Companies / Data vendor
Chakra Labs
Builds expert-authored tasks, benchmarks and training datasets for companies training frontier AI models. Recorded here: what it sells, in its own words.
chakra.devLatest record 15 Oct 2026
What Chakra Labs sells
In its own words, read on its own site 15 Oct 2026. On the vendor register
- What it sells
RL environments: “Each domain runs as its own factory: tasks written by experts, graders built for real work.” chakra.dev
Expert and human data: “Legal tasks authored by seasoned practicing counsel in a wide range of practice areas.” chakra.dev
Evaluations and benchmarks: “Chakra Labs has built a new benchmark to measure this capability directly: MAGI (MCP and GUI Benchmark).” chakra.dev
Datasets: “This dataset contains authentic computer interaction data collected from users performing real work tasks in productivity applications.” huggingface.co
- Where its data comes from
Made by experts and contractors: “Tasks in MAGI are written by expert operators who spend two to three days immersed in the full Company Soul environment before writing a single prompt.” chakra.dev
Replicas of real software: “MAGI consists of 1,000 (200 test / 800 train) tasks mimicking real work spread across 8 cloned enterprise applications.” chakra.dev
Recorded from people and the real world: “The data was collected through Pango, a crowdsourced platform where users are compensated for contributing their natural computer interactions during actual work sessions.” huggingface.co
Generated: “Synthetically generated task instruction for training purposes. Represents the context of the full task.” huggingface.co
- Buys data from owners
- Its pages do not say it does.
- Areas its pages name
- Legal, Finance, Business work, Computer use
Cite this
AIData.Broker, “Chakra Labs in the AI training data market”, reviewed 15 Oct 2026. https://aidata.markets/companies/chakra-labs
Free to reuse under CC BY 4.0 if you credit AIData.Broker and link to this page. Each row names the publication that stated it, and words quoted from a source keep that source's own terms.