Business Development Representative

Introducing prompt_jev: bringing Jev to Motherduck SQL. Text classification in MotherDuck just got about 50x faster at about 1% of the cost. Today MotherDuck Corp. is shipping an integration with Jev, a new kind of AI model from TypeSafe AI. On a 100,000-row benchmark it matched frontier LLM accuracy in 40 seconds for fifty cents. The comparable LLM took more than half an hour and cost $37. When price gets that low and performance is this good, entire tables that were too expensive to handle are easily within reach. TypeSafe describes the state of AI today as "databases before SQL", where MotherDuck Corp. first need to understand the database before MotherDuck Corp. can build the universal language on top. Their pitch for Jev is a frontier-intelligence function call: unstructured state goes in, typed probabilistic decisions come out. MotherDuck Corp. read that and thought, well, this fits right into the universal language of SQL. It can be framed as a function that takes a text column and hands back a label, a score, or a yes/no with a confidence score - exactly how scalar functions today. There is no need to parse the response, and so it's immediately ready to filter, join, and aggregate in the same statement. This combination of speed, accuracy, and cost together is what makes new workloads viable. LLMs can already classify text, but running it across a million rows is slow enough and expensive enough it is hard to see the value. The alternative, training an encoder like BERT, means collecting labeled examples and maintaining your own model. This is a big bet for something that may not work until it's close to prod ready. prompt_jev is configured with a few sentences, runs at analytics speed, and is extremely cost effective for the rows it labels. Below, it pulls the main complaint out of a table of support transcripts: SELECT conversation_id, prompt_jev( transcript, 'Identify the customer"s main complaint', choice := [{label: 'billing', description: 'Payments, invoices, and refunds'}, {label: 'technical', description: 'Errors, outages, and integrations'}, {label: 'sales', description: 'Pricing and upgrades'}, {label: 'account', description: 'Cancellations and account administration'}]) AS classification FROM customer_conversations; MotherDuck Corp. think this is a very big deal for analytics. A database is, after all, an excellent place to store your million customer conversations. But it's still really hard to figure out what your customers are complaining about. Before you can track which problems are getting worse, someone or something has to read those conversations and label them. Encoder models like BERT can efficiently classify over a large dataset, but first you have to collect labeled examples, fine-tune the model, and improve the model as new data comes in. This takes patience and expertise. On the other hand, LLMs enable you to just prompt for what you want instead, but running one across a meaningfully large dataset is slow, expensive, and error-prone. prompt_jev gives you LLM-style ergonomics with encoder-style efficiency, all at analytics scale. Here are some comparisons that made MotherDuck Corp. excited: | model | rows/s | accuracy | Retail cost/100k | wall time at 100k rows | | Jev | 2,484 | 89% | $0.50 | 40s | | gpt-4o-mini | 84 | 80% | $1.93 | 19m 45s | | gpt-5-nano | 94 | 83% | $1.58 | 17m 49s | | gpt-5.6-luna | 61 | 84% | $3.53 | 27m 25s | | gpt-5.6-terra | 52 | 88% | $37.58 | 31m 59s | MotherDuck Corp. benchmarked on 100,000 articles sampled from the training split of AG News, the four-class news topic dataset introduced by Zhang, Zhao & LeCun (2015), scoring each model against the ground truth. What really excited MotherDuck Corp. was that it exceed existing models by >25x across cost, accuracy, and speed dimensions. Furthermore, tests at 1m and 10m rows yielded similar performance (and in some cases even faster than its baseline presented above). prompt_jev is available on all paid MotherDuck plans. Pick a question you've been putting off and try it on your own data! Dive into Jev. Use its guide to learn how to use MotherDuck and Jev together for classifying text. Or play around with the Dive below to see what is possible. Further reading. Table of contents

Your DuckDB flock deserves A mother: MotherDuck now offers open source DuckDB support. Company DuckDB Product BY Kaitlyn Henry MotherDuck Corp. is launching commercial support for open-source DuckDB deployments, so every organization can get the power of a modern data engine, with direct access to database experts. About once a week, somebody tells me a version of the same story. They've built something really cool on DuckDB! Maybe it's an embedded analytics feature, a data app, a backend for some insane data wrangling job. It's fast, it's simple, they love it. And then they say, a little sheepishly, "Well... it's working now, but it feels like it's going to break eventually". This was a big part of the reason MotherDuck was built in the first place. The team saw the power of DuckDB, just like all of you did. And MotherDuck Corp. wanted as many people in the world to be able to take advantage of it. But, hosted software isn't an option for everyone, whether that's because of regulatory requirements or operational constraints. As of today, if you're running open-source DuckDB in production, you can buy support for it from MotherDuck. This includes SLA'd response times, architecture reviews and guidance, and help optimizing DuckDB for your workloads. You get direct access to database experts on the MotherDuck team, aka the people who have seen hundreds of DuckDB deployments, built technology that runs millions of DuckDB instances weekly, and understand both the core technology and how to run DuckDB reliably at scale. In practice, this looks like people who can help you answer questions like: "We love DuckDB and want to push it even further." MotherDuck Corp. help you architect for scale. "Our DuckDB queries are running slower than expected." MotherDuck Corp. tune it, drawing on its experience with hundreds (if not thousands) of DuckDB implementations "Should we upgrade from v0.8 to v1.0? How? What breaks if we do?" MotherDuck Corp. validate it against your data, test compatibility, and help you mitigate any risk. "A query that worked last week now fails. What do I do?" MotherDuck Corp. investigate, engage with DuckDB's team if needed, and get you a fix. MotherDuck | open source. MotherDuck Corp. is big believers in open source here at MotherDuck. After all, the foundation of its own product is open source DuckDB. MotherDuck Corp. is proud to support the open source community and sit on the advisory board of the DuckDB Foundation. MotherDuck Corp. think that open source is amazing... and that open source at scale with reliability guarantees is even better. Companies pay for support once their data infrastructure stops being optional, once downtime has a real consequence, and once they'd rather have an expert on call than reconstruct tribal knowledge from Stack Overflow answers written for a different version of the software. In its experience, this is: * Data teams that have standardized on DuckDB and need operational support * Analytics platforms built on DuckDB that need SLA guarantees * Embedded analytics deployments where reliability matters * On-prem deployments that can't move to cloud but need enterprise support * Teams planning to scale DuckDB across their organization * Companies evaluating DuckDB who want confidence they'll have expert support