Pienso

Pienso is a no code platform for exploring language data and training machine learning and large language models on it. It is aimed at subject matter experts, analysts and data scientists who hold the domain knowledge but do not write code, and who want to find themes in large volumes of unstructured text such as emails, contracts, call transcripts and forum posts. The workflow has four stages. Ingest accepts raw text and sorts it into topics. Train and Fingerprint lets the user refine those categories in an interactive workspace and train models on their own judgments. Annotate applies the trained models to label new datasets. Data Set Analysis provides visualizations and adjustable thresholds for deeper exploration. A companion tool, PromptFactory, helps users build, test and deploy production grade prompts without code. The vendor says customers can fine tune models with visibility into their parameters, own the resulting models, and pay for fine tuning only when they deploy. Typical uses named on the site include analyzing customer conversations at scale, content moderation and tracking how language changes over time. The company grew out of MIT research on content moderation and puts human expertise in the loop at the center of its design. Pricing is not published. The site directs prospective customers to book a demo.