STATODE

AI-powered food intelligence for the global restaurant economy

Restaurant chains and food companies finally get the competitive intelligence they've never had: per-dish margins, competitor benchmarks, menu gaps, and pricing strategy. All automated, all fresh daily. Built on proprietary AI that understands food at the ingredient level. Starting with India's 500K+ menu items across 100K+ restaurants.
March 2026  •  Pre-seed  •  statode.com
The Problem

Restaurant owners price by gut. Food companies sell by guesswork. This is true everywhere.

For restaurant chains

  • Zero visibility into competitor pricing, menus, or discounting
    A 15-outlet chai chain doesn't know what the chai shop next door charges
  • No reliable way to estimate their own food cost per dish
    Most chains under 50 outlets don't have a food cost system
  • Menu decisions based on intuition, not data
    Which items to add, price, or remove is a guessing game
  • No semantic food search or discovery
    Apps still use keyword matching; "chicken tikka" doesn't find "murgh tikka"
  • Consulting firms charge 5-10L for a competitive audit
    Takes weeks. Stale by the time it arrives.

For food suppliers & FMCG

  • No structured data on what restaurants actually serve
    Field sales teams rely on physical visits and anecdotal intel
  • Can't identify which chains use their ingredients
    A cheese company can't find all pizza chains in Bangalore
  • Pricing intelligence is nonexistent
    No way to know how ingredient cost trends affect menu pricing
  • Market sizing for food categories is manual and slow
    Consultants charge 20-50L for category reports
What We Do

We build proprietary AI systems that understand food. Then we sell the intelligence.

Scrape

500K+ items from Swiggy daily. Menus, prices, ratings, discounts.

→

Understand

Food embedding model classifies, normalizes, and semantically maps every item.

→

Cost

AI matches to 63K+ recipes. Ingredient-level food cost estimation.

→

Benchmark

Competitor graphs, area benchmarks, pricing gaps, menu engineering.

→

Deliver

PDF reports, datasets, dashboards, API, embedding-as-a-service.

Product 1: Competitive Reports

  • Per-dish margins, competitor pricing, menu gaps
  • Fully automated, payment to inbox in 15 min
  • INR 1,999 - 11,999 per report
B2C LIVE

Product 2: Data Intelligence

  • Structured menu, pricing, food cost data
  • For FMCG, suppliers, consulting firms
  • Datasets, dashboards, or API access
B2B ACTIVE SALES

Product 3: Food Embedding Model

  • Semantic food search and recommendation
  • Global dish model, works across cuisines
  • Already live: powering abCoffee's app search
AI PRODUCT IN PRODUCTION
AI + Data Engine

What our AI collects and computes, every day

500K+
Menu items tracked daily
100K+
Restaurants across 9 cities
63,516
AI-generated recipes with ingredient-level costing

Computed nightly by 10 automated AI + analytics jobs

Pricing Position

Per-item pricing vs area median, tier peers, and direct competitors

AI Food Cost

Estimated food cost, MRP margin, and post-discount margin for every matched item

Competitor Graphs

Chain and outlet-level competitor pairs scored by cuisine overlap

Menu Engineering

Star / Plowhorse / Puzzle / Dog quadrants by popularity and margin

Category Mapping

16 subcategories with penetration rates, powered by food embeddings

Discount Strategy

Discount penetration, avg discount %, effective pricing after offers

Area Benchmarks

Median price, rating, food cost by area with 5-level fallback

Expansion Scoring

Area-level opportunity scores by demand, density, and rating gaps

Platform-agnostic architecture. Same AI pipeline works on any food delivery marketplace globally.
Proprietary AI Systems

AI estimates food cost for any dish, without the restaurant's cooperation

No restaurant shares their recipes or costs. Our AI reverse-engineers them by matching every menu item to a proprietary recipe database, then pricing each ingredient at city-specific wholesale rates.

THE AI PIPELINE

  • Normalize: 10 specialized NLP modules strip noise, fix transliterations, detect proteins, parse combos
  • Embed: Proprietary food embedding model (384-dim) maps dishes to semantic space
  • Match: FAISS nearest-neighbor + multi-signal re-ranking (embedding + string + weighted food-token similarity)
  • Guard: Hard veg/non-veg gate, protein conflict rejection, category mismatch blocking
  • Verify: LLM re-ranking on uncertain matches. Binary classification + top-5 re-rank
  • Cost: Ingredient quantities x city-specific wholesale prices x tier multiplier

WHY THIS COMPOUNDS

  • Every unmatched item triggers AI recipe generation (40K → 63K recipes organically)
  • Each new recipe permanently improves matching at zero marginal cost
  • The system builds itself: more data → better AI → more accurate → more products

Match accuracy: 98.5%

  • Up from 67% baseline through 3-phase AI improvement
  • 242K verified matches, avg cosine score 0.962
  • Chain coverage (tested): 95%+ of menu items

Proprietary AI systems that build themselves

  • Recipe generation AI: creates ingredient-level recipes for any dish
  • Grey-zone verifier: LLM classifies uncertain matches
  • Re-ranker: selects best match from top-5 candidates
  • Alias generator: expands recipe catalog with regional name variants
  • Total R&D cost for all AI systems: $72

Wholesale pricing (supply side)

  • Scrape B2B marketplace across 11 cities for real wholesale prices
  • City-specific cost multipliers + price tier adjustments
  • The denominator of every food cost calculation
Product 1

Competitive Intelligence Reports

Restaurant owner searches on statode.com, pays, receives a PDF in their inbox within 15 minutes. Fully automated.

WHAT'S IN THE REPORT

  • Per-dish margin analysis: food cost, MRP margin, post-discount margin
  • Competitor pricing benchmark: vs direct competitors and area median
  • Menu gap analysis: high-demand items competitors sell that the chain doesn't
  • Menu engineering quadrants: Stars, Puzzles, Plowhorses, Dogs
  • Discount strategy: discounting vs competitors, with margin impact
  • Expansion opportunities: areas ranked by demand and competitor density

Unit economics

Revenue per report (avg)~INR 5,000
AI generation costINR 50
Email + storageINR 0.60
Gross margin~99%

Pricing (one-time, no subscription)

1 outletINR 1,999
2-5 outletsINR 3,499
6-15 outletsINR 5,999
16-50 outletsINR 8,999
50+ outletsINR 11,999

Delivery pipeline

  • Payment → data extraction → AI narratives → template → PDF → email
  • Validated against source data for zero hallucination
  • ~15 min, payment to inbox
Product 2

Data Intelligence for Food Companies

The same AI engine that powers reports contains structured intelligence that food suppliers, FMCG companies, and consulting firms can't get anywhere else.

Menu & Pricing Data

Every item, price, category, veg/non-veg across 100K+ restaurants. Daily.

AI Food Cost Estimates

Ingredient-level cost breakdown for 242K+ items. City and tier-adjusted.

Geographic Intelligence

Restaurant density, cuisine penetration, competitive heatmaps. 9 cities.

Category Analysis

16 subcategories. Menu share, price distribution, discount patterns.

B2B Supply Pricing

Wholesale ingredient prices across 11 cities. 24 fields per product.

Trend Data

Daily snapshots with 7/30/90-day deltas on pricing, ratings, discounts.

WHO BUYS THIS

Ingredient suppliers
Dairy, frozen foods, sauces, spices. "Which chains in Bangalore serve cheese items?"
FMCG brands
Amul, McCain, Veeba, ITC. Track ingredient penetration.
Consulting firms
Category reports, market sizing. Replaces weeks of primary research.
Restaurant tech
POS, cloud kitchen platforms. Enrich products with market data.
100 target companies identified across 12 categories. Active outbound pipeline.
Core AI: Food Embedding Model

First-of-its-kind: a global food embedding model that understands every dish

We built a proprietary embedding model trained on food semantics across cuisines and languages. It doesn't just match strings. It understands that "murgh makhani" and "butter chicken" are the same dish, that "gosht biryani" means mutton, and that a "loaded nachos" is closer to "cheese fries" than to "tortilla chips."

WHAT IT DOES

  • Semantic food matching at 98.5% accuracy across 242K+ items
  • Multilingual: handles Hindi, Urdu, regional transliterations natively
  • Cross-cuisine understanding: maps dishes across Indian, Chinese, Italian, Mexican, and more
  • Global dish taxonomy: 16 subcategories, 65 cuisine clusters, protein/veg classification
  • Runs on CPU: 384-dim vectors, zero API cost at inference

WHY THIS IS HARD

  • Food names are chaotic: abbreviations, transliterations, regional variants, combo names
  • Generic embedding models fail on domain-specific food semantics
  • We built 10 specialized NLP modules (protein detection, spelling normalization, combo parsing, etc.) that feed the model

Accuracy

98.5%
Match accuracy
(up from 67% baseline)
0.962
Avg cosine score
(242K verified matches)

Monetizable beyond Statode

  • Food platform search: semantic "what do you feel like?" queries
  • Recommendation engines: "if you liked X, try Y"
  • Menu deduplication: identify same dish across restaurants
  • Ingredient intelligence: infer ingredients from dish names

Already in production

abCoffee, a specialty coffee chain in India, uses our food embedding model to power semantic search on their consumer app. First external customer, validating the model's commercial value beyond our own platform.

Market

Global food delivery is $300B+. No one owns the intelligence layer.

$300B+
Global food delivery market (2025), growing 10%+ YoY
$7B+
India alone, growing 20%+ YoY. Our starting market.
800K+
Restaurants on Swiggy and Zomato combined

The problem is identical everywhere. Uber Eats, DoorDash, Deliveroo, Grab, iFood all have the same menu data structure. Our AI pipeline is platform-agnostic. India is where we start; the approach scales globally.

Why no one has built this

  • Platforms don't sell competitive intelligence.
    Swiggy/Zomato show a restaurant its own performance, never competitors or food costs.
  • Food cost estimation requires domain-specific AI.
    You need AI-generated recipes, ingredient prices, city adjustments, and a multilingual food embedding model. Nobody has assembled all four.
  • The data needs to be fresh.
    Menus and prices change daily. A one-time scrape is worthless.
  • Consulting firms are slow and expensive.
    A competitive audit costs 5-10L and takes weeks. We do it for INR 2K-12K in 15 minutes.
Defensibility

Proprietary AI systems that compound. Each layer makes the next one stronger.

1

Proprietary food embedding model

  • 98.5% accuracy, global dish coverage, works across cuisines and languages
  • Already monetized: abCoffee uses it for app search
  • Replicable to any food platform for search, recommendations, deduplication
2

AI-generated recipe database

  • 63,516 recipes with gram-level ingredient breakdowns
  • Generated by LLMs, verified through multi-stage quality gates
  • Self-improving: 40K grew to 63K organically via AI flywheel
3

Replicable AI systems

  • Each AI subsystem (recipe gen, grey-zone verifier, re-ranker, alias expander) is modular and reusable
  • Adding a new country = plugging in new data, not rebuilding the stack
4

Daily refresh + data history

  • 10 automated jobs run every 24 hours
  • Competitors starting today are 12+ months behind on accumulated data
5

Platform-agnostic, globally extensible

The AI pipeline works on any food delivery platform's data. Swiggy today. Zomato, Uber Eats, DoorDash, Grab, Deliveroo next. Same models, same code. The food embedding model is already language-agnostic.

Where We Are

Early, but the hard parts are built.

WHAT'S LIVE

  • Full product, end to end
    statode.com: search, pay, report in inbox. Razorpay integration.
  • Daily AI + data pipeline
    500K+ items, 100K+ restaurants, 9 cities. 10 automated jobs.
  • Food embedding model at 98.5% accuracy
    63K recipes, 242K matches. Global dish model.
  • First external AI customer
    abCoffee uses our embedding model for search on their app.
  • AI-powered report generation
    INR 50 per report. Template + AI narratives. Validated.
  • B2B sales pipeline
    100 target companies across 12 categories. Outreach prepared.

KEY METRICS

Infrastructure cost$20/mo
Per-report COGS (AI)~$0.60
Avg report price~$60
Gross margin~99%
Report delivery time~15 min
Food embedding accuracy98.5%
Recipe catalog63,516
Cities covered9 (scraping) / 11 (B2B prices)

NEXT 6 MONTHS

  • Close first 5-10 B2B data contracts
  • Monetize food embedding model to more platforms
  • Add Zomato + scale to 15+ Indian cities
  • Recurring subscription tier for chains
  • Begin international platform research
Timing

Three things changed.

AI made this category possible

  • Generating 63K recipes with ingredient detail would have required food scientists. AI does it for $72.
  • Embedding models are now accurate enough for domain-specific food matching.
  • LLM classification pipelines can verify uncertain matches at scale.
  • This is a new product category that AI created.

Restaurant chains are professionalizing

  • Swiggy and Zomato IPO'd.
  • Cloud kitchens scaled to hundreds of outlets.
  • Chains are raising capital, hiring analysts, looking for data.
  • The buyer didn't exist five years ago. Today a 15-outlet chai chain has a growth team.

Supplier-side data demand is real

  • FMCG companies entering food service need to understand what restaurants serve.
  • Field sales teams run on anecdotal data.
  • The food service supply chain is digitizing, but demand-side intelligence doesn't exist.

The combination is new: AI capable enough for food cost estimation and semantic food understanding, mature delivery platforms globally to scrape, and a restaurant industry with data buyers on both sides. Our AI pipeline is platform-agnostic: Swiggy today, Uber Eats, DoorDash, or Deliveroo tomorrow.

Team

Small team. Full stack.

Aditya Patni

Founder & Engineer

  • Built the entire platform: food embedding model, scraping infrastructure, food cost engine, analytics pipeline, report generation, payment integration, API, and sales pipeline
  • Full-stack across data engineering, ML/NLP, backend, and infrastructure

Koja (KJ)

Engineer

  • Pipeline operations and infrastructure

What we've built with this team

98.5%
Food embedding accuracy
63K
AI-generated recipes
10
Automated daily AI jobs
$20/mo
Total infra cost
The Ask

Pre-seed round

Use of funds

  • Scale the food embedding model
    More platform integrations, international cuisine coverage, embedding-as-a-service product
  • Expand data coverage
    Add Zomato, scale to 15+ Indian cities, begin international platform research
  • Hire 1-2 AI/ML engineers
    Improve embedding models, expand recipe generation, scale AI systems
  • B2B sales
    Close first enterprise data contracts with FMCG and supplier companies

What you get

  • Working product with 99% gross margins
    Not a prototype. Customers can buy reports today.
  • Proprietary AI with first external customer
    Food embedding model already in production at abCoffee.
  • Three revenue streams
    B2C reports + B2B data + AI embedding model. Same engine, three markets.
  • Globally replicable AI systems
    Platform-agnostic. India today, any food delivery market tomorrow.
  • Capital-efficient team
    Entire platform built and run for $20/month in infra.
Aditya Patni
aditya@statode.com  •  statode.com