Lume raises $4.2M Seed Round led by General Catalyst

We're excited to announce our $4.2M Seed Round, led by General Catalyst, with participation from Khosla Ventures, Floodgate, Y Combinator, Soma Capital, and angels from Opendoor, Predibase, Orrick, and more.

The Complex Challenge of Data Integrations

Since the birth of databases 60 years ago, moving data seamlessly between systems has been an unsolved problem. Despite technological advancements, data integration has largely remained a manual process due to the unique nature of every data system. 

Companies and vendors structure and interpret data differently, making reformatting data (known as data mapping or data transformations) a complex, human-centric problem. These variations are infinite and cannot be encapsulated in any one algorithm. Thus, data integrations are iterative and unscalable today. Companies dedicate disproportionate resources to integrating data, bottlenecking their scalability. Engineers, including data scientists, data engineers, customer success engineers, and solutions engineers, spend considerable time manually onboarding new customer data systems or creating data integrations for new applications.

The data speaks for itself: on average, prospects spend ~5 weeks to map to a new system, whether it’s a customer’s system or a new internal integration. Some Lume customers used to spend up to 4 months per customer strictly on data integration, with reports of teams spending over 11 months on the process.

After speaking with hundreds of software companies, it’s clear that data integrations, and specifically data mapping, are massive bottlenecks that drain engineering resources and limit growth. The core challenge is twofold: understanding the new data system and executing the engineering work to move data between systems. Lume’s AI addresses both challenges.

Lume: Automating Data Integrations with AI

At Lume, we are solving manual data integrations for the first time due to the advancements of large language models and our unique approach to data transformation. The key foundation lies in Lume’s semantic understanding of data, allowing our AI to understand the nuances between data systems and create the mappings between them. With this, Lume’s mission is to be the universal translation layer between systems.

Lume offers an API and Web Platform to easily integrate data between systems. Core features include:

• AI generation of mapping logic

• Comprehensive review and editing capabilities for mappers

• Scaffolding to manage and maintain thousands of mappers

• Embeddable API for integrating data mappers in code

• Web Platform supporting fully no-code data mapping workflows

• Connectivity to systems for direct data writing and reading

Core Use Cases

Lume handles three core use cases:

• Onboarding Client Data

• Normalizing Data from Multiple Sources

• One to Many Integrations

We serve multiple industries, including ecommerce, insurance, manufacturing, and financial services. Examples of our impact include:

• For a CRM & ATS mid-size enterprise, Lume achieved a 95% mapping accuracy for recent customer migrations, significantly reducing onboarding time by mapping and classifying data ontologies from customer systems to their internal system.

• For a growth-stage battery analytics company, Lume accelerated the mapping process by 75%, reducing it from several weeks to just 4 days. For this customer, Lume maps incoming battery data to their internal sensor specifications templates.

• For an early-stage financial product startup, Lume manages ~1500 live data pipelines piping millions of data batches, which are externally facing, continually adapting to new incoming data.

These examples highlight the common underlying theme of mapping data between unique schemas, where even discrepancies as minor as column name variations make this process time-consuming and near-impossible to automate. This worsens at scale. Clients previously allocated engineers, customer success teams, or offshore labor to analyze, map, and route incoming data to new systems. This process took months and incurred significant costs.

As data systems continue to evolve, Lume ensures that businesses are not just keeping up with AI, but staying ahead, as shown by the use cases above. Whether it's drastically reducing onboarding times or managing hundreds of data pipelines with ease, Lume empowers teams to achieve their goals faster and with greater confidence. In the long term, Lume will be the glue that connects any data system.

Funding and Future Plans

The funding will enable us to make strategic hires, particularly in AI research and data infrastructure, to manage the increasing data load across customers and address the core, pervasive challenges this industry has faced for decades around data mapping and integration. 

We have an exciting roadmap ahead, including an enhanced testing and validation suite, embeddable UIs for customer data uploads and edits, and a direct connector suite to applications and databases.

If your team spends time mapping data for customer onboarding, data normalization, or data integrations, we can help! The best way to learn more about Lume is to book a demo - find a time here.

See Lume's exclusive feature on TechCrunch!

Lume Cofounders Nicolas Machado (left), Robert Ross (middle), Nebyou Zewde (right)