Research

We read the papers so the decisions in your build are not vibes. Every claim we make about what works is traceable to something someone published and put their name on.

What it is

The research desk is a public, dated log of the papers and benchmarks behind how we build, each credited to its real authors and annotated with what it changes in production. It is the reason our recommendations come with citations.

We also take on research work directly: feasibility studies, literature reviews, and the unglamorous job of finding out whether the thing you want to build has already been tried and why it failed.

How we build it

  1. Find the primary source

    A claim in a blog post is not evidence. We go to the paper, the benchmark, or the repository, and we record where it came from.

  2. Separate the result from the marketing

    Benchmarks are run under conditions nobody ships in. We read the setup before we quote the number, and we say so when a result does not transfer.

  3. Write the production consequence

    Every entry ends with what changes in the systems we build. Research that does not alter a decision is reading, not research.

Who it is for

A good fit if

  • Teams that need a defensible reason for an architecture choice.
  • Anyone evaluating a model or a vendor on claims they cannot verify.
  • Organisations that want the literature on a problem read by someone who will tell them the honest answer.

Not a fit if

  • Work that needs a literature review to reach a conclusion already agreed.
  • Requests for a citation that supports a decision you have already made.

See it

Three from the deskLive entries from the research desk, newest first. Every one is a real paper with real authors, a DOI or arXiv id, and our own note on what it changes.

2023

Lost in the Middle: How Language Models Use Long Contexts

The single most useful result for anyone shipping retrieval. A bigger context window does not fix relevance: if the right chunk lands in the middle of twenty, it may as well not be there. Re-rank and put the best passage last.

A reference build

The research desk

The desk is not a content marketing exercise. It is the reference layer we use when a client asks why we are choosing one approach over another, and it is public so the answer can be checked.

  • Every entry credited to its real authors, ours marked separately
  • BibTeX on every page so a claim can be cited properly
  • Annotated with what it changes in production, not a summary of the abstract

This is a system we run, not a client case study. We do not publish client numbers we cannot show you the workings for.

Talk to someone who builds this

No qualification call, no discovery deck. Describe the work and we will tell you whether this is the right line for it.

Start a conversationlegal@ejentic.xyz

Other lines

AI agentsAutomationsConsultation