pablonm

I predict 2026 is going to be the year of disillusionment with LLMs, when we realize that this technology has reached its theoretical limit and won't keep improving and I wanted to be early to the party so wrote an article where I analyze the current stage of affairs

The Einstein of chimpanzees

I noticed Claude Code (on a 100$ max subscription) has become slower for me in the last few weeks. Just yesterday it spent hours coding a simple feature Which I could have coded myself faster.

Yes, In order to provide features like semantic search and autocategorization/clustering the system will need to store and process private information. The guidelines that I follow are about using trustworthy technology companies and reduce the number of services that handle data to a minimum: - Store data on S3 buckets and digital ocean servers protected by all types of security offered: 2FA, ssh key, IAM, etc. - For AI models at first I will use openAI and anthropic APIs. In the future I would like to run my own models for better privacy. - Don't use any other service, everything else runs locally inside digial ocean server, eg: vector databases, mongodb, etc.

So it's not an app to store passwords but I think these measures should do for general notes and files.

While working as an NLP freelance engineer I found several clients who wanted to train their own text generation model for different applications like Creative Writing, comment generation, etc. I've developed a service that takes your text dataset and trains a text generation model that then you will be able to use through a simple API. Any feedback is more than welcomed!