invaria (YC S24)

Attention lovers of structure logs (and the rest of you): we are now invaria (YC S24). To celebrate, we posted an update on Show HN. invaria.io Patchwork is a frickin' great name, obviously. Unfortunately several other tech founders thought the same thing in the last year or two, so it's time to carve out a new brand for the future of structured telemetry.

Senior devs wanted to turn off INFO logs in prod. Cue my next (Ted?!) Talk on modern observability :D - Structure your events for queryability - High cardinality is your friend, not enemy - Debugging without context is a nightmare Aaaand its 2024, there is no need to optimize during instrumentation. Use smart "control plane" features to filter and manage! Last9

Congratulations to all Oxford-affiliated ventures presenting at Y Combinator's demo day this week and graduating from #S24! These startups all have a founder that was either a matriculated or visiting student at the University of Oxford - a record number, likely thanks to Tom Blomfield's visit this year to Sa?d Business School, University of Oxford and efforts to engage with the UK startup ecosystem. We wish all of the ventures the very best with fundraising! ?? Arva AI (YC S24) - AI agents for instant business verification ?? Blast - Helping enterprises build safe and compliant LLM apps/agents ? Conductor Quantum - Quantum computers on silicon chips using AI ?? Dodo (YC S24) - AI employees for specialty clinics ?? ēvolvere Biosciences - Next-gen antibiotics outpacing bacterial evolution ? Fortress (YC S24) - Tenant isolation simplified for SaaS ? Ionworks - Simulation software for battery & automotive companies ? Ligo Biosciences - AI designed enzymes for the chemical industry ?? Olive Legal - AI medical summaries for injury lawyers ? Overstand Labs - Cross-selling engine for enterprise SaaS companies ?? Patchwork Technologies (YC S24) - Next-gen logging & observability ?? Planbase (YC S24) - Workforce management for modern healthcare ?? PLUME - Home energy renovation co-pilot ?? Redouble AI - Scale human-in-the-loop for AI in regulated industries ? Remo (YC S24) - AI for compliance teams at banks ?? Simple AI - A better, AI-powered Siri that just works ?? Simplifine (YC S24) - AI-powered workspace for research ? Storia AI - Open source AI that knows your company's code & context ?? Tabular - AI co-pilot for accounting firms, starting in Europe ???? Taxo (YC S24) - Automated medical billing and coding ?? Zeit AI (YC S24) - Insights from spreadsheets with just a few words

What sucks about observability is that buying the tools is the easy part. You want whatever you buy to just work but like buying a hammer won’t make you a carpenter, there is a lot to learn before you’ll get value from your purchase. Often, and something I see a lot, is companies making a push to rollout observability across the org. Usually after a series of incidents that made executives say ‘something’ must be done, where that eventually lands as ‘add some metrics’ Great! You add an o11y project to each teams roadmap, approve budget for the new toolchain, then you’re done. Except… you’re not. Teams get to their projects, often with varying levels of skepticism about the value, and proceed to run headfirst into the blank page problem. “So many tools, so much code. Could do anything, really. Add metrics here, some tracing there, I gues that could do with some telemetry…” It can be a huge waste of time. Worse, it might generate loads of telemetry data that skyrockets your bill, all of it unused. No value delivered, engineers annoyed, incidents continue. What you need is a strategy, a good understanding of the problems you want to solve, and a vision of where you want to go. You need patterns that people can apply consistently to add observability to systems, and an example of ‘good’ that people can aim toward. The internet is short of content about exactly how teams rollout observability and what the end result looks like. Which is why I’m super happy Martha has written a post that shares exactly how we did this at incident, including real screenshots and videos of our dashboards and our philosophy for creating them. These are patterns that can be adopted by any team with similar aims. If you want to get past the blank page, I strongly recommend you read this post!

Patchwork Technologies (YC S24) automates logging from code to storage. Observability is broken, and fixing it means tackling it at the source — your code. Adding logging to code is currently a time-consuming manual task. Unlike building features, it doesn’t earn promotions, so it’s often neglected. It’s at 2 am, when your team is diving through your application logs that the value of actionable, context-rich structured logs hits home. Patchwork works in the background, reviewing your existing codebase and stepping in during continuous integration. At all times, it ensures that your logging code meets a rigorous standard. With Patchwork, developers can focus on building features while logging is taken care of, from the implementation within your code, through to storage. Co-founders Benjamin H., Samuel Kramer, and Alex Maass met while working at Palantir Technologies. They learned the challenges of delivering reliable software firsthand, and want to make it easier to debug incidents. Congrats to the team on the launch! ?? https://lnkd.in/gFDEFrDc

If you went to therapy for engineers, the first topic would be: how do you feel about how much you're paying for your observability? ?? ICYMI - here's Charity Majors on the Orb podcast talking about Observability 1.0 vs. 2.0. It's such a good point - today, observability relies on engineers to *insert* a lot of context into triage by hand-correlating graphs rather than the system doing the work for you.