Anton the Search Relevance Evaluator
Hey there HN, I'm Pablo from Objective. We're working on an AI-native Search platform. Today we're releasing Anton (https://www.objective.inc/anton) an API that you can use to evaluate relevance for any search system. We would love your feedback. Here's a quick demo by my co-founder Lance: https://youtu.be/geX8mwBGqIo Why we created Anton: In every company we've worked, it was very slow and costly to spot check a new model, or to thoroughly evaluate a release candidate, or to compare a few competing approaches for how to improve search going forward. In each of…
In plain words
Anton is an API for evaluating search relevance across any search system. Built by Objective, it helps teams quickly assess new models, compare different approaches, and test release candidates without requiring expertise in machine learning or information retrieval. The tool addresses the traditionally slow and expensive process of spot-checking search quality and enables software engineers to improve search results more efficiently.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hey there HN, I'm Pablo from Objective. We're working on an AI-native Search platform. Today we're releasing Anton (https://www.objective.inc/anton) an API that you can use to evaluate relevance for any search system. We would love your feedback. Here's a quick demo by my co-founder Lance: https://youtu.be/geX8mwBGqIo Why we created Anton: In every company we've worked, it was very slow and costly to spot check a new model, or to thoroughly evaluate a release candidate, or to compare a few competing approaches for how to improve search going forward. In each of those cases, we would absolutely have loved to have had Anton! Today, in order to ship search that doesn't disappoint users, software engineers need to become experts in AI/ML, information retrieval, data engineering, and more. Folks in our team have worked for some of the largest companies out there (Apple, Google, Youtube, Amazon, Twitch, etc.), building products that you use every day. We've all had to build and stitch together similar components everywhere, and we're now building Objective to lift some of that burden for you. We want to make it super easy for every developer to build great search experiences into their app/site. To build Anton, we spent a lot of time evaluating different LLMs, prompt engineering, fine-tuning and really throwing the kitchen sink at the LLM-as-a-judge problem. We've discovered many ways in which LLMs make it hard to make systematic progress, and only a few that didn't make us question our sanity. That work culminated in the creation of Anton. We hope this will provide you with a cost-effective way to get near-human quality relevance judgements, so that you can delight your users every day with search that actually understands them. Can't wait to see what you'll build with it. And as always, would love feedback!
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, September 2024
the whole month →
BeforeSunset AI 2.0▲1,267Personalized AI daily planning that suits your life
AI · 2024 · beforesunset.ai
