We scored 50k PRs with AI
I'm a CTO with a ~16-person engineering team. Last year I wanted real data on what was actually shipping, not guesswork or story point theater. So we built GitVelocity. Every merged PR gets scored 0–100 by Claude across six dimensions: scope (0–20), architecture (0–20), implementation (0–20), risk (0–20), quality (0–15), perf/security (0–5). Six dimensions added up, then scaled by change size — a 10-line fix scores lower than a 500-line refactor even at the same complexity. Full formula at gitvelocity.dev/scoring-guide. After scoring 50,000+ PRs across TypeScript, Python, Rust, Go, Java, Elixir, and more, some things surprised us: Big PRs don't automatically score high. An 800-line migration with low complexity scores worse than a 200-line architectural change. Size gets you the full multiplier, but the base score still has to earn it. You can't score well without tests. The quality dimension (0–15) won't give you points without test coverage. At similar experience levels, this was the clearest separator between engineers. Juniors started outscoring some seniors. They adopted AI tools faster and took on harder problems. Once they could see their own scores, they aimed higher. We score AI-generated code the same as human-written code. Code is code. An engineer who uses AI to ship more complex work faster is more productive, and their scores reflect that. Scoring consistency was the hardest technical problem. Without reference examples anchoring each dimension, Claude's scores drifted 15+ points between runs. With 18 calibrated anchors (three per dimension at low/mid/high), we got it down to 2–4 points on the same PR. The thing we didn't expect was behavioral. We call it the Fitbit effect — the tool doesn't make you ship better code, but seeing the score does. Engineers started referencing their own scores in 1:1s unprompted, because the numbers matched what they already felt about their work. A junior who shipped a tricky concurrency fix could point to a score that proved it wasn't "just a small PR." We recently added team benchmarks (gitvelocity.dev/demo/benchmarks). Once you're scoring PRs, you can see how your team compares to others across the dataset — about 1,000 engineers on 60 teams so far. Headline's team ships faster than roughly 95% of them, which was nice to confirm but also made us wonder who the other 5% are. The competitive angle surprised us: teams that were skeptical about individual scores got genuinely curious once they could measure themselves against the field. Every score is fully visible to the engineer who wrote the PR, with per-dimension breakdowns and reasoning. There's no hidden dashboard that management sees and engineers don't. Free, BYOK (your Anthropic API key). We default to Sonnet 4.6, which scores nearly as well as Opus 4.6 at a fraction of the cost — but you can switch models if you want. Pennies per PR either way. No source code stored, diffs analyzed and discarded. Works with GitHub, GitLab, and Bitbucket. Ask me anything about the scoring methodology, how we solved calibration, or what it was actually like rolling this out to a team.
AI Analysis
Analysis coming soon.
Similar Products
Capgo
Instant updates for Capacitor apps. Ship fixes in minutes, not weeks. Push OTA updates to users without app store delays.
OpenAlternative
Open source alternatives to popular software. Over 1 million users replaced their proprietary tools with open source software. Discover the best alternatives and join the movement.
SubSmith
I've been learning Japanese for a few years and kept running into a similar problem. I'd find a video I wanted to learn from, hear a useful sentence, and then realise that turning that sentence into something I could study later was both time consuming and draining at times. I would end up jumping between a video player, subtitles/transcription, a dictionary, screenshots, audio clips and Anki. So I built SubSmith to bring that workflow together. You can drop a video or audio file into it, generate a transcript locally and then use the transcript alongside the media to: * look up words and sentences * replay individual lines * edit the transcript * save useful sentences with their original context/audio * export them as Anki cards The important part for me is that it works with your own media. It isn't tied to a particular streaming service or library, so I can use the random anime episode, podcast, lecture, etc. that I'm actually interested in studying. It's an offline-first desktop app, and transcription happens locally rather than sending the media to a transcription API. I'm sharing it here because I'm now more interested in finding out where this workflow breaks down for other people rather than adding features randomly now that I have solid core/base. For example: * Would you actually save sentences from your own media? * Which part of this process feels like too much work? * Does having the audio/context attached make creating an Anki card more useful? * Would you prefer this to work inside your existing video player/browser? * Is installing a desktop app a significant barrier? * And does requiring an account before starting the free trial make you give up? The current version does require an account to start the trial, and I'm trying to work out whether that's meaningful friction for the people who would actually use this. It's free to try, and I'd particularly appreciate feedback from people who already learn languages through their own videos, anime, films, podcasts or other media. I'm the developer, so I'll be around in the comments to answer questions and discuss how it works. https://subsmith.app
Ancestree
Every time I talk to my older family members, especially my grandparents, I find out a new super interesting fact about them. Last week: my Croatian grandfather served in the French marine corps... Not worth mentioning I guess. I realized that these stories are passed down only by re-telling them. Just think about how much you don't know about your family from 2 generations ago. For this reason, I created ancestree.marindedic.com It's a really cool family tree creation app, where each person gets their own book. Inside it, write what they did, what happened to them, what they were like... whatever you want. Little biographies of your loved ones. Apart from exporting normal and detailed versions of your family tree, you can even export someone's chapters as one long biography. Of course, open-sourced, no account, no server. Nothing you draw or write ever leaves your browser.
Keenable
Hey HN! We built https://keenable.ai, a different web search API for AI agents. Keenable searches our own 100B+ page index. We are focused on low cost and latency (p95 <250ms from us-east). We don’t believe in benchmaxxing, so we open-sourced our internal benchmarking suite, NEEDLE (available at https://keenableai.github.io/needle): a live benchmark that compares Keenable with other search APIs on fresh agent-like queries. I spent seven years at Amazon as a scientist working on web grounding for Alexa/AGI, and my co-founder Andrey previously led search at Yandex. We started Keenable because agents search differently from humans, and we wanted to build around those patterns directly. The API is available now and we provide a free allowance of 100,000 requests a month. It also exposes a novel SQL-like interface to the web, which is useful for structured extraction and agent workflows. Happy to answer questions about the index, crawl, ranking, latency, or benchmarking.