Tomorrow Unpacked is an independent publication by Rajath Vanakudare about AI developments, everyday applications and research.
My current independent research explores selective verification and structured editing: when another attempt helps, when a tool cannot express a request, and how to evaluate those boundaries without confusing a tidy output with a correct one.
This notebook also follows AI beyond those projects: new models, everyday tools, and changes in how people use them. The aim is to make a technical development understandable through a concrete example and a useful question.
Editorial approach
Research notes distinguish proposed methods, synthetic examples, preliminary observations, and completed findings. Sources are linked where they support a claim. Substantive corrections will be dated on the affected article.
AI tools assist with source research, coding, and drafting. Individual articles describe how they were prepared. Reporting from documentation is distinguished from hands-on testing, and company benchmark claims are identified as such.
The work presented here is personal and independent. The site contains public-source explanations and synthetic examples; it does not describe confidential employer projects or represent an employer’s views.
Follow the work
Subscribe to the RSS feed or browse the notebook. New notes will follow the questions as the evidence develops.