Martin

Every so often Silicon Valley devs re-discover that the vast majority of programmers use Windows and are re-surprised.

"[ #EventSoucing ] is intuitive. If you register everything that happens, figuring out what should be the current state of the system is accounting, the other way around, is _forensics_." - @JYCabello This encapsulates why we can offer free bug fixes and others can't.

Wow, diffusion models (used in AI image generation) are also game engines - a type of world simulation. By predicting the next frame of the classic shooter DOOM, you get a playable game at 20 fps without any underlying real game engine. This video is from the diffusion model.

Snowflake may soon find itself in the uncanny valley of database price-to-performance, serving neither the cold nor hot tiers of data well. To wit: * Snowflake isn't cheap enough to compete with @databricks for "cold tier" data that belongs in a data lake, for slow reporting / batch processing use cases (the @tabulario acquisition further positions Databricks as winning here), and yet... * Snowflake isn't fast enough to compete with databases like @ClickHouseDB for the "hot tier" of data that powers user-facing applications / real-time ML systems and requires high performance. Snowflake is serving the "lukewarm" data tier, and unless they make some strategic moves, I predict more and more companies will decide to take their data elsewhere.

Future of work prediction: People will work.

We at @FoundationCap believe there is $4.6T of work to be automated. AI companies are leading a transition from Software-as-a-Service to Service-as-Software, turning the table on the very essence of SaaS. We look at the areas to be automated in two buckets: 1.) Salaries of jobs globally ($2.3 trillion in sales & marketing, software engineering, security, and HR) 2.) The amount spent on outsourced services and salaries—both IT services and business process services ($2.3 trillion, per Gartner) In the software business, a company may sell access to its platform or tool, but customers are still responsible for using that tool to achieve the desired outcome. In the services business, responsibility for achieving the desired outcome sits with the company selling the service.

Interactive visualization of chatgpt completion probabilities, I ask it to to complete the prompt "When " hundreds of times.

Maybe it’s time to start an open working group with @tldraw, @obsdmd, @excalidraw, @MuseAppHQ, +others to develop a minimal, open and interoperable format so canvases can approach the longevity of text files. I’d be happy to organise this. Done it before. And I’m very invested.

I worked on Google's M&A team when we were doing 40+ acquisitions a year 21 things founders should know about getting acquired 1. your team will likely have to pass interviews at the new company, so hire well. 2. every time your valuation increases, the number of potential acquirers decreases. 3. deals that come in through the corp dev team have a <1% success rate, so talk to actual product people. 4. build relationships with product teams years in advance of a potential acquisition. 5. your largest customers and partners are the best potential acquirers. 6. M&A is a FOMO game, so it's good to play acquirers off each other (when you are legally able to). 7. the deal isn't over until the $$ is in the bank (lots of deals dying very far along these days). 8. don't let your team know you're running a process until the very end. they'll leak the news or become wildly unproductive when they find out and even more unproductive when the deal falls through. 9. the best time to get acquired is when you don't need to or want to be. 10. when you talk to acquirers, you need to show them how you will supercharge their product/business – this can involve actual design and code. 11. full acquisition prices you see in the headlines often come with strings attached – usually integration, revenue/user growth, and employee retention milestones. 12. decide with your cofounder what the conditions would be (e.g., $$) for you to accept an acquisition offer before you have your first conversation. cofounder misalignment here can really really hurt. 13. running a fundraiser at the same time as an M&A process can help both – buyers can offer more if your fundraising options are good and investors love to see real exit opportunities. 14. there are 3 types of deals: acquihire (just the team), asset (just the tech), and full (entire company, team & assets). 15. acquihires can range from just getting a normal job (and the ability to say you were "acquired") to also including $10M+ payouts. don't trust folks bragging about getting acquihired because it's usually the former. 16. your liquidation preferences will largely determine your financial outcome. your investors get their money back before you get your payout from an acquisition, so if you've raised a lot (on bad terms) and you don't have traction, you're probably not going to make much at all. 17. some acquirers will hold back equity you've already vested as part of the deal. some will accelerate unvested equity AND throw in retention bonuses. 18. falling into depression after an acquisition is not uncommon. take care of your mental health and make friends who have gone through it. 19. make sure you're clean legally and financially. investing in good bookkeeping early on can save you tons of time and prevent your deal to get derailed. 20. acquisitions have complex personal tax consequences. hire someone good for that. 21. the best acquired founders at Google only stayed an average of 2.5 years before leaving to start their next thing. life isn't over after you sell!!

We’re still in the “let’s retrofit AI into existing UIs” phase. The future is AI-native UX.

Very excited to see where this goes engineering.fb.com/2022/11/15/ope…

Here's a consolidated thread of all new GitHub features that are relevant to ML/Data people that were shipped recently 1. Notebook Diffs twitter.com/HamelHusain/st… 1/8

Today, we're excited to launch @github code search - a new way to search and navigate code. We’re introducing a brand new search and code navigation view and they are jam packed with new features. Here are some I’m excited about.... 🧵 1/6

My new design policy

🧵Here are some quick observations on what I’m seeing within distributed systems in 2022 -- largely focused on edge computing, Rust/Wasm, and databases

Data-centric AI requires keeping data secure. Today we're proud to announce that Continual is SOC 2 Type 2 compliant with help from @DrataHQ. hubs.ly/Q01jdZlz0

Excited to announce our investment in @continual_ai! The rise of the modern data stack has laid the foundation for a new massive company to be built in the MLOps space, and we think @tristanzajonc and Tyler Kohn are the right team to go after it. medium.com/innovationende…

I'm excited to announce that @continual_ai is now generally available and we've raised a $14M Series A to bring AI to the modern data stack. Here's what that means. continual.ai/post/announcin…

Exciting news! The stealthy startup that @sriramsubram and I started in Jan is looking for a senior backend/infra eng with expertise in building API products. If this is you, and you want to join us in the exciting "under 10 people" stage of a startup, DM me.

@schmidtdisturbr @unbounce What a journey! 🙌 Looking forward to what you do next 🚀

Today is my last day as an @Unbounce employee. It's been an amazing 13 years, and I count myself lucky to have come to know so many amazing people (employees and customers!) through this journey. Not sure what's next for me yet, but excited to explore!