The company that decided the answer was never in the rows and columns - it was in the connections between them.
LYNX ANALYTICS
The primary brand mark, photographed against studio white. Fifteen years, one idea: think in networks.
In a business-school classroom at INSEAD, a group of students and professors kept circling the same frustration. The tools of the day flattened the world into spreadsheets - customers in rows, attributes in columns - and then wondered why they kept missing the thing that actually mattered: how everything connected to everything else. Lynx Analytics was their answer. Apply graph theory, they reasoned, and the hidden structure of a business would finally become visible.
Lynx Analytics turns large, messy datasets into graphs - webs of entities and the relationships between them - and then runs analysis that traditional row-and-column tools simply cannot see. A telecom subscriber is not a record; she is a node connected to the people she calls, the towers she pings, and the plans she leaves behind. Modeled that way, churn, fraud, and high-value segments stop hiding.
The engine is LynxKite, a graph AI platform built to handle very large datasets in a no-code environment. Its processing core was built on Apache Spark for scale, and its newest release moves hundreds of algorithms onto GPUs.
For years, the customer list read like an Asian enterprise roll-call: DBS, Singtel, HKT, Vodafone - telecoms and banks with millions of customers and every reason to understand the graph beneath them.
More recently, the company has pointed the same machinery at a harder target: life sciences. Under an AI-native consulting model, Lynx now embeds engineers and scientists inside pharma and biotech teams, building tools for commercial and clinical decisions - from HCP engagement to drug discovery.
Figures compiled from company press materials and public databases (Crunchbase, Tracxn, PitchBook). Revenue and headcount estimates vary by source; treat as approximate.
“AI is the foundation of how we think, build, and deliver.”
Fraud, churn, and drug targets look like unrelated problems. Lynx treats them as the same one - a network no one has mapped yet.
By modeling relationships between entities, LynxKite surfaces anomalies and fraudulent behavior that slip past traditional, record-by-record analytics.
Customer graphs capture touchpoints across channels, revealing churn risk, high-value segments, and upsell opportunities that flat tables obscure.
With biomedical knowledge graphs, NVIDIA BioNeMo, RDKit, and Graph Neural Networks, Lynx targets preclinical and early-stage pharma pipelines.
Where many platforms bolt a graph feature onto a tabular core, Lynx built around graphs from day one - and open-sourced LynxKite 4.0 in 2020 to widen adoption rather than lock it up.
Rather than handing over a deck, Lynx embeds cross-functional teams inside client organizations, iterates rapidly, and takes ownership of the result in production.
The alternatives are familiar names - graph databases like Neo4j and TigerGraph, Amazon Neptune, broad data-science platforms, and a growing field of AI-driven drug-discovery specialists. Lynx's wager is that deep graph expertise plus embedded delivery is harder to copy than any single feature.
A complete graph AI platform that transforms very large datasets into graphs and runs complex analysis in a no-code environment. Version 4.0 was released as open source.
A GPU-optimized, composable AI orchestration platform for drug discovery - 600+ algorithms (100+ GPU-accelerated via cuGraph), NetworkX compatible, chaining Python, LLM agents, and tools like Claude Code.
A proprietary generative AI platform for building assistants, chatbots, and workflow automation for pharma commercial and clinical teams.
AI-based customer graph analytics for telecom, banking, and retail - churn prediction, segment discovery, upsell, and fraud detection, built on Apache Spark.
A hybrid of enterprise software and AI-native consulting: Lynx licenses and deploys the LynxKite platform while embedding cross-functional teams inside client organizations to build and run tailored solutions - charging for engagements and outcomes rather than seats alone.
Students and professors found Lynx Analytics to apply graph theory to complex business problems.
A funding round fuels the growth of the graph customer analytics platform.
The graph data science platform is released as open source to democratize graph AI.
CIOReview APAC names Lynx a top data analytics solution company.
The company repositions around pharma commercial and clinical decisions with LynxScribe.
A GPU-optimized graph AI platform for drug discovery debuts with NVIDIA BioNeMo integration.
Several founders went on to become professors and faculty directors of analytics centers at leading US universities.
Co-founded Lynx in 2010 with an INSEAD team and a vision to solve real problems with big-data graphs.
Drives the research and innovation agenda rooted in the company's graph-analytics heritage.
Chairman of the board and faculty at Columbia Business School.
Part of the founding team from INSEAD helping translate research into a business.
Integrates NVIDIA BioNeMo and cuGraph into LynxKite 2000:MM for GPU-scale, multimodal biomedical workflows.
Runs graph AI workloads on Nebius GPU infrastructure, featured in a Nebius customer story on scaling graph AI.
The LynxKite engine is built on Apache Spark for scalable, real-time graph computation.
In an AI market fixated on large language models, Lynx sits deliberately upstream - in the knowledge graphs and network structure that give models something reliable to reason over. Rooted in Singapore, with software that runs on-prem and in-cloud, it is well placed for the sovereignty-conscious enterprise AI now taking shape across Asia and life sciences.
The name is a pun. “Lynx” plays on “links” - fitting for a company obsessed with the connections in data.
Classroom origins. It began as the brainchild of INSEAD students and professors, several of whom later ran university analytics centers.
It talks to agents. LynxKite 2000:MM can chain Python code, LLM agents, and tools like Claude Code into reusable workflow boxes.
Speed as a feature. One GPU migration reportedly delivered up to 1,000x speed-ups on graph workloads.
It builds graph AI software and delivers AI-native consulting, turning large datasets into graphs to solve problems like churn, fraud, customer analytics, and - increasingly - pharmaceutical drug discovery.
LynxKite is Lynx's flagship graph AI platform. It converts data into graphs and runs hundreds of graph algorithms in a no-code environment; version 4.0 was open-sourced in 2020, and the newest LynxKite 2000:MM is GPU-optimized for drug discovery.
It was founded in 2010 by INSEAD students and professors, including Gyorgy Lajtai (CEO), Gabor Benedek, Miklos Sarvary, and Sander Swinkels.
Historically large telecom and banking enterprises such as DBS, Singtel, HKT, and Vodafone; more recently, global pharma and biotech organizations in life sciences.
The company has raised about $10M, with its disclosed round dated to 2016. It is headquartered in Singapore with a team of roughly 56-60 people.
Compiled from public sources: company website and press releases, PR Newswire, Help Net Security, Nebius, CIOReview APAC, Crunchbase, Tracxn, and PitchBook. Figures are approximate and drawn from public databases. Video links point to search results rather than a single official channel.