Case Study
Jsonify Unlocks Global Market Intelligence for Bacardi
Satwik
May 16, 2025
Highlights
8,500 AI agent hours logged — enabling real-time, structured insights from multiple sources
Cocktail extraction success where every other provider struggled
10x growth in data scale during the pilot—from 1.8K drinks on Day 1 to 1.5M+ by the end
Key Results
1.5 Million
Drinks Extracted
98%
Data Accuracy
6.5 Million
Pages Analysed
“The opportunity is validated. You did a very good job—your flexibility, the extractions, the speed… it’s quite impressive. This could really become a new capability for Bacardi.” — Jesus Checa, Chief Innovation Officer, Bacardi

Introduction
For Bacardi-Martini B.V., one of the world’s leading spirits companies, gaining real-time visibility into outlet listings is no longer a luxury—it’s a necessity. The company’s global footprint spans hundreds of markets, and understanding which brands are present, where they appear, how they are priced, and which competitors are gaining traction is vital for strategic sales execution and market growth.
This data is available on the internet, but it is spread across millions of constantly changing websites, documents, images, and other web properties, which makes it impossible to analyze at scale with the frequency required. Bacardi needed a new kind of solution that could visually interpret messy online data, extract structured insights at scale, and quickly deliver actionable intelligence. Jsonify’s AI-powered web agents and visual data extraction engine was purpose-built precisely for challenges like this.
Challenges
The main hurdles were firstly finding up to date venue data, across multiple geographies, then for each discovered venue, traversing the fragmented web to navigate, parse and extract data given the highly visual and varied nature of outlet website and menus across the globe.
This task required turning unstructured, inconsistently formatted websites and documents into structured, queryable insights. Traditional scraping tools struggled to parse visual PDFs, artistic layouts, or multilingual formats. Human-led data collection was slow, inconsistent, and costly. Legacy solutions lacked the flexibility to navigate websites, interpret diverse formats, and extract in a scalable way.

Above: A sample menu found during the pilot
Solution
Jsonify brought a new paradigm to Bacardi’s data operations: AI-powered, agentic, visual-first web extraction. Unlike traditional scraping tools that depend on strict HTML structure, Jsonify uses its proprietary vision models and large language models to interpret information as a human might—seeing information and reading them contextually.
Jsonify built sophisticated agent workflows that could navigate, understand and extract data from the web. Starting at key listing directories such as Tripadvisor or Timeout, the agent navigated through each directory, discovered venues, then located their website. It would then open millions of newly discovered venue websites, navigate each one dynamically, then find the menu listings on each.
This could be a webpage, document, image, or other format — it didn't matter. Jsonify was able to read and extract a common structure from each menu found.
Crucially, the Jsonify agent system does not have to be "taught" specifically about each website. It is able to understand and navigate the same way that a human does, looking visually at the page using a proprietary computer vision model. This means that it is resistant to page changes.

Above: A no-code workflow created in the Jsonify editor
Key benefits of the Jsonify approach included:
Comprehensive Discovery – Jsonify automatically discovered outlets using different sources.
Intelligent Data Access – Jsonify’s AI agents could locate and extract data buried within complex websites, whether in HTML or PDF format.
Product Matching & Classification – Jsonify matched extracted text to known Bacardi brands and products.
Multilingual & Global Scalability – Jsonify’s AI parsed content across languages and adapted to a wide range of regional formatting conventions.
Results
In 2025, Bacardi and Jsonify conducted a rigorous pilot across three key markets.

Above: A sample of extracted venues from London
Metric | Result |
---|---|
Agent hours logged | 8,500 |
Pages read | >6.5 million |
Venues discovered | >180,000 |
Output data rows | >3 million |
Data accuracy per eval | 98% |
Collaboration and Adaptability
Jsonify’s close collaboration with Bacardi ensured continuous refinement of the platform throughout the pilot. Feedback loops between both teams led to several product enhancements:
Improved PDF parsing logic to handle dense, stylized documents
Menu location tuning for sites with buried or unusual structures
Cocktail classification enhancements to better identify branded spirits within ingredient lists
Venue intelligence management for filtering out closed or renamed venues
Bacardi praised Jsonify’s flexibility and responsiveness, noting that the platform adapted rapidly to the demands of global data extraction.
The Path Forward
With the pilot proving Jsonify’s value, Bacardi is entering the first phase of implementation—actively working with Jsonify to ensure smooth adoption and integration into global commercial workflows.
To learn how Jsonify can help your business intelligence operations and achieve similar results, contact us here.
“We really liked how responsive you were and how quickly you adapted to our needs. The cocktail extraction, in particular, was a real differentiator—we hadn’t seen that level of depth before.” — Jesus Checa, Chief Innovation Officer, Bacardi
