Personal AI assistants
Private assistants built around one person's work and data — a sharp intern that knows your world and stays inside your lines.
Angad Ahluwalia · Operator & AI Builder · Bengaluru
AI software,
built end-to-end.
Production-grade AI software, apps and enterprise products — designed, built, secured and shipped in record time. Not an agency: one operator who builds it, hands you an AI Control Centre to run yourself, and trains your team to own it.
The kettle, on the fire. An honest pot of black.
Built & shipped
- Fita Ai · Personal AI Stylist
- ·
- SkilletCookie · AI Product Builder
- ·
- Nursery Meal Planner · Childcare ops
- ·
- ESPLRecon · Reconciliation
- ·
- AvaGames · AI Game Builder
- ·
- Personalised CEO's AI Assistant
What I build
From one assistant to whole enterprise products.
One builder, every segment. I build AI that fits the brief — a personal assistant, the software a company runs on, or a full consumer AI app — native or on the web, private or in the cloud.
Company AI software
The everyday software a business runs on, powered with AI — automations, internal tools and agents that take work off your desk.
Consumer AI apps
Full, production-grade consumer products — complex apps shipped to real users at scale, in record time. Native or on the web.
Why build this way
Production software, at the speed of a conversation.
Record time to market
What used to take quarters ships in weeks. Built from battle-tested AI Skills, not from scratch — so you reach real users while the idea is still hot.
Flexible by design
Change your mind, change the product. AI-built software bends to new requirements fast, without a rewrite or a long change-request queue.
Your team vibe-codes changes
Tweaks, copy, new screens, small features — your own people describe what they want in plain language and the software changes. No ticket, no wait.
You own and operate it
Code, infrastructure and the AI Control Centre are yours. Run it, extend it, modify it — on your accounts, inside your rules, forever.
How I work
Steeped slowly, served warm.
A clear path from a conversation to a live, production product — built alongside you by one person, then left in your hands with the controls to run it.
-
01
Discover
I learn your world first — the real work, the priorities, the constraints — then scope the build around it.
-
02
Design
Architecture, data model and experience. I decide what to build and how it stays simple, fast and safe.
-
03
Build
Backend, frontend and the AI itself — assembled from proven AI Skills, shipping in record time.
-
04
Secure
Auth, code audit, API hardening and data guardrails — production-grade from the first commit.
-
05
Hand over
Shipped to your cloud, with your AI Control Centre — and your team trained to own, operate and extend it.
The promise
Speed to market without the lock-in. I build fast, ship to production, and then hand you the keys — an AI Control Centre your team owns to operate, build, modify and grow, with the training to do it confidently.
My work
Real products, in the real world.
Two consumer products I shipped end-to-end — and the engine that builds them. All built hands-on with AI. The same AI Skills are waiting for your build.
Built in Claude Code
Built in 52 days · weekends off
Fita Ai — the wardrobe that works for you.
fitaai.comAn AI personal stylist that sees every item you own and styles you for any occasion in seconds. On-device vision, conversational styling with “Ms. Fit”, virtual try-on, and social circles — a full consumer product, shipped to production.
AI Skills used
My own personal library of AI Skills & libraries — so I build these containers and combine them in seconds.
- Next.js · React
- Supabase database (RLS)
- Auth — email OTP
- Google sign-in
- Meta / Facebook sign-in
- WhatsApp Business OTP
- Multi-model AI (Claude · Gemini · GPT)
- Model selection & fallback
- BYOK key encryption
- On-device vision (MediaPipe)
- AI customer service
- Payments & billing
- API security & rate limiting
- IDOR & signed-URL hardening
- GCP Cloud Run · CI/CD
Beta release
SkilletCookie — the engine that builds the rest.
Coming soonThe reason Fita shipped in 52 days and Ava in 32. SkilletCookie is my platform: declare what an AI product needs — inputs, outputs, rules — and it scaffolds the whole backend: database, auth, billing, security, memory, deployment. The 70% every AI product shares, built once. The 30% that's yours, swapped in. The AI does the coding; SkilletCookie makes it correct by construction — model-agnostic, so you swap AI models freely.
And it never sleeps: always-on listeners watch every change while the platform constantly learns — each build harvests new AI Skills, the dependency graph repairs itself, and the library grows on its own.
The AI Skills, themselves
Not AI Skills used on one project — this is the library itself. Declare a product; SkilletCookie wires these together and ships it.
- Composable AI Skill contracts
- Typed I/O — skills chain like pipes
- Multi-model AI gateway · swappable · BYOK
- pgvector memory & open recall lenses
- Always-on listeners
- Self-repairing dependency graph
- Constant learning loop
- Skill-creator — clone by analogy
- Auth · API · security orchestrators
- Governance invariants — correct by construction
- Billing & entitlements
- Cloud deploy — GCP · AWS · Azure
- + many more — new AI Skills auto-shipped hourly
Nursery Meal Planner
Nursery Meal Planner — safe meals, at scale.
Private · client-internal · Staging ↗A meal-operations platform for a children's meal service running across 60 nurseries and thousands of meals a day. From parent meal choices and dietary flags to allergens, it turns the day's orders into two things kitchens actually run on: a daily cooking plan and a daily dispatch plan — resolved down to the individual student, their preferences and each nursery.
Safety-first by design: the AI flags, narrates, cross-verifies and pre-suggests — but never decides the food. Deterministic rules compute every number, so allergens are never guessed.
AI Skills used
My own personal library of AI Skills & libraries — so I build these containers and combine them in seconds.
- AI-powered throughout
- Alert detection & anomaly identification
- Allergy alerts & cross-verification
- AI chatbot · role-scoped
- BYOK for AI · swappable models
- Multi-user · multi-role access
- Postgres + row-level security (multi-tenant)
- AI intent → deterministic SQL → AI narrate
- Roster import + column-match AI
- Daily cooking plan + dispatch plan
- Per-student · per-preference · per-nursery
- Label / sticker generation
- Encrypted at rest · self-host or cloud
India's first AI-native kids' gaming platform. Children describe a game in plain words and the platform builds it — safely. A config-driven game engine, AI hosts with cloned voices, multi-level content moderation, and creator payouts.
AI Skills used
My own personal library of AI Skills & libraries — so I build these containers and combine them in seconds.
- Next.js · React · Phaser engine
- Supabase database
- Auth — email OTP
- Google sign-in
- Parental consent · DPDP
- Multi-model AI (Claude · Gemini · GPT · Groq)
- Model selection by task
- 4-level AI moderation
- ElevenLabs voice
- Replicate image generation
- Creator payouts
- Admin 2FA & roles
- API security & rate limiting
- Cost governance & caching
- GCP Cloud Run · CDN · CI/CD
An AI reconciliation system that closes the vendor-billing leak in large facilities operations. It ingests the messy paper trail — proposals, POs, cost sheets, attendance musters, vendor invoices — reconciles every line against a canonical baseline, and surfaces each discrepancy before payment, with the rupee impact and the evidence to act on it.
Property → run → journey: every number traces to a validation and a ₹ impact. The AI proposes the fix and explains itself; the manager decides.
AI Skills used
My own personal library of AI Skills & libraries — so I build these containers and combine them in seconds.
- Multi-format document parsing (Excel · PDF · OCR)
- Identity reconciliation & fuzzy matching
- Canonical baseline normalisation
- Financial validation & variance detection
- Cost-ceiling enforcement
- AI-proposed rules with reasoning
- Conversational “ask the data”
- Working-capital & recovery surfacing
- Multi-model AI · swappable · BYOK
- Audit trail · approval routing
- Role-based access
The AI Control Centre
Predefined AI Skills, handed to you.
Every build is assembled from my library of production AI Skills I've already made — and handed over as your control centre to own, operate and extend. Deployed privately or in the cloud. Native or web. Custom AI Skills built for your needs.
Backend development
APIs, services, data models and agent logic — the engine room of the product.
Frontend development
Clean, fast interfaces — native apps or on the web — people actually want to use.
Databases
Schema, row-level security, soft deletes — your data modelled to last.
Authentication
Email OTP, Google, Meta, WhatsApp, parental consent — identity done right.
Integrations
Google, Meta, payments, messaging, voice — wired into the tools you already use.
Model selection & BYOK
The right model per task — Claude, Gemini, GPT — with fallback and your own keys.
AI customer service
Automated, on-brand support that knows your product and escalates when it should.
Security & API shield
Guardrails, rate limits, IDOR checks, signed URLs and safe-by-design defaults.
Code audit
Adversarial review of correctness and risk before anything reaches your users.
Cloud & deployment
Private or public cloud — GCP, Cloud Run, CDN — with CI/CD that ships reliably.
Moderation & governance
Multi-level content moderation and cost ceilings that keep AI safe and affordable.
+ custom AI Skills
I build new AI Skills for your business — and hand you the controls to run them.
Training & company-wide enablement
Enable AI
Train your whole company to use & run AI.
Beyond building software, I train and enable your people. Enable AI is a guided, hands-on program that brings your whole company along — shaped to your processes, priorities and people. It ends with working AI in your business and your own team running it as an AI Ops team.
“I build the AI I help you adopt.”
Explore Enable AI training →- 1
See it
A hands-on half-day. Everyone confident with the everyday AI tools, on real work.
- 2
Use it
Go deeper with chosen teams and pick the processes worth powering up.
- 3
Build it
Power up real processes with simple AI tools, running on your data.
- 4
Run it
Your tech team grows into an AI Ops team — the nerve centre that runs AI for everyone.
Swipe →
Put the kettle on.
Tell me what slows you down, or what you want to build. You're talking to the person who'll build it. That conversation is where it begins.
Meet me
I speak your language — I just happen to build the AI too.
I'm Angad — and I'm not a techie. I've spent 15 years exactly where you're standing: running companies and operating across FMCG, real estate, pharma, aerospace, manufacturing, healthcare, advisory — and now AI & technology. So when you tell me what's slowing you down, I get it the first time.
Then I do the rare part — I build the AI myself. Or rather, I've built the AI that builds AI — single-handed: complex software and applications, shipped in record time. Industry instinct and hands-on building is the brew that makes a real product, not a demo.
So I'm not a vendor translating your problem for a team of engineers. I'm the operator who already understands your business — and ships the AI.
I build. I father. I golf.
