01 The observation

Intelligence understood the request. Commerce never caught up.

A student moves from Rajasthan to Mysuru with ₹5,000 a month and a list of things she needs — groceries, stationery, daily essentials. The intelligence to plan it already exists. The commerce layer to fulfil it does not. Deserta Lab is building that missing layer.

Deserta Lab
The company — building commerce intelligence infrastructure.
LAONI
The first interface to that system, beginning in Mysuru.
Approach
Infrastructure before applications. Systems before features.
02 The broken system

To buy what she needs today, she has to become the integration layer herself.

Every step below is something a person does manually because no system connects them. This is the workflow that exists right now.

Open ChatGPT Generate shopping list Open Google Search products Compare apps Check inventory Build basket Checkout
What works
Artificial intelligence understands the request.
What is missing
Commerce cannot act on that understanding.

This missing layer is what Deserta Lab is built to provide.

03 The company thesis

Deserta Lab is building AI-native commerce intelligence infrastructure.

Most companies build an application and add intelligence later. We are doing the opposite. We are building the intelligence layer for local commerce first — the system that understands products, merchants, inventory, locations and decisions — and treating each product as an interface onto it.

A worldview comes before a product. The worldview is simple: local commerce is enormous and almost entirely uncoordinated, and the durable advantage is not another app — it is the infrastructure underneath every app.

Deserta Lab
The company
Commerce Intelligence Infrastructure
The platform
LAONI
First interface
Future products
Built on the same system
04 The first interface

LAONI is what the infrastructure looks like when a person uses it.

It begins with intent, not a storefront. You describe what you want; the system resolves the right products, the nearest merchants who have them in stock, and the most sensible way to buy. The same infrastructure will power products beyond LAONI in time.

Product-first discovery
Intent leads to products; products lead to merchants. You are never asked to pick a shop before you know what you need.
Merchant-first architecture
Merchants are the source of supply. The system reduces their operational complexity while increasing their visibility.
Local intelligence
Everything is resolved against what is genuinely nearby, in stock, and reachable — not a national catalogue with a pin on a map.
05 Why we begin with manual commerce

Starting simple is the architecture, not a compromise.

LAONI begins as reliable, manual commerce — order, pay, deliver. That is a deliberate decision. Intelligence is earned from behaviour, and behaviour has to be observed before it can be modelled. We refuse to build AI systems on assumptions we have not validated.

Every transaction in this first phase is high-quality behavioural data that later makes the intelligence layer defensible.

01User behaviour is already understood — we build for patterns we can see, not ones we imagine.
02Merchant onboarding is simpler — adoption is faster when the first version is genuinely easy.
03Operational complexity is lower — fewer moving parts, fewer ways to fail early.
04Product–market fit is validated faster — we learn what matters before we automate it.
05Behavioural data compounds — and that data is precisely what later enables the AI systems.
06 The central concept

The commerce graph is the system everything else depends on.

Traditional databases store records. Intelligence needs relationships. The commerce graph is the memory of the ecosystem — customers, merchants, products, inventory, locations and transactions, connected by the relationships between them. Without it, none of the intelligent capabilities are defensible.

viewed sold by searched in stocked in purchased PRODUCT node CUSTOMER MERCHANT LOCATION INVENTORY ORDER
Concept diagram — architecture in development
RecommendationDemand forecastingMerchant intelligenceBasket optimisationAgent systemsRelationship understandingInventory prediction
07 The intelligence model

We are not building conversational commerce. We are building decision infrastructure.

Intelligence here is not a chat box bolted onto shopping. It is a loop that understands a need, reasons about it, plans against real supply, optimises the outcome, executes the transaction, and learns from what happened — so the next decision is better.

01
Understanding
02
Reasoning
03
Planning
04
Optimisation
05
Execution
06
Learning

Learning feeds back into understanding — the loop compounds with every transaction.

08 Architecture

Five layers. Frameworks are an implementation detail.

Architectural thinking communicates engineering maturity; a list of frameworks communicates only what we happened to install. The system is organised as layers, each with a single responsibility. The tools sit underneath, and can change.

Presentation layerHow people and merchants interact with the system.
Flutter · Next.js · TypeScript
Commerce layerOrders, payments, fulfilment — reliable transactions.
NestJS · Node.js · PostgreSQL
Intelligence layerSearch, recommendation, forecasting, optimisation.
Vector search · embeddings · ranking
Knowledge layerThe commerce graph — relationships and memory.
Graph store · event log
Infrastructure layerReliability, scale, observability, security.
Redis · queues · containers
09 Capability roadmap

The roadmap is a progression of capability, not a count of merchants.

Each phase earns the next. Complexity is introduced only when the system is ready to support it — and disciplined execution matters more than ambitious promises.

Phase one
Reliable commerce
Order, pay, deliver. Earn trust and observe real behaviour.
Phase two
Merchant intelligence
Give merchants insight into demand, inventory and performance.
Phase three
Commerce understanding
Recommendation and forecasting from accumulated behaviour.
Phase four
Knowledge graph
The commerce graph as the relationship memory of the system.
Phase five
Agent systems
Specialised agents that plan and act across the graph.
Phase six
Infrastructure platform
The intelligence layer other products are built upon.
10 Research directions

Infrastructure companies do research continuously.

These are the questions we are actively working on. They define Deserta Lab as an engineering and research organisation, not an application company.

Commerce intelligence
Turning raw transactions into structured understanding of demand.
Knowledge graphs
Representing commerce as relationships that systems can reason over.
Demand forecasting
Predicting what a neighbourhood will need, and when.
Regional language search
Understanding intent in the languages people actually use.
Inventory prediction
Helping merchants stock what will sell before it sells out.
Recommendation systems
Ranking against real local supply, not a national catalogue.
Human decision modelling
Understanding how people actually make buying decisions.
Multi-agent commerce
Coordinating specialised agents across a transaction.
Intent understanding
Translating a sentence of need into an executable plan.
11 Engineering principles

How we build is part of what we are building.

01Infrastructure before applications.
02Systems before features.
03Progressive complexity — earn every layer.
04Composable architecture.
05Event-driven thinking.
06Observability by default.
07Security by design.
08AI-ready data models.
12 Differentiation

Durable advantages are structural, not adjectives.

"AI", "hyperlocal", "nearby" and "delivery" are not advantages — anyone can claim them. Our advantages are architectural decisions that are hard to copy.

Not durable

  • AI as a label
  • Hyperlocal as a label
  • "Nearby"
  • Delivery speed alone

Structural advantages

  • Intent-first interaction
  • Merchant-first architecture
  • Product-first discovery
  • Commerce knowledge graph
  • Progressive AI adoption
  • Local intelligence & decision optimisation
13 Founder's letter

The shops on your street already hold almost everything you need. The problem was never supply. It was that no system could see it, connect it, or reason about it.

For a long time the tools meant to help local commerce arrived as marketplaces that flattened merchants into interchangeable listings. They optimised for the platform, not for the neighbourhood. I never believed that was the real opportunity.

Deserta Lab starts from a different conviction: the answer is not another marketplace, it is an intelligence layer — one that begins with what a person actually needs, understands the local supply around them, and coordinates the decision end to end. LAONI is the first interface to that system, and it is intentionally simple today because intelligence has to be earned from real behaviour, not assumed.

We are early, and we are honest about it. There is no finished dashboard to show — only a clear thesis, a disciplined architecture, and a commitment to build with merchants rather than over them. If that resonates, the first neighbourhoods we work with will shape everything that follows.

Kabir Bishnoi
Founder · Deserta Lab
Merchant research · ಮಾರುಕಟ್ಟೆ ಸಂಶೋಧನೆ

Before we build, we are listening to merchants.

Research, not a sales pitch — under three minutes, twelve questions, in English and Kannada. Your answers directly decide what the infrastructure prioritises first.

Tell us about your business — your challenges, your delivery radius, and what would actually help.
ನಿಮ್ಮ ವ್ಯಾಪಾರದ ಬಗ್ಗೆ ತಿಳಿಸಿ — ನಿಮ್ಮ ಸವಾಲುಗಳು, ವಿತರಣಾ ವ್ಯಾಪ್ತಿ ಮತ್ತು ನಿಜವಾಗಿ ಸಹಾಯ ಮಾಡುವ ವಿಷಯಗಳು.
Take the merchant survey · ಸಮೀಕ್ಷೆಯಲ್ಲಿ ಭಾಗವಹಿಸಿ →
Scan to open the survey
ಸ್ಕ್ಯಾನ್ ಮಾಡಿ
Get involved

Be early to the network.

Join the early-access waitlist as a shopper or rider, or register your store to be among the first merchants when LAONI goes live.

Careers

Build the intelligence layer for local commerce.

We are a small, early team building foundational systems from first principles. If infrastructure, AI and real-world commerce excite you, we would like to talk — even if your exact role is not listed.

careers@desertalab.com
Founding Engineer · Backend
NestJS · PostgreSQL · search
Remote / India
Flutter Engineer
Customer · Merchant · Rider
Remote / India
AI / ML Engineer
Search · graph · recommendation
Remote / India
Merchant Growth · Pilot
On the ground
Mysuru
Security & questions

Trust is the foundation of a commerce network.

The platform is designed around security by default — authentication and role-based access, encryption in transit and at rest, fraud detection, rate limiting, audit logging, and recoverable backups.

Is this a company or an app?

Deserta Lab is the company — it builds commerce intelligence infrastructure. LAONI is the first interface to that infrastructure. Future products will be built on the same system.

Why launch with manual commerce instead of AI?

Intelligence is earned from behaviour. Manual commerce lets us validate product–market fit, onboard merchants simply, and collect the high-quality behavioural data that later makes the AI systems defensible. It is a deliberate architectural decision.

What exactly is the commerce graph?

It is the relationship memory of the ecosystem — customers, merchants, products, inventory, locations and transactions, connected by the relationships between them. Recommendation, forecasting, optimisation and agent systems all depend on it.

Is LAONI live yet?

Not yet. We are pre-product and preparing a pilot in Mysuru. Everything here is the thesis and architecture we are building toward — there are no live numbers or dashboards to show, and we will not fabricate them.

How is my data handled?

Survey and registration responses are used only for research, product development, merchant onboarding and pilot communication.

14 · The takeaway

Deserta Lab is building foundational intelligence for local commerce. LAONI is the first interface to that system.

Read the thesis, take the three-minute merchant survey, or join early access. The first neighbourhoods set the direction.