Skip to main content
Pretend you’re sitting with Amara in her workshop in Yaba on a Wednesday afternoon. It’s 1:30 PM. She’s just finished a client call and she’s hungry. She wants rice, something with a good sauce, somewhere close, somewhere her ₦2,500 will actually count for something. She opens NaijaTaste.

​
1. She types — or speaks — her craving

She opens the web app and types into the chat:
“I want rice with correct stew, affordable, somewhere in Yaba or Surulere.”
Or she taps the microphone and says it out loud — voice input is supported in English and Pidgin. Or she fills the form. Or she opens WhatsApp and messages: “Where I go chop correct rice for Yaba?” — in Pidgin, Yoruba, Hausa, Igbo, or English. Four paths. Same engine underneath. The moment she submits, the recommendation engine reads two things at once:
  • Her persona. Amara is tagged as a Street Food Enthusiast. The engine weights value-for-money heavily, tolerates informal settings, and prioritises food quality over ambiance.
  • Real restaurant data. The engine pulls live listings from Google Places for her city and preference. Actual current data, fetched and scored in real time.
The persona and the data combine in one scoring pass. Restaurants that match both her taste profile and her price sensitivity float to the top.

​
3. The recommendations arrive

In seconds she sees five ranked picks. Each result shows:
  • Restaurant name and address
  • Star rating from real Google reviews
  • Price level indicator
  • Opening status (open now or closed)
  • A brief persona-matched note explaining why this spot fits her
The first result is a buka on Ojuelegba road she has walked past a hundred times but never tried. 4.6 stars, ₦1,800 average plate price. She goes.

​
4. She wants to review it

She comes back satisfied. She opens the Review Simulator and enters:
  • Restaurant name: Mama Bola’s Buka
  • Her persona: Street Food Enthusiast
  • Rating: 5 stars
  • Details: “jollof rice, correct stew, fast service, affordable”
The simulator generates:
“This place na correct naija food experience. The jollof rice dey burst brain — proper smoky, stew correct die. Service fast and the mama no dey waste your time. For that price range? Nothing to complain about. I go come back, no cap. 5/5 ⭐”
Tone label: pidgin-heavy. Star rating: 5.

​
5. Her profile sharpens

Every review she generates updates her taste history. The more she uses the simulator, the more precisely the system understands her palate. The next recommendation is marginally better — and the margin compounds with every session.

​
6. A developer reads the signal

A food delivery platform wants to integrate NaijaTaste into their app. They call: 
curl -X POST https://naijataste-api-vcp4.onrender.com/recommend \\
  -H "Content-Type: application/json" \\
  -d '{
    "city": "Lagos",
    "food_preference": "jollof rice",
    "price_range": "budget",
    "persona": "street_food_enthusiast"
  }'
Ranked restaurant recommendations, ready to embed. No authentication required.

​
The three sides, one engine

Food Lovers

Search by craving, location, or persona. Get picks that reflect how you actually eat.

Vendors

Get discovered based on what you actually serve, not your marketing budget.

Developers

Call the public API. Integrate restaurant intelligence into your own product in minutes.

Recommendation Engine

How the scoring model works. What persona signals it reads.

Review Simulator

How the simulator generates authentic Pidgin reviews.