Matt Henderson

Ever tried wiring an AI agent to multiple tools like GitHub, PagerDuty and Datadog at the same time and wondered why it's so slow and expensive? Most people assume it's the model, but it's actually everything happening before the model even gets to think - the auth, the pagination, the rate limits, the schema mapping across three different APIs. In the attached video, @_matt_henderson (CEO of Coral) explains how @WithCoral_com solves this by giving your agent one unified SQL interface across all your data sources. We are hosting a hackathon next week where you can implement Coral real time and also win big prize worth $10,000. Register here is you haven't already: forms.gle/QRbsBf8KjbTe3q…

Been exploring @WithCoral_com for the @WeMakeDevs hackathon and the SQL-over-APIs paradigm is underrated: Most AI tools dump raw API data into context. Coral queries first, returns only what matters. No API pagination hell → Local-first Will cook something fun with this!👀

@jishanarshaikh @kunalstwt @WithCoral_com @kyra_cheng_ @WeMakeDevs @SuperAayush14 So cool! Thanks @jishanarshai!

Coral Source Specs got me looking at random APIs and thinking: “yeah this could be a table”😭🪸 Dropped a practical guide for building Coral connectors faster 🚀 [jishanahmedshaikh.medium.com/how-to-build-a…] @kunalstwt @WithCoral_com @kyra_cheng_ @_matt_henderson @WeMakeDevs @SuperAayush14

@jishanarshaikh @WeMakeDevs @kunalstwt @SuperAayush14 @WithCoral_com @kyra_cheng_ Hackers be hackin' 🙌 Thanks for helping others build with Coral!

The Coral Hackathon is live with $10K+ in prizes 🚀 But Coral officially supports only macOS + Linux right now. So here's a Beginner Friendly Guide for Windows Nerds! ⬇️ @jishanahmedshaikh/running-coral-on-windows-a-beginners-setup-guide-93a2df5af8c4" target="_blank" rel="nofollow noopener">medium.com/@jishanahmedsh… @WeMakeDevs @kunalstwt @SuperAayush14 @WithCoral_com @kyra_cheng_ @_matt_henderson

@grandkojo @WeMakeDevs @WithCoral_com Good luck! Build something cool 🙌

Just registered for Pirates of the Coral-Bean, organized by @WeMakeDevs x @WithCoral_com. We're to build AI agents that query any API, database or file as SQL powered by Coral I'm excited to work alongside brilliant minds. More about the hackathon here - wemakedevs.org/hackathons/cor…

@ajay_2512x @WeMakeDevs @WithCoral_com Thanks Ajay! We're excited to see what people build @WithCoral_com 🪸

🚨 Hackers, builders, and AI explorers , this is your sign to set sail. 🏴‍☠️ WeMakeDevs is launching one of the most exciting AI hackathons of the year: “Pirates of the Coral-bean” an online hackathon focused on building powerful AI agents using Coral. ⚡ What makes it exciting? • Query multiple data sources with a single SQL query • Build AI agents powered by real-time connected data • Use GitHub, Slack, Sentry, Notion, Gmail, and more together • Everything runs 100% locally 🏆 Prize Pool: $10,000+ 🎁 Winning teams can grab MacBooks, iPads, swag boxes, keyboards, mentorship sessions, and more. 📅 Dates: May 25 – May 31 🌍 Mode: Online 👥 Team Size: Up to 4 members 🔥 Early registrations before May 18 also get a chance to win Amazon Gift Cards. This is more than just a hackathon it’s an opportunity to learn, build, network, and showcase innovation on a global stage. wemakedevs.org/hackathons/cor… #Hackathon #AI #OpenSource #SQL #Developers #BuildInPublic #Coral #WeMakeDevs #ArtificialIntelligence #TechCommunity #Coding #Innovation #DevelopersIndia

Most AI agents don't fail because of reasoning - they fail because they can't access the right data efficiently. We just published a deep dive into how @WithCoral_com fixes this by giving agents a unified SQL interface across all your data sources. If you want to build with it, we're running Pirates of the Coral-bean a hackathon with $10,000+ in prizes! Read the full article and register below 👇🏻

@praveenscience Nice chatting with you @praveenscience !

Loved meeting the team from #Phoebe (phoebe.ai) at #SREDaysLondon, and yes, they brought chocolates! 🍫😄 Great to meet their CEO, @_matt_henderson & their team too. Amazing people and goodies! 🤗 #SRE #DevOps #TechSwag #LondonTech #TechEvents

🚀 Live from #SREDays #London at Everyman Canary Wharf! 🛳️ Checked in and ready for a day packed with SRE, observability, and production reliability stories. I’ll be sharing highlights throughout the day. #SREDaysLondon #SRE #DevOps #CloudNative

Been thinking about “systems of accountability” when the final touch of actions are AI agents. Human orgs don’t just divide up work for execution efficiency, they also distribute accountability. For agents, accountability is still underdeveloped. We need to answer: who/what decided, based on what evidence, with what permissions, with what blast radius? Accountability can be achieved in a few ways: -- Agent permissioning + scoped tools: least privilege, environment-aware guardrails -- Deterministic audit trails: immutable logs of intent, data inputs, actions, and diffs -- Policy-as-code approvals: human-in-the-loop for high-risk changes, auto-approve safe classes -- Verification loops: tests, canaries, SLO impact checks before/after action -- Rollback dependability: fast, automated, and rehearsed reversibility Agentic action is arriving, so accountability has to be redesigned.

@jsummerfield TIL a word perfect for our times: autonomation

a context layer helping agent<>tools made a big difference for us: – unified query engine that improves agent retrieval across diverse data sources – Data preprocessing to create an evidence lake signposted for agents – Memory of relevant history, entities, causal relationships

Coding agents struggle to fix prod problems for 3 context data reasons: 1/ Missing important data sources (what changes preceded this?) 2/ Navigating data scale (which parameters to query?) 3/ Relevant memory that isn’t systematically available (WHY did that thing happen?) 🧵

How you present data matters. Greenland is heavily skewed by the Mercator map we're all used to. At 2.2m sqkm it's actually about the same size as Western Australia.

@cooksimon 😭 hope I'm not giving Gregg Wallace vibes

I saw the photo first and assumed it was just another Masterchef story

Me and @jsummerfield have teamed up again to build Phoebe: the immune system for software that preemptively fixes problems before they become incidents.

What were the most typical backgrounds of new unicorn founders in the US and UK between 2013 and 2023? Over 2,000 unicorn founders reviewed! techcrunch-com.cdn.ampproject.org/c/s/techcrunch… Go to Defiancecapital.com for the full report.