I have a growing conviction that 2026 will be the year of multi modal AI. There are a handful of trends all coming together at the same time that are set to converge. Multi-modal models get good enough. Inference is getting cheaper and faster (cost curve is important). And the real world starts showing up as first class input. I really believe AI will stop predominantly living in text boxes and instead in places humans actually are. For the last few years, AI has been overwhelmingly text first, and for good reason. Text was the fastest path to usefulness. It was easy to collect, easy to tokenize, relatively cheap to serve, and generally didn’t have the same latency requirements. If you were building an AI product in 2023 or 2024, starting with text was the rational choice. But text always

As always, these posts are more of a brain dump of “what I’m thinking” about…And lately I have been thinking a lot about “legacy SaaS",” systems of record, etc. I wanted to write another post today in a similar vein.When I think about how people and companies interact with software today (and for this post when I say software I’m talking about legacy SaaS systems of record) the pattern is generally pretty simple. The system of record is a single, organized place where a human goes to look something up, understand the state of the world, and then take some sort of action based on the information they gathered from the system of record. Like opening Salesforce to check pipeline before updating a forecast, or pulling up NetSuite to reconcile numbers before approving a close. And more often th

I have been thinking a lot lately about education in AI. More often than not, I am seeing education as a core go to market problem (but sometimes an advantage) in AI. The companies that are best able to educate the world and their prospects are winning. Part of the reason for this is that everything in enterprise AI is still so new. Many people, teams, and organizations are facing what I think of as a blank canvas problem. They know AI is powerful, but they don’t even know where to start. Where to use it, where to augment existing workflows, where to create net new workflows, etc.What surprises me the most - even when teams think they know what functionality they want build, it’s not obvious to them how to build it. Build versus buy is just where it starts. Even inside “build,” there are d

Every AI conversation eventually drifts toward the same set of questions. Which model should we use. Build or buy. How much will inference cost. What happens to accuracy at scale. These are all real questions, but they are also strangely orthogonal to whether AI actually changes anything meaningful inside a company. You can answer every one of them well and still end up with an AI strategy that feels impressive in demos and immaterial in practice.There is a more important decision sitting underneath all of this, and most companies are not talking about it directly. They are making it implicitly, often without realizing it. That decision is whether AI is allowed to be authoritative or merely assistive.Most enterprise AI today is firmly in the assistive bucket. AI drafts the email, but a hum

Cloud Giants Update:AWS ( $Amazon.com(AMZN)$ ): $132B run rate growing 20% YoY (last Q grew 17%)Azure ( $Microsoft(MSFT)$ ): ~$93B run rate (estimate) growing 39% YoY (last Q grew 39%)Google Cloud ( $Alphabet(GOOG)$ $Alphabet(GOOGL)$ includes GSuite): $61B run rate growing 34% YoY (last Q grew 32%, neither are cc)ImageFor SG users only, a tool to boost your purchasing power and trading ideas with a Cash Boost Account!Welcome to open a CBA today and enjoy access to a trading limit of up to SGD 20,000 with upcoming 0-commission, unlimited trading on SG, HK, and US stocks, as well as ETFs. Find out more here.Other helpful