Everyone at @artie_labs can see our revenue and burn because people make better calls when they know the actual numbers. Got into this and more on @Composio’s Agents at Work.
Thanks for having me @juliafedorin!
Show more
Every single customer told Jacqueline Cheong (@JacquelineSYC19) they didn’t want the feature. She built it anyway. Six months later, 40% of them had turned it on.
She’s the co-founder and CEO of @artie_labs, which moves data between databases in real time: the unglamorous plumbing that lets AI agents reason over what’s happening now, not what happened yesterday.
We got into:
•Building the feature customers said they didn’t want
•Founding a company with your spouse (and why she’s loud about it)
•Why her whole team sees the revenue and burn numbers
•Why she stopped watching the metric and started watching the inputs
•Why real-time data is the layer AI agents actually need
𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Transamerica, and Why This Spot
(02:00) What Artie Actually Does
(05:00) Why Real-Time Data Is Hard
(08:00) The AI Agent Angle
(12:00) The History Table Everyone Said No To
(18:00) Founding a Company With Your Husband
(24:00) Radical Transparency With the Team
(30:00) The “Working With Me” Doc
(34:00) Focus on Inputs, Not the Score
(38:00) The Wedding Cake Metaphor
This is a @Composio “Agents at Work” podcast, where I chat with founders building the next leap of AI.
Follow for more :)
Show more
This year has been the most exhausting of my career. It’s a great sign for Artie.
In 2024, roadmap planning was @robin_codes, our founding engineer Dana, and me asking each other what we should build next. Nobody was asking us for anything. Engineering backlog was nonexistent, and it was depressing at times.
This year, we barely have time to ask what to build next. Customer requests are so overwhelming that can we no longer deliver on new features in a few days time. Most of our roadmap now comes from teams running Artie in production, and balancing between customer requests and building towards our vision is a constant balancing act.
You'd think I would complain about being so exhausted all the time, but I wake up everyday grateful we're in this position now. The alternative is way more stressful.
Show more
Throwback to March 2025, when every new Artie customer was greeted by a dog in a cone of shame.
Looking at the old dashboard next to what our product designer, @RobbieBernal, built this year, it’s hard to believe it’s the same product. More to come!
Show more
A startup can get bigger but slower at the same time. This is actually the normal path, and you have to work hard to overcome that reality.
More than doubling Artie’s headcount this year showed me that a good chunk of my job is alignment: making sure everyone knows what we’re building and why.
When everyone’s aligned, the math is simple. 1 + 1 + 1 = 3. Three people fully aligned means you get 3x the amount of work done - sometimes even more.
When someone’s not fully aligned, even a great hire adds less than a full 1. Good work goes in the wrong direction, and the team spends time getting back in sync. Add enough people that are misaligned and a team of 6 ships less than a team of 3.
That’s why we're keeping Artie as lean as we can as we grow. More people only makes the team faster when everyone’s pointed the same way.
Show more
Real-time data used to be a hard sell. I can't tell you the number of times from 2023-2025 that someone told me on a sales call that "hourly latency is completely fine for us, we don't need real time".
Before last October, latency barely came up when teams reached out to us at @artie_labs. They wanted to replace tools that were unreliable and kept breaking.
In the last 12 months, about one in five have reached out because hours-old data isn’t good enough anymore.
What changed is what the data is used for. When the main consumer was a dashboard someone checked every morning, a nightly batch did the job. Now teams are feeding data into products, models, and agentic workflows that act on changes as they happen. Stale data there means a wrong decision, made automatically, at scale.
The world is changing incredibly quickly right now. Exciting times!
Show more
Firecrawl is the API for the web. More than 1.5 million developers build on it.
The company is growing fast, and the data is growing with it. With one data engineer, Rahul Singh, @firecrawl uses Artie to move around 4 billion rows a month from Postgres into BigQuery. That data feeds every dashboard, usage report, and spending decision the company makes.
Since Firecrawl started on Artie in December 2025, it has raised a $75M Series B, and the pipeline hasn't had to change. When Rahul moved a large share of the data from Supabase to PlanetScale, he used Artie for that too, and the replication side was done within two days.
In his words: "It just runs in the background, and I don't have to worry about it."
Show more
A team who has tried every data pipeline tool on the market told us Artie is the best one they've used.
Their data team is experienced but small. They ran their pipelines on Stitch for years and have tested numerous alternatives like Fivetran and Airbyte as their use case and scale evolved.
They found Artie on their own, signed up, and had pipelines running before our first call. Less than three weeks after signing up, they became a customer.
"The best service we've looked at" hits differently from a team that's looked at all of them.
Show more
We built a film studio inside the Artie office.
Before this, recording a video meant our marketing team had to find and book a space, haul gear across town, and lose half a day to setup. It was operationally heavy, so most video ideas died in Slack.
When we moved offices, we decided to turn a corner room into a permanent studio with acoustic panels and a full lighting rig. Now anyone on the team can walk down the hall and start recording.
Nobody explains Artie better than the people who built it, and that's who we want in front of the camera. An ad tells you what CDC does, but an engineer showing you how we catch a schema change tells you whether to trust it.
You'll be seeing a lot more of our engineers 👀
Show more
Aurora PostgreSQL can now query Apache Iceberg tables in S3 directly. You can join live Postgres rows with years of history in your lake in a single query, without a pipeline copying that data back into the database. It uses DuckDB embedded in Aurora, following AWS’s acquisition of DuckLabs (the team that maintains DuckDB).
This solves the read side. But something still has to keep those Iceberg tables current, through schema changes, backfills, and high write volume. That’s what we do at @artie_labs – we stream changes from operational databases into Iceberg-backed tables in Amazon S3 Tables, so the history Aurora reads stays fresh.
Show more
Artie supports fan-in across Postgres schemas, MySQL databases, and @PlanetScale keyspaces. Consolidate every tenant’s operational data into one continuously updated destination table, without centralizing your application database.
Multi-tenant applications often keep each customer, region, store, or business unit in its own database namespace. Instead of managing separate copies of the same data across every tenant, Artie continuously fans in matching tables into one destination table.
Use an explicit allow list when the set is stable, or use regex rules to automatically include new tenant databases as they launch, with no manual pipeline updates needed.
That means fewer one-off pipelines, less downstream UNION logic, and one canonical dataset for customer-facing applications, automated workflows, AI applications, and analytics.
Show more
People often ask me if Artie's logo was inspired by Postgres. In some ways, it was.
Postgres's history is the history of CDC. It turned 30 this year, and for its first 18 years, it had no clean way to tell you what changed. You either polled tables for new timestamps and missed every delete, or put triggers on every table and traded off with additional load and storage.
Then in 2014, Postgres 9.4 shipped logical decoding and replication slots, and it could finally stream row-level changes straight out of the WAL. Nearly every CDC tool that supports Postgres today, Artie included, exists because of that release.
Postgres picked an elephant because elephants never forget. Artie exists because Postgres remembers every change in the WAL, and we read it.
Happy 30th, Postgres! 🐘
Show more
2024 was the year I went into my slump.
We missed the revenue goals I'd set over a few consecutive months, and every lost deal took a bit of my confidence with it. The problem was that I'd turned those revenue targets into a scorecard for myself, but no matter how well I ran each deal, “close a deal every week” is an outcome I don’t fully control. I knew that, yet every miss felt like a personal failure.
Losing confidence is a massive problem for a founder, and bleeds into many negative downstream effects, like no longer being able to make decisions under uncertainty. I knew I needed to change.
So I changed what I counted as a personal win. Did I think through the next step on every deal that day? When I noticed a problem, did I say it right away instead of sitting on it? Did I ask the question that felt uncomfortable to ask? Those are yes/no questions, and the answer is entirely within my control.
I gave myself a target: three yeses in a week, and I'd won the week, whether or not we closed anything.
I gained my confidence back by focusing on the right inputs, not the outcomes.
Show more
Our team at @artie_labs is a mix of high alpha and low beta people.
I watched an interview of @stanine (CPO at Rippling) and this is how he described hiring. I came from a hedge fund, so it really resonated with me.
Alpha is outperformance relative to the index (they're "spiky"). High alpha people have a ton of upside and thrive in environments with low process. They’re great at doing 0 to 1.
Beta is how much you swing (volatility). Low beta people are very reliable and predictable, and thrive with process. They ship what they said they'd ship, when they said they would, and you never have to check.
A company needs both, but you almost never get both in one person. So I use it as a 2x2:
- High alpha, low beta: extremely rare. If you meet one, hire them.
- High alpha, high beta: good. You're get the spikes and you budget for the misses.
- Low alpha, low beta: good. Dependable, and every company runs on them.
- Low alpha, high beta: no. All the volatility, with none of the upside. This is the hire that kills companies.
Everyone here is one of the first three.
Show more
someone signed up for @artie_labs using INSTINCT!!
I was always curious what other people are using Instinct for and never ever imagined someone would try to set up a data pipeline with it
I saw something last week that blew my mind – @artie_labs showed up in job descriptions that weren’t ours.
I was looking through open roles at Clio and Alloy and saw Artie appear in both. Our name, right next to Snowflake, Databricks, and dbt.
Case studies and logos on a website are things we control, but showing up in another company's job description isn't.
I'm so honored and proud of the team. Back to work.
Show more
Great day at Rows & Columns Summit – was awesome seeing two talks mention @artie_labs!
1. Our founding engineer William Haw joined James Sutton, CTO at ShipScience, to share how ShipScience uses Artie’s streaming CDC pipeline to move row-level changes from its production MySQL database to MotherDuck. The pipeline handles roughly 30 tables and 3 billion row updates each month, keeping analytics and agent workloads off the production database.
2. Shalaka Borker, Director of Data Engineering and Analytics at Roofstock, mentioned Artie in a discussion of moving reporting workloads off operational databases and onto a continuously updated analytical layer. Their report-rendering times dropped from minutes to seconds.
Big thanks to the @clickhouseDB team for bringing together a room of data practitioners to discuss the evolving world of transactional and analytical systems. We’re already looking forward to next year’s summit.
Show more
Artie is at Rows & Columns Summit today at the Contemporary Jewish Museum in San Francisco!
At 2 PM, our founding engineer William Haw will join James Sutton, CTO of ShipScience, for a talk on how ShipScience moved 3 billion MySQL row updates per month out of production and into MotherDuck with Artie.
If you’re at the summit, come say hi!
Show more
A few days ago a friend called and asked, “What makes someone qualified to be a founder?” She had been thinking of starting her own company.
I thought about it for a minute, then said: "It comes down to whether or not you believe that you can do it". She was a bit shocked - is it really that simple?
I think it is. But believing is one of the hardest things to build up.
Show more
We've heard this from customers for two years: the hardest part of CDC is the months of plumbing to get it running. Now you can get your pipeline live with a single prompt.
Standing up a production CDC pipeline used to mean designing source access, replication, backfills, destination writes, schema-change handling, monitoring, and recovery.
Now, it can done with a single prompt: “Create a pipeline from the staging Postgres database to Snowflake.”
With Artie MCP available in Claude Code, Cursor, and Codex, an AI agent can use Artie’s tools to configure and manage that workflow. So engineers can start with what they need, instead of spending months assembling and maintaining the pipeline themselves.
Show more
A few weeks ago I wrote that our whole team at Artie uses Hermes (an AI agent harness from @NousResearch) to do their work.
Here's how we got there: what we tried first, what it cost us, how we built a Hermes for each team, and why they dream at night.
Show more
