Generate any data, in any format, of any structure, in any database. All on demand. You can now get complex, realistic synthetic data just by asking for it. We built an AI Agent that turns natural language into comprehensive datasets for AI and software development. 1. Create structured SQL databases and unstructured files (PDF, DOCX, JSON) in the same workflow. 2. Describe the data shape, edge cases, and relationships in plain English. 3. Download immediately as SQL, JSON, CSV, and raw files or send data to any destination database of your choosing like @Oracle , @databricks , @Snowflake , @Supabase, Neon, @postgres , @MySQL , @SQLServer, and many more I'd love your support on Product Hunt: producthunt.com/products/ton… You can also try the Agent out for yourself. tonic.ai/fabricate

Shout out to @vercel for their outstanding streamdown library: streamdown.ai/. No more layout/style shift as markdown chunks stream in from Fabricate's data agent! Today we released Fabricate 3.2.0, which allows users to export to: - Parquet - Databricks - MongoDB - Oracle Generate synthetic data using Fabricate's data agent: tonic.ai/products/fabricate

The @tonicfakedata team is back and energized from our annual offsite glamping at Zion National Park with @theautocamp We spent the week working, hiking, and bonding in nature. We're returning with fresh ideas and clear focus for 2026. It’s going to be a great year.

Generating a synthetic FHIR from @HL7 resource is a foundational capability for developing healthcare software and AI systems, but generating a realistic patient journey is the true challenge for building advanced agents. A real patient journey has temporal and causal relationships. Symptom → Diagnosis → Procedure → Medication. The events and the timing between them are critical for testing clinical AI. Most synthetic data tools miss this completely. Today, a Fortune 500 healthcare company asked us at @tonicfakedata if our Data Agent could model this entire care continuum, including the complex inter-resource relationships. The answer is yes. It's not just a data generator; it's a system simulator. It understands the logic of a clinical pathway and can generate the chain of #FHIR resources to match.

The AI race just shifted again. Qualcomm’s new chips are gunning for Nvidia & AMD. Meta’s spending $27B on data centers. Open AI lined up deals for $1.5 million in chips over the next several years. It means the battle is now infrastructure over models. Whoever owns the silicon, supply chain & rack scale footprint is setting the pace. Whoever controls the racks controls the race.

LLMs have already consumed all human knowledge. Now the frontier labs are paying 7 figures for human generated data. Why not just use high quality synthetic data that is indistinguishable from the real thing?