Who They Are Now
The tiger has been the company's mascot since 2017. In June 2025 it finally got top billing. Timescale, the New York database company that spent a decade making PostgreSQL faster at time-series, dropped the old name and adopted the one its team had been doodling on whiteboards for years. Tiger Data is not a pivot. It is what happens when a product outgrows its own name.
Walk into the company's Madison Avenue office today and the conversations are not about ingest rates and continuous aggregates - though those are still there. They are about AI agents that need to fork a 200GB production database in milliseconds, about hedge funds running RAG pipelines on tick data, about industrial chatbots that have to answer in real time without lying. The product still sits on PostgreSQL. The use cases sit somewhere new.
The Problem They Saw
The premise was almost embarrassingly simple. The world was going to keep generating timestamped data - sensors, markets, logs, clicks - and the prevailing wisdom in 2015 was that you needed a special-purpose database to handle it. InfluxDB, Cassandra, a long list of bespoke time-series stores. Each one a new operational burden. Each one a fresh way to lose your data at 3am.
Ajay Kulkarni and Mike Freedman noticed something most people had not bothered to notice: developers did not want a new database. They wanted Postgres to handle the load. Postgres they trusted. Postgres they could hire for. Postgres had transactions, joins, the entire SQL surface area, and forty years of accumulated wisdom. The problem was that vanilla Postgres choked on high-ingest time-series workloads.
The fix - if you can call inventing a hypertable a fix - was an extension. TimescaleDB shipped as open source in 2017, partitioned big tables automatically, compressed old chunks ruthlessly, and let developers query everything with normal SQL. It was the kind of solution that, in retrospect, seems obvious. Most good solutions do.
What is more interesting is what happened next. The team kept noticing that customers showed up for time-series and stayed for everything else. A user would arrive to store sensor data and then realize their reporting workload, their fraud detection job, and their nascent vector search all wanted to live in the same database. Each conversation widened the company's understanding of what Postgres could be asked to do.
The Founders' Bet
Kulkarni is a serial founder with an MIT degree and a tolerance for long arcs. Freedman is a tenured computer science professor at Princeton. The two of them did not invent the time-series problem; they refused to treat it as a separate problem at all. Their wager was that the database market would eventually consolidate back around PostgreSQL, and that whoever made Postgres fastest at the hardest workloads would inherit the future.
Ten years on, that bet looks less like a hunch and more like a thesis statement. Tiger Cloud now handles transactional, analytical, vector, and agentic workloads inside the same Postgres. The team didn't need to rebuild the database. It needed to keep extending it.
The culture grew out of that posture. Tiger Data is remote-first, engineering-led, and unusually patient about its roadmap. The team writes long internal essays before shipping major features, runs a vocal open-source community on GitHub, and treats the Postgres mailing list as something between a town square and a sparring gym. The tone is, broadly, polite skepticism aimed at every claim that something other than Postgres is the answer.
The Product
TimescaleDB remains the engine. Hypertables, continuous aggregates, columnar compression - all the original tricks - now sit alongside hybrid row-columnar storage, vector search, BM25 lexical search, point-in-time recovery, and workload isolation. Tiger Cloud wraps it all in a managed service with high availability and the kind of operational polish that lets a startup CTO sleep through the night.
Agentic Postgres
The most striking addition - launched late in 2025 - is Agentic Postgres. The idea is that AI agents want what humans don't: hundreds of throwaway copies of a real database to test against, run loops on, and discard. So Tiger Data built Fluid Storage, which performs instant copy-on-write forks of production data. Add an MCP server and a terminal-native CLI, and an agent can spin up a sandboxed Postgres in less time than it takes to refill a coffee.
A Brief Tiger Calendar
The Proof
The customer list is not a vanity wall. Around 2,000 paying organizations now run on Tiger Data's platform, spanning hedge funds and private equity firms doing market analytics, industrial manufacturers building chat-based operator interfaces, SaaS companies wiring up real-time dashboards, and AI-native game studios storing player telemetry. The company has reported mid-eight-figure ARR with growth exceeding 100% year over year - the sort of number people stop calling a "vanity metric" once it appears on an audit.
Funding by round (cumulative, USD)
The capital backing is heavy enough to take seriously - Tiger Global, Benchmark, Redpoint, Icon Ventures, and others have stacked $184.8M into the company across rounds, including a $110M Series C in February 2022. Partnerships with AWS and Azure mean Tiger Cloud lands wherever the customer's compliance team last waved a flag.
The Mission
The company describes its mission, when pressed, in stubbornly unglamorous terms: make PostgreSQL the fastest, most capable database for modern transactional, analytical, and agentic workloads. The phrasing matters. It does not say "build a new database." It does not say "disrupt the data layer." It says: take the one engineers already trust, and keep extending it until it is the only one they need.
This is not the most exciting positioning in a market full of vector-native upstarts and AI-flavored data platforms. It is, however, the position with the longest half-life. Postgres has outlived its own obituaries roughly once per decade since 1996.
Why It Matters Tomorrow
The next five years of software are going to be written, in large part, by software. Agents will spin up databases, run queries, branch state, roll it back, and try again. They will want what humans never quite did: instant disposability, machine-readable schemas, search that returns embeddings and keywords in one round trip, and a billing model that does not punish curiosity. Tiger Data is one of the few companies building for that user explicitly.
None of which would matter if the underlying database were exotic. The whole point - the entire wager - is that it isn't. It is Postgres. The same Postgres your last three jobs ran on. Just faster, larger, and considerably more comfortable around a tiger.
So return to that Madison Avenue office. The tiger on the wall is no longer just a mascot. It's the logo, the philosophy, and the bet: that the database of the future will look an awful lot like the database of the past, only it will know how to run.