ParadeDB takes on Elasticsearch as interest in Postgres explodes amid AI boom

Open source database management system Postgres is nearly 40 years old, but has recently started seeing explosive demand due to being very well-suited for AI applications. Despite this rise in popularity, search and analytics functionality remain limited. ParadeDB is changing that.

ParadeDB is an open source Postgres extension that facilitates full-text search and analytics directly in Postgres without users needing to transfer data to a separate source. The platform integrates with other data infrastructure tools, including Google Cloud SQL, Azure Postgres, and Amazon RDS, among others.

Philippe Noël, the co-founder and CEO of ParadeDB, told TechCrunch that he and his co-founder, Ming Ying, CTO, got the idea for the company by running into their own Postgres search woes while running their first startup, cloud-hybrid browser Whist.

“Postgres is becoming the default database of the world, and you still can’t do good search over that information, believe it or not,” Noël said. “There’s just a lot of pain points. So we decided to go and start a company to do that after talking to more people and realizing this was a very shared problem.”

ParadeDB isn’t the first company to try to solve Postgres search. Open source Elasticsearch is the most notable legacy player in the space since it was founded in 2012. Noël said that Elasticsearch works by moving data back and forth between itself and Postgres, which can work, but this system isn’t great for heavy workloads or processes that require frequent updating.

“That breaks all the time,” Noël said. “The two databases are not meant to work together. There’s a lot of compatibility issues, there’s a lot of latency issues, higher costs, and all of that deteriorates the user experience.”

ParadeDB claims to eliminate a lot of those challenges by building as an extension on top of Postgres directly, no data transfer required.

The startup was founded in 2023 and released the first open source version of the product later that year. Noël said that the group wanted to focus first on building out the open source product before it thought about sales or marketing.

That was short-lived. In early 2024, Chinese e-commerce giant Alibaba reached out and became ParadeDB’s first customer in May 2024. From there, the company turned toward building out its enterprise version of the software. Now it works with enterprises, including Modern Treasury, Bilt Rewards, and TCDI, among others.

ParadeDB just raised a $12 million Series A round led by Craft Ventures with participation from existing investors, including Y Combinator.

The company will use the funds for hiring. ParadeDB is currently a team of four and is looking to grow to at least 10 employees. Some of the funds will also go to continue to improve the platform’s user interface and analytics capabilities.

Noël said the company wasn’t looking to raise capital, but when Craft reached out to ParadeDB after being recommended to them by Supabase, the Postgres developer platform that has raised nearly $400 million of venture capital, they thought the timing made sense.

“It’s clear people are recognizing that Postgres matters a lot, and they want to get behind it,” he added.

Postgres’ popularity is clearly on the rise if recent M&A is any indicator. The recent acquisitions of Crunchy Data by Snowflake and Neon by Databricks were both ploys for the acquirers to gain more Postgres offerings, Devin Pratt, a research director at IDC, told TechCrunch last month.

“It is widely used across the industry, and I would say that was one of the main targets for their acquisitions,” Pratt told TechCrunch at the time. “The different strengths of Neon versus Crunchy [Data] were definitely debated between both companies and what would suit their companies best, but Postgres was their first goal in those acquisitions.”

ParadeDB hopes it can ride this wave of interest.

“We think building on Postgres is actually a meaningful shift, right?” Noël said. “Because that’s where the data is, and you can make a meaningful dent in Elasticsearch’s market share by meeting users where their data is, rather than building something that’s marginally faster or marginally cheaper.”