Show HN: Long term memory agent for "real world" data work

Hi HN!

I am an FDE and in all my clients' projects I have encountered the same problem: real world data preparation is the bottleneck. Company data doesn't only come with the usual Kaggle problems, it also has a wide range of other problems: corrupted migrations, multiple versions of taxonomies, duplicate columns and many more. Each problem requires a decision and portia helps accelerate all those and keep everything in a long term memory making work sessions persist from one chat to another. Today portia works with Claude Code, Codex or a local model (llama.cpp or ollama). The data can be a Postgres database, CSV files or hosted in the cloud (Snowflake or BQ).

Happy to answer any questions!

Comments URL: [https://news.ycombinator.com/item?id=49993790](https://news.ycombinator.com/item?id=49993790)

Points: 1