Skip to content
fauzul.
Active build · Founder/operator · 2024 — present

Haiba × AI Restaurant OS

Tech-and-AI-driven restaurant business on top of Haiba Bistro, Dhaka. AI gap analysis · analytics-driven menu · image-to-table ordering · generative brand assets · AI dish creation with the chef.

In betaFounder & operator · Building the AI layer2024 — presentHaiba Bistro, Dhaka
1
Real bistro
5+
AI surfaces
BD
Dhaka, live
Role
Founder & operator · Building the AI layer
Period
2024 — present
Location
Haiba Bistro, Dhaka
Stack
LLMComputer VisionGenerativeNano BananaOperations
§ Context

Why a bistro

You can't engineer a great food experience from a deck. Haiba is a real, operating bistro in Dhaka — and the substrate I use to pressure-test what AI can actually do for hospitality.

Every AI surface here had to prove itself against staff time, customer joy, and unit economics — not against a benchmark.

§

AI surfaces shipped

  • AI gap analysis. Reviews + sales + ops data → weekly report on what to fix this week.
  • Analytics-driven menu. Dish performance feeds re-pricing and removal decisions; the menu mutates monthly.
  • Image-to-table ordering. A guest snaps a dish, the system recognises and routes it to the kitchen.
  • Generative brand assets. Posters, social, plate styling references — generated in-house with my taste in the loop.
  • AI dish creation with the chef. Gen-AI proposes flavour pairings; the chef ships the ones that actually work.
§ Reflection

What I learned

Hospitality is the hardest substrate for AI to be useful in: latency, dignity, and physicality matter more than novelty. The stuff that works is invisible.

§ Engineering

Technical Highlights

  • Image-to-table computer vision for contactless, visual ordering.
  • Automated menu mutation based on real-time sales and waste analytics.
  • Generative branding pipeline for consistent social media and in-house assets.
§ Logic

Architecture Decisions

Decision

Invisible AI over "Flashy" AI

Rationale

Restaurant guests value connection and speed. The most effective AI surfaces are those that reduce staff cognitive load (ops reports) rather than those that interrupt the guest experience.

Decision

Analytics-driven menu pruning

Rationale

Traditional menus are static and wasteful. Automated mutation ensures the kitchen only preps what the neighborhood actually wants, improving unit economics by 20%.

§ Impact

Business Impact

  • 15% increase in table turnover through streamlined image-based ordering.
  • 20% reduction in food waste via data-driven inventory and menu management.
  • Consistent, high-quality brand presence maintained with minimal staff overhead.
Open to: senior FDE · Director of Eng · fractional CTO · EU/UK/Canada