech stack: LangGraph, LangChain, Claude API, OpenAI, Groq, Replicate, Whisper, Edge TTS,
Document AI, FastAPI, Python, WebSockets, Playwright.

## Stack-Specific Pages

Each covers the questions of someone who has already chosen the technology, rather than
repeating the parent service page.

### Next.js Development

URL: https://wizcodes.site/services/web/nextjs

App Router and React Server Components, with rendering chosen per route rather than applied
to a whole project. Covers WordPress and legacy migrations with redirect mapping, incremental
adoption alongside an existing React app, and Core Web Vitals treated as a budget agreed
before the build. Next.js does not require Vercel: a static export runs on any static host,
and a server-rendered app runs anywhere Node runs. Deployment goes to the client's account.

### React Native Development

URL: https://wizcodes.site/services/mobile/react-native

React Native and Expo for iOS and Android from one codebase. Renders real platform UI
components rather than a web view in a wrapper. Expo is the default — modern Expo supports
custom native code through development builds — with bare React Native where a specific
native dependency requires it. Native modules in Swift or Kotlin when the JavaScript layer
runs out. Over-the-air updates ship JavaScript fixes without waiting on store review. Logic
and TypeScript types share with a web app; UI largely does not.

### Flutter Development

URL: https://wizcodes.site/services/mobile/flutter

Flutter and Dart for iOS and Android. Flutter ships its own rendering engine and draws every
widget itself, so a custom or animation-heavy interface is identical on both platforms.
Includes 2D games via the Flame engine and custom rendering via CustomPaint — most of the
studio's shipped mobile and game work is Dart. Also compiles to desktop; Flutter web is not
recommended for pages that must rank in search, since it renders to a canvas rather than HTML.

### AI Agent Development

URL: https://wizcodes.site/services/ai/ai-agents

Agents that take actions rather than answer questions: given a goal and a set of tools, an
agent decides which to call and in what order until the goal is met or it hits a boundary.
Built on LangGraph so state persists per step and a failed run resumes rather than repeating
actions. Guardrails are defaults, not upgrades — tools scoped to the narrowest permission that
works, outputs validated against a schema, human approval on anything irreversible, and every
run traced. Deployed to the client's own cloud account with their own model keys, so model
usage is billed directly to them rather than marked up.

## Comparisons

Written only where the studio ships production work in both options, which is the stated
condition for publishing one. Each gives both sides a real "pick this one if" case.

### React Native vs Flutter

URL: https://wizcodes.site/compare/react-native-vs-flutter

Nearly every practical difference follows from one architectural decision: React Native maps
to the platform's native components and inherits their behaviour; Flutter draws every pixel
itself and inherits nothing, which removes constraints but makes native feel a deliberate
choice. Pick React Native for teams already in React/TypeScript, for sharing logic with a web
app, and for the larger hiring pool. Pick Flutter for heavily custom or animated interfaces,
for games, and for pixel-identical rendering. Performance is close enough to be a tie for
business applications. For most apps the decision matters less than teams assume.

### Next.js vs WordPress

URL: https://wizcodes.site/compare/nextjs-vs-wordpress

Not the same category of thing: WordPress is a CMS you configure, Next.js is a framework you
build with. WordPress remains the better answer when non-technical people publish and
restructure content regularly, when budget and time are tight, and when mature plugins already
cover the requirements. Next.js wins for products rather than page collections, where
performance is commercially material, and for a small security and maintenance surface — a
static export has no server, database or plugin ecosystem to patch. Headless (WordPress as
editor, Next.js as front end) is a genuine middle path. In a replatform, the risk is unmapped
redirects, not the technology.

## What Drives Software Cost

URL: https://wizcodes.site/pricing/what-drives-the-cost

The site publishes no rate card, deliberately (see Pricing Philosophy below). This page
answers the question underneath the price question — what a quote actually depends on —
ranked by relative impact.

Highest impact: the number of external systems the product must integrate with; how much of
it is genuinely custom versus assembled from solved problems; and whether it needs an admin
or CMS layer, which is effectively a second application and the most commonly underestimated
line in a project.

Medium impact: the number of distinct user roles, since permission logic grows with role
interactions rather than role count; real-time, offline-sync or high-volume data requirements;
how AI features are designed, where calling a model to classify something and building an
agent that acts across systems differ by an order of magnitude; regulatory or contractual data
obligations, which are far cheaper designed in than retrofitted; and how quickly decisions come
back from the client, the one factor entirely in their control.

Low impact: the choice of framework or language, which matters more for hiring and maintenance
than for build cost; and the number of pages or screens, which is the most common basis for a
DIY estimate and among the least reliable, because cost follows behaviour rather than surface
area.

Biggest available savings, in order: defer the admin panel and handle operations manually
until you know what it needs to do; integrate with two systems rather than five at launch; use
managed services for authentication, payments, email and storage; and cut scope rather than
quality, since removing testing, deployment and documentation produces a later bill rather
than a lower price.

## Working Across Borders

URL: https://wizcodes.site/markets

WizCodes is based in Ahmedabad, India, and delivers for clients across the US, UK and Europe.
Three things make the distance workable: the client talks to the engineer writing the code
rather than an account manager; everything deploys to accounts the client already holds, so
they can leave at any point with everything; and scope is agreed in writing after a free
prototype, which removes the misunderstandings that cause most remote-project failure.

Time zones: a few hours ahead of the UK and Europe, giving a working day that covers most of
theirs. For the US, hours overlap with the American business day — the East Coast comfortably,
West Coast mornings well. Same-day replies are the normal case rather than the best case.

### United States

URL: https://wizcodes.site/markets/us

Invoiced in USD against a fixed scope. Standard documentation for a US company paying an
overseas supplier is provided, typically a W-8BEN-E. Client paperwork — MSAs, SOWs, NDAs, IP
assignment, security questionnaires — is signed rather than replaced with our own. IP is the
client's, reinforced structurally: code lives in their GitHub organisation and deploys to
their cloud accounts from day one rather than transferring at handover.

### United Kingdom

URL: https://wizcodes.site/markets/uk

Invoiced in USD as standard, with GBP available on request. As an overseas supplier we do not
charge UK VAT; B2B services are typically handled by the client's accountant under the reverse
charge, though that depends on their circumstances. UK GDPR is treated as an architectural
question rather than a compliance checkbox: because deployment goes into the client's own
cloud accounts, they choose the region, and UK or EU data residency is a configuration choice.
Data processing agreements are signed. Where AI features process personal data, what is sent
to a model provider is a deliberate design decision, including whether a self-hosted model is
the better answer.

## Founder

URL: https://wizcodes.site/about/divya-patel

Divya Patel, Founder and Lead Engineer. Founded WizCodes in 2025 and remains the engineer
behind every project it ships — scoping on the call, building, and handing over personally,
with no account managers and no junior team the work is passed down to. Works across React,
Next.js, TypeScript, Node.js, FastAPI and PostgreSQL on web; Flutter, Dart, React Native and
Expo on mobile; and LangGraph, LangChain, Claude, OpenAI and Whisper for AI systems. Based in
Ahmedabad, India, working with clients across the US, UK and Europe. Every claim on that page
points at something publicly verifiable — shipped projects, open-source repositories, or named
client testimonials — and it deliberately lists no degrees or years-of-experience figures,
since none are published anywhere checkable.

## Process — How a Project Works

Every engagement follows the same five-step process:

1. **Free Prototype**
   A clickable design of the client's core product (key screens + main flow), built around
   their idea, at no cost and no obligation. If they look at it and walk away, they owe
   nothing.

2. **Adjust Together**
   Refined from the client's real reactions — not a sales pitch, but proof they can evaluate
   directly.

3. **Fixed-Scope Plan**
   An accurate quote with no padding. Scope and price agreed in writing up front; anything
   beyond is quoted first. No hourly billing, no retainers, no surprise invoices.

4. **Build & Test**
   Engineered to production standards. The client has direct access to the engineers building
   it — not an account manager. Progress is shared continuously; the client can review and
   give feedback at every stage.

5. **Handover**
   Delivered to the client's own hosting, cloud, and app-store accounts. The client receives
   every repo, credential, API key, and deployment config. They own everything — code,
   accounts, store listings, and infrastructure. WizCodes does not host on its own
   infrastructure and does not retain access after handover unless asked.

## Pricing Philosophy

- Fixed scope, one quote: every project gets a single all-inclusive price before work begins.
  No hourly billing, no retainers, no change-order surcharges for reasonable scope.
- More affordable than a traditional software company: no overhead from account managers,
  sales teams, or office costs. Modern tooling and AI let a small senior team ship faster,
  and that saving is passed directly to the client.
- Free prototype first: the client sees working software before paying anything.
- Affordable, not cheap: quality, security, accessibility, and SEO are part of every build.
  WizCodes competes on efficiency, not on cutting corners.

## Open Source

WizCodes publishes five open-source projects, all under
https://github.com/wizcodes-studio. Every released repository is MIT licensed. Each has
its own page under https://wizcodes.site/open-source/.

ClarivueXAI — explainable-AI library for Python. Feature importance, decision paths and
human-readable explanations for black-box models; works with scikit-learn, XGBoost and
most standard ML pipelines. 1.68K PyPI downloads.
  Install: pip install clarivuexai
  PyPI:    https://pypi.org/project/clarivuexai/
  Repo:    https://github.com/wizcodes-studio/clarivuexai
  Page:    https://wizcodes.site/open-source/clarivuexai

Bulkhead — a Model Context Protocol (MCP) server that catches security problems in code
written by AI coding agents. You attach it to the agent once; it then reviews what the
agent produces and flags security issues before they reach the codebase. No workflow
change and nothing to run alongside it. Not publicly released yet.
  Page:    https://wizcodes.site/open-source/bulkhead

Cyber Bug Bounty Agent — an autonomous security-research pipeline: two LangGraph state
graphs plus a permission watcher, coordinated by a supervisor. The intake graph runs
ResearcherAgent (OSINT and company profiling) then OutreachAgent (permission email) and
halts. A permission watcher polls IMAP and runs ClassifierAgent (Claude Haiku) to decide
whether access was granted. On approval the scan graph runs ReconAgent, ScannerAgent,
AggregatorAgent (Claude Opus, deduplication and CVSS enrichment), ReporterAgent (PDF via
ReportLab) and DeliveryAgent. Supporting modules handle token-aware API calls, RAM-aware
staged execution, network/VPN resilience, SQLite persistence in WAL mode, push
notifications and SMTP delivery.
  Repo:    https://github.com/wizcodes-studio/cyber-bug-bounty-agent
  Page:    https://wizcodes.site/open-source/cyber-bug-bounty-agent
  Case study (the narrative version): https://wizcodes.site/work/cyber-agent

Cartoon Diffusion — image-to-cartoon model and Python library, built on Stable Diffusion
techniques and fine-tuned on the Google Cartoon Set avatar dataset from Kaggle. Published
as both a trained model and a library that wraps it, so converting an image is a two-line
call. Model id: wizcodes12/image_to_cartoonify.
  Install: pip install cartoon-diffusion
  PyPI:    https://pypi.org/project/cartoon-diffusion/
  Model:   https://huggingface.co/wizcodes12/image_to_cartoonify
  Repo:    https://github.com/wizcodes-studio/cartoon-diffusion-library
  Page:    https://wizcodes.site/open-source/cartoon-diffusion

Chrome Extensions — full source for the browser extensions WizCodes builds, so anyone can
read what an extension does before installing it, fork it, or build it themselves.
TypeScript, Manifest V3. The first extension is Tab Copilot, currently in development.
  Repo:    https://github.com/wizcodes-studio/chrome-extensions
  Page:    https://wizcodes.site/open-source/chrome-extensions

## Contact

- Email: hello@wizcodes.site
- Website: https://wizcodes.site/contact
- LinkedIn (company): https://www.linkedin.com/company/wizcodes/
- LinkedIn (founder): https://www.linkedin.com/in/divya-patel-957236283
- X / Twitter: https://x.com/wizcodes_co (@wizcodes_co)
- GitHub: https://github.com/wizcodes-studio
- Instagram: https://instagram.com/wiz_codes
- WhatsApp: https://wa.me/917862036533
- YouTube: https://www.youtube.com/@WizCodes
- Facebook: https://www.facebook.com/profile.php?id=61591785185792
- dev.to: https://dev.to/wizcodes
- Hashnode: https://wizcodes.hashnode.dev/

Independent review profiles (third-party verified):
- Clutch: https://clutch.co/profile/wizcodes
- GoodFirms: https://www.goodfirms.co/company/wizcodes
- The Manifest: https://themanifest.com/company/wizcodes

<!-- BEGIN:case-studies (generated by scripts/gen-llms.mjs — do not edit) -->

## Case studies

Each entry below is a self-contained factual summary of one real project.

### Destiny AI Journal
URL: https://wizcodes.site/work/destiny-ai-journal
Platforms: iOS, Android.
Built with: React Native, Expo, FastAPI, Supabase, Groq, RevenueCat.

Business problem: People abandon journaling apps when the app interrupts them. Destiny AI Journal had to add transcription, summarisation and spoken narration to a journal entry without making the writer wait for any of it.
Technical challenges: Destiny AI Journal chains three separate AI services per entry — speech-to-text, summarisation, then text-to-speech. Run in sequence and in the foreground, that chain is long enough for a user to notice, and it has to work on a phone with an unreliable connection.
Engineering solution: Destiny AI Journal saves the entry to Supabase immediately and returns, then processes transcription, summary and narration on background workers, streaming each result back to the app as it completes.
Architecture: React Native (Expo) → FastAPI → Groq | Replicate | Edge TTS → Supabase → Background workers.
Core features: Voice journaling with automatic transcription; AI-written summary per entry; Spoken narration played back in-app; Mood tracking with a historical heatmap; Social feed and user matching; Subscription billing via RevenueCat; Push notifications.
Technical highlights: Entries persist before any AI work begins, so nothing blocks the write; Results stream back progressively rather than arriving in one response; One Expo codebase targets both iOS and Android; Supabase realtime carries the social feed without a separate WebSocket layer.
Business value: Destiny AI Journal gives a writer a transcribed, summarised and narrated entry while keeping the app as fast to write in as a plain notes app. The AI is additive rather than something the user waits on.

### CuePilot
URL: https://wizcodes.site/work/cuepilot
Platforms: Web.
Built with: Python, FastAPI, Whisper, WebSockets, React.

Business problem: Support agents lose time mid-call deciding how to answer a difficult customer. CuePilot had to put a usable suggested response in front of the agent fast enough to be read while the customer is still speaking.
Technical challenges: CuePilot chains audio capture, speech-to-text and an LLM inside a budget of roughly 200 milliseconds end to end, while holding open WebSocket connections for every agent on a support floor at once.
Engineering solution: CuePilot streams browser audio over WebSockets to a FastAPI service, runs Whisper incrementally on GPU workers so transcription keeps pace with speech, and pushes each LLM suggestion to the agent UI as soon as it is ready.
Architecture: Browser → WebSockets → FastAPI → Whisper → LLM → React Dashboard.
Core features: Live transcription of an in-progress call; AI-suggested responses, updated as the call moves; One-click copy or adapt from the suggestion panel; Session history and per-call analytics; Multi-agent team mode with parallel sessions.
Technical highlights: Connections are multiplexed per session rather than per agent; Whisper runs on GPU workers with pooled connections, so no request pays model load time; Transcription is incremental, so the LLM starts from partial text instead of waiting for silence; The suggestion panel updates in place rather than re-rendering the transcript.
Business value: CuePilot lets a support agent concentrate on the customer instead of on composing a reply. The suggestion is already on screen by the time the agent would have started thinking about one.

### Cyber Agent
URL: https://wizcodes.site/work/cyber-agent
Platforms: Web.
Built with: Python, FastAPI, LangGraph, Claude API, Resend.

Business problem: A thorough website security assessment means running roughly ten separate tools, reading ten separate outputs, and writing the findings up by hand. WizCodes was doing that manually on every audit, and most of the effort was collation rather than analysis.
Technical challenges: Cyber Agent has to run many scanners concurrently without them interfering, normalise ten incompatible output formats into one finding shape, decide which findings actually matter, and produce a report a non-technical reader can act on.
Engineering solution: Cyber Agent runs its scanners in parallel behind an orchestrator, normalises every result into a single finding schema, has an LLM rank findings by business impact rather than by raw tool severity, and generates the written report from the ranked set.
Architecture: Scan Orchestrator → Security Tools → AI Analysis → Report Generator → Dashboard.
Core features: Parallel scanning across ten security tools; Normalised findings from incompatible tool outputs; AI ranking by business impact rather than raw severity; Generated PDF report, written for a non-technical reader; Scheduled re-scans and on-demand runs; Dashboard for reviewing findings across scans.
Technical highlights: Scanners run concurrently, so total time is the slowest tool rather than the sum; One normalised finding schema behind ten incompatible tool outputs; LangGraph holds the orchestration state, so a failed scanner degrades the run instead of ending it; The report is generated from ranked findings, not from raw scanner output.
Business value: Cyber Agent turns a day of tool-wrangling and write-up into a single scan and a finished report. The engineering time goes into interpreting findings rather than into collecting them.

### Nullzec
URL: https://wizcodes.site/work/nullzec
Platforms: Web.
Built with: React, Next.js, TypeScript.

Business problem: Nullzec needed a web interface for an enterprise browser-isolation product. The engineering effort was going into the isolation technology itself, and the client-facing application through which the product is configured and monitored needed a dedicated frontend team.
Technical challenges: The Nullzec console presents dense security information — policies, live session activity, audit logs — that has to stay legible and fast to scan. Audit views in particular grow without bound, and a naive list renders tens of thousands of rows into the DOM.
Engineering solution: WizCodes built the Nullzec frontend as a component-driven React and Next.js application with one design system, virtualising the heavy list views and lazy-loading the rest so the interface stays responsive regardless of data volume.
Architecture: Next.js → React → TypeScript → API Integration.
Core features: Client-facing web application UI; Policy management and admin dashboard; Session activity and audit log views; Account, settings and onboarding flows; Reusable design-system component library; Responsive layouts and accessibility passes.
Technical highlights: Virtualised list rendering, so audit views stay smooth at volume; Lazy-loaded routes keep the initial console load small; One design system rather than per-screen styling; TypeScript throughout, so the client’s team inherits a checkable codebase.
Business value: The Nullzec engagement gave the client a production-grade interface without diverting their own engineers away from the isolation engine. The frontend was built to be handed over and extended by their team.

### SolarSathi
URL: https://wizcodes.site/work/solarsathi
Platforms: Web.
Built with: React, Node.js, PostgreSQL, AWS.

Business problem: Buying commercial solar in India means finding vendors one at a time, describing the same requirement repeatedly, and comparing quotes that arrive in different formats over days. SolarSathi was built so a buyer describes a requirement once and receives comparable quotes in one place.
Technical challenges: SolarSathi coordinates an asynchronous conversation between one buyer and many vendors, where each vendor responds on their own schedule. The platform has to track every request independently, notify both sides at the right moments, and stay simple enough to ship as an MVP.
Engineering solution: SolarSathi models each quote request as a small state machine — requested, reviewed, quoted, accepted — with every transition firing an email notification. Vendor listings and search run on PostgreSQL, and the whole platform is deployed on AWS.
Architecture: React → Node.js → PostgreSQL → AWS Services.
Core features: Multi-vendor listings with search and filtering; Quote requests that fan out from one buyer to many vendors; Four-state workflow tracking per request; Email notifications on every state change; Vendor profile and verification pages; Admin panel for listing management.
Technical highlights: A four-state machine keeps every request auditable rather than free-form; Notifications are driven by state transitions, not by scheduled polling; PostgreSQL search is tuned for vendor discovery by capacity and region; A mainstream stack chosen so the client could hire for it later.
Business value: SolarSathi turns solar procurement from a sequence of separate vendor conversations into one request with comparable answers. Vendors reach buyers they would not otherwise meet, and buyers stop repeating themselves.

### AI Lead Generation Agent
URL: https://wizcodes.site/work/ai-lead-agent
Platforms: Web.
Built with: Python, LangGraph, Playwright, FastAPI, SMTP.

Business problem: Finding prospects, researching each one, and writing a proposal that is actually about them is slow, repetitive work that scales only by hiring. The AI Lead Generation Agent was built to run that entire sequence without a person in it.
Technical challenges: The AI Lead Generation Agent has to read sites that only exist after JavaScript runs, survive anti-bot measures across wildly different domains, decide whether a prospect is worth contacting, and write outreach specific enough to be worth reading — all without per-domain maintenance.
Engineering solution: The AI Lead Generation Agent runs as a LangGraph state machine where a classifier node identifies the site type first and routes it to the right extraction strategy, then an LLM writes a proposal from the extracted context and SMTP delivers it, with every step logged and deduplicated.
Architecture: LangGraph Agent → Playwright → LLM → SMTP → Monitoring Dashboard.
Core features: Autonomous scraping of JavaScript-rendered sites; Site classification that routes extraction strategy; Lead enrichment for contact, role and company context; LLM-written proposals grounded in the prospect’s own site; Automated email delivery; Deduplication against previously contacted prospects; Scheduled and on-demand runs with a monitoring dashboard.
Technical highlights: A classifier node runs first, so per-domain engineering is close to zero; LangGraph holds run state, so a failed node retries instead of restarting the run; Deduplication is persistent, so no prospect is contacted twice across runs; Playwright with randomised fingerprinting handles JavaScript-heavy pages.
Business value: The AI Lead Generation Agent turns prospecting from a headcount problem into a scheduling one. The work that scaled by hiring now scales by running more often.

### Tocablox World
URL: https://wizcodes.site/work/tocablox-world
Platforms: iOS, Android.
Built with: Flutter, Flame Engine, Dart, Hive.

Business problem: Most virtual pet games retain players with notification pressure, streak anxiety, and monetisation aimed at children. Tocablox World was built to hold attention through the game itself, with no account, no connection, and nothing engineered to create guilt.
Technical challenges: Tocablox World has to keep a care loop rewarding with no online hooks at all, persist every bit of progress on the device, and animate smoothly on mid-range Android hardware where a dropped frame is immediately visible in a slow, calm game.
Engineering solution: Tocablox World runs its pet state locally on a lightweight Flame game loop with Hive for on-device persistence, and paces content unlocks so there is always a visible next goal without any external prompt.
Architecture: Flutter → Flame Engine → Local Hive Database.
Core features: Real-time virtual pet with needs, moods and reactions; Home decoration with unlockable furniture and items; Progression loop with in-game currency; Fully offline play with on-device save persistence; Animation tuned for mid-range Android and iOS; Child-friendly interface with no dark patterns.
Technical highlights: No network layer at all, so there is no loading state anywhere in the game; Hive persistence means the game opens straight into its last state; One Flutter codebase renders identically on iOS and Android; The game loop is deliberately small, which is what keeps frames stable on modest hardware.
Business value: Tocablox World gives families a game that works on a plane, in a car, and on an old phone, and that no parent has to supervise for manipulative mechanics.

### MindMaze Junior
URL: https://wizcodes.site/work/mindmaze-junior
Platforms: Android, iOS, Web, Windows, Linux, macOS.
Built with: Flutter, CustomPaint, Dart, Local Storage.

Business problem: Parents wanted an educational game that helps young children learn through interactive gameplay without requiring reading skills or internet connectivity.
Technical challenges: MindMaze Junior needed to deliver child-friendly interactions with smooth performance across six platforms while creating reusable game logic and ensuring intuitive gameplay.
Engineering solution: MindMaze Junior draws its entire interface with Flutter’s CustomPaint API and no third-party packages, so behaviour is identical across Android, iOS, Web, Windows, Linux and macOS.
Architecture: Flutter → CustomPaint Rendering Engine → Local Storage.
Core features: Four educational game modes; Child-friendly touch interactions; Offline gameplay; Cross-platform from one codebase; Custom rendering engine; Local progress storage; No reading required; Ages 3 and up.
Technical highlights: MindMaze Junior uses CustomPaint to render all game graphics directly, avoiding third-party packages that lag on Linux and Windows or break after Flutter updates.; The game runs on six platforms from one codebase with zero per-platform compatibility patches.; All game state lives in local storage, so MindMaze Junior works fully offline with no server dependency.; Child-friendly interactions were designed to work without text labels or reading ability..
Business value: MindMaze Junior delivers engaging educational gameplay while minimizing maintenance and maximizing platform reach.

<!-- END:case-studies -->

<!-- BEGIN:blog (generated by scripts/gen-llms.mjs — do not edit) -->

## Blog

Each entry is one published article: what it covers and the questions it answers.

### Game Development Studio Cost: What You Pay (2026)
URL: https://wizcodes.site/blog/game-development-studio-cost-2026-real-numbers
Published: 2026-08-12.
Topics: Game Development, Mobile Apps, Cost Breakdown.

Game development studio cost ranges from $15k to $60k for the same scope. Here's what drives the price and where budgets get wasted.

Questions answered:
- What actually drives the cost of building a mobile game
- Where most game budgets get wasted
- A realistic cost breakdown by game type
- How to get more game for less without cutting corners
- When to invest more in a game studio - and when not to

### FastAPI vs Django for Startups (2026)
URL: https://wizcodes.site/blog/fastapi-vs-django-for-startup-which-framework-2026
Published: 2026-08-05.
Topics: FastAPI, Django, Startup Tech Stack, Python Frameworks.

FastAPI vs Django for startup products: one handles AI and real-time workloads, the other ships admin panels and mature tooling. Here's how to choose.

Questions answered:
- What each framework actually offers
- Django wins on stability and the admin panel
- FastAPI wins on speed and modern async patterns
- Decision criteria most framework guides skip
- How to pick the right one for your MVP

### Enterprise Browser Isolation: What It Costs
URL: https://wizcodes.site/blog/enterprise-browser-isolation-vendors-pricing-guide
Published: 2026-08-03.
Topics: Enterprise Security, Browser Isolation, Security Budget.

Enterprise browser isolation vendors quote $8-95/user/month, but real costs run 40% higher. Here's what drives pricing and where to push back.

Questions answered:
- What actually drives enterprise browser isolation costs
- Where most security budgets get wasted
- A realistic pricing breakdown for browser isolation
- How to evaluate vendor contracts without overpaying
- When expensive is worth it and when it's just markup

### Choosing a Database for a SaaS Product
URL: https://wizcodes.site/blog/choosing-a-database-for-saas-product-postgres-vs-mongodb
Published: 2026-08-01.
Topics: SaaS Development, Database Architecture, Technical Decision.

Most founders pick a database based on familiarity. Here's how to choose based on query patterns, scale, and cost - before migration becomes impossible.

Questions answered:
- Why database choice is harder than it looks for SaaS founders
- The three variables that decide your database architecture
- A practical decision framework for SaaS founders
- The migration scenarios that cost founders six figures
- How to validate your database choice before you build

### Build vs Buy Customer Support Software
URL: https://wizcodes.site/blog/build-vs-buy-customer-support-software-cost-breakdown
Published: 2026-07-22.
Topics: Customer Support, SaaS Strategy, Cost Analysis.

Most teams choose wrong. See the actual cost breakdown for build vs buy customer support software decisions — including hidden lock-in costs.

Questions answered:
- Why this decision is harder than it looks
- The four variables that change the answer
- A practical decision framework for support software
- Common mistakes that lead to the wrong choice
- How to validate your decision before committing

### What custom software actually costs in 2026
URL: https://wizcodes.site/blog/what-custom-software-actually-costs-in-2026
Published: 2026-07-14.
Topics: Pricing, Web, AI.

A plain-English breakdown of what drives the price of custom software in 2026, where the money really goes, and why AI lowers cost, not quality.

Questions answered:
- What are you actually paying for?
- What makes one project cost more than another?
- Why did the arithmetic change?
- Why do quotes vary so wildly?
- What questions should you ask?
- How do you keep the cost down honestly?
- The short version

### Why we build a free prototype first
URL: https://wizcodes.site/blog/why-we-build-a-free-prototype-first
Published: 2026-07-10.
Topics: Process, Startups.

A working prototype before a contract sounds risky for the studio. Here is why we do it anyway, and how it de-risks a project for both sides.

Questions answered:
- What is wrong with the normal way?
- How does a prototype fix it?
- What do you actually get?
- Why give it away free?
- What happens after the prototype?
- Is there a catch?
- The short version

### How we pick React Native or Flutter
URL: https://wizcodes.site/blog/react-native-vs-flutter-2026
Published: 2026-07-07.
Topics: Mobile, Engineering.

Not a spec-sheet war: how a studio actually decides, project by project, for the apps we ship to the App Store and Play Store.

Questions answered:
- What is actually different between them?
- When do we choose React Native?
- When do we choose Flutter?
- How do you decide in ten minutes?
- What about performance?
- What does the choice cost you later?
- What we would tell a founder
- The short version

### How to add an AI agent to your product
URL: https://wizcodes.site/blog/add-ai-agent-to-your-product
Published: 2026-07-03.
Topics: AI, Engineering.

AI agents are easy to demo and hard to ship. Scoping, guardrails, cost control, and the failure modes to plan for before launch.

Questions answered:
- Do you actually need an agent?
- How narrow should the scope be?
- What guardrails does it need?
- How do you keep the cost and the wait under control?
- How do you know it actually works?
- How should you launch it?
- The short version

### Building a real-time voice AI copilot
URL: https://wizcodes.site/blog/building-real-time-voice-ai-copilot
Published: 2026-06-30.
Topics: AI, Case Study.

CuePilot suggests responses mid-call in under 200 ms. The architecture behind that number, the hard parts, and what we would do differently.

Questions answered:
- What is the chain you are racing against?
- Why does streaming change everything?
- Where did the milliseconds actually come from?
- What was the hardest part?
- What would we do differently?
- What transfers to other products?
- The short version

### The automation audit: 7 workflows to start with
URL: https://wizcodes.site/blog/startup-automation-audit-7-workflows
Published: 2026-06-26.
Topics: Automation, Startups.

Before you buy another tool, run this audit: seven workflows that quietly eat a small team alive, and what it takes to automate each one.

Questions answered:
- Which seven workflows should you check?
- How do you decide which one to fix first?
- What does that look like on a page?
- What should stay human?
- What goes wrong most often?
- How do you know it worked?
- The short version

### Idea to MVP SaaS: our end-to-end playbook
URL: https://wizcodes.site/blog/idea-to-mvp-saas-in-six-weeks
Published: 2026-06-23.
Topics: SaaS, Startups.

A week-by-week look at how a SaaS idea becomes a deployed MVP: what we build, what we deliberately skip, and the traps that add months.

Questions answered:
- What is an MVP actually for?
- What happens in week one?
- What gets built in week two?
- How is the core loop built?
- Why spend a whole week on edge cases?
- What does handover look like?
- When is six weeks the wrong number?
- The short version

### Own your code: what agency lock-in really costs you
URL: https://wizcodes.site/blog/own-your-code-agency-lock-in
Published: 2026-06-19.
Topics: Ownership, Process.

Their servers, their accounts, "we will handle that for you" — the small conveniences that trap you, and how to keep ownership from day one.

Questions answered:
- How does lock-in actually happen?
- Why does it matter if the work is good?
- What does real ownership look like?
- What if it is already too late?
- What should you ask before you sign?
- Why we deliver this way by default
- The short version

### The 5 UX decisions that make or break an MVP
URL: https://wizcodes.site/blog/five-ux-decisions-that-make-or-break-an-mvp
Published: 2026-06-16.
Topics: UI/UX, Product.

Most MVPs fail on five UX decisions made in the first week. What they are, why they matter more than polish, and how to get them right early.

Questions answered:
- Is the one thing obvious in five seconds?
- What does the screen say when it is empty?
- What happens when something goes wrong?
- How many steps to the payoff?
- Does it use your words or theirs?
- Why do these beat visual polish?
- The short version

### Digital transformation without an enterprise budget
URL: https://wizcodes.site/blog/digital-transformation-small-business
Published: 2026-06-12.
Topics: Digital Transformation, Business.

Digital transformation is not just for seven-figure IT budgets. A step-by-step approach for small businesses that want results, not theatre.

Questions answered:
- What does it actually mean at your size?
- Why not start by picking a tool?
- How do you decide what to do first?
- What tends to pay off first?
- Why is this affordable now when it was not before?
- What should you insist on?
- The short version

### LLM integration done right: RAG and guardrails
URL: https://wizcodes.site/blog/llm-integration-rag-prompts-guardrails
Published: 2026-06-09.
Topics: AI, Engineering.

The gap between a working prompt and a production LLM feature is bigger than it looks. Retrieval, prompt design, evaluation and guardrails.

Questions answered:
- Why does a working prompt fail in production?
- Why is retrieval the first thing to fix?
- What makes a production prompt good?
- How do you know a change made it better?
- What guardrails does a live feature need?
- What about cost and speed?
- The thread running through all of it
- The short version

### Offshore vs in-house vs studio
URL: https://wizcodes.site/blog/offshore-vs-in-house-vs-studio
Published: 2026-06-05.
Topics: Startups, Business.

In-house, offshore or a studio: each solves a different problem. An honest comparison of cost, speed, risk and ownership for founders.

Questions answered:
- What are you actually choosing between?
- When should you hire in-house?
- When does offshore make sense?
- Where does a studio fit?
- How do you actually decide?
- What does this cost in practice?
- The short version

<!-- END:blog -->

<!-- BEGIN:facts (generated by scripts/gen-llms.mjs — do not edit) -->

- Countries with delivered client work: 11 (Belarus, Canada, Estonia, India, Netherlands, Nigeria, Poland, Spain, Ukraine, United Kingdom, United States)
- Client projects delivered: 26
- Published client testimonials: 13
- Typical delivery for a defined MVP: 3-4 weeks, up to 6 with heavier integrations
- Every project starts with a short call and a free clickable prototype, both at no cost
- All code, credentials and store listings are handed to the client's own accounts

<!-- END:facts -->