Blog

A markdown notes file plus a vector index works for lookups, but fails on real business tasks that span entities, time, systems and decisions. Why retrieval is an attention problem, and when a context graph is worth it.

AI coding agents let one senior engineer do the work of a small team for several companies at once. What a fractional engineering firm looks like, why SMB owners hesitate, and what it does to the IT job market.

An AI model settled open questions across seventeen areas of mathematics by reading fields that specialists rarely read together. The same idea, with outcomes as the feedback loop, works on a B2B sales pipeline.

Websites, search boxes and checkout pages are the part of the internet we see, and personal agents are about to remove them. What that means for consumers, SMBs and the still-open enterprise question.

Knowledge-representation researchers split knowledge into a rulebook and a diary forty years ago. Keeping the T-box and the A-box apart is what turns an agent with a long context window into one with a real memory.

The processes that produce revenue are reasoning habits, not workflows. Borrowing Wolfram's computational reducibility to find the pockets inside them — and to let the record assemble itself.

A business world model, the org chart as a routing protocol, the end of manual CRM entry, and the end of hardcoding. Read together, they describe one architecture — and one uncomfortable org change.

Every era of architecture reduced how much of the world we decide in advance. The next step is a tiny constitution humans control — and everything else negotiated by agents.

AI cut the cost of shipping a product by orders of magnitude — for you and for the ten thousand companies next to you. So what's your business model, startup?

The agent isn't the model, the memory, or the tools. It's the rules and context — everything else is replaceable infrastructure.

For decades we built corporate software focused on one thing: keeping a record. The next era is about reasoning over context.

Most modern SaaS platforms are graveyards of disconnected data. A context graph reconnects it.

Notes on the technical side and business application of the context graph concept, and where the open research questions are.

In the industrial era, Ford's genius wasn't the Model T — it was the assembly line. The same shift is happening with AI.

We're approaching a point where AI agents become functionally indistinguishable in quality and capacity. What then?

Your beautiful hero video means nothing to an AI agent. The era of bot engine optimization is here.

Falling markets for legacy SaaS and the daily advance of agents and models — an attempt to structure what's happening.

A thought experiment about the future of B2B sales when your SaaS platform becomes an active coach.

Imagine a sentry, ever-watchful — what it takes to give a SaaS platform genuine memory and reasoning.

For two decades the paradigm was clear. We're now standing on the precipice of a fundamental shift in how we build and consume software.

We've lived in the Golden Age of the App. AI may be about to redefine the interface of the internet itself.

A hands-on test of Google Antigravity — a next-generation IDE combining editor, agent manager, Gemini CLI and a browser sandbox.

The era of the monolithic, one-size-fits-all cloud LLM is facing a quiet but definitive revolution.

A question that touches the evolving architecture of AI systems. The short answer: yes and no.

We've spent two years asking how far behind open source is. With Moonshot AI's Kimi K2.5, the answer may have changed.

Physics has always been a game of finding the source code of reality. AI is changing how we play it.

For years experts argued that scaling alone is a dead end. That debate is effectively over — what comes next?

In every gold rush there's a difference between those selling shovels and those digging. The AI gold rush is no different.

For two decades the holy grail in B2B SaaS was to become the System of Record. That moat is crumbling.

For decades the tax on software engineering has been syntax. The future of coding is about context, not semicolons.

The future of software isn't a smarter text editor — it's a digital assembly line where humans manage and AI executes.

We're stepping over the threshold into a new internet era — and it will get messier before it improves.

If 2024 was the year of the chatbot and 2025 the prototype, 2026 has ushered in agentic industrialization.

A fascinating paradox: as consumer tech becomes more visual, pro-level coding seems to run the opposite way.

In commerce and retail, agents have hit a wall of fragmentation. The Universal Commerce Protocol aims to fix it.

Digital content creation is at an inflection point, moving from human-in-the-loop to machine-to-machine.

Agentic AI is a fundamental challenge to the cloud-native, stateless paradigm that has dominated the last decade.

A follow-up to "Video content creation prosumer platform" on orchestrating generative AI pipelines.

Video content marketing keeps growing — and personalisation is where much of its untapped value sits.

It's still early days for the next generation of prosumer video tools. A look at what they need to become.

If you manage many real estate objects, creating attention-grabbing content at scale is a real problem. Here's an approach.

If your product data lives in e-commerce or PIM platforms, you can automate product video creation.

The past decade was driven by AI tooling and finding applications for it. AVICA applies that to video content.

Is digital product content management done right in large enterprises? An attempt to answer.

Large BSS/OSS vendors offer enterprise product catalog solutions — but PIM for telecom has its own challenges.

Idea and business case for AVICA: automated video content creation, boosted by growing demand across scenarios.

The Metaverse as a metaphor for extending the internet as a computing platform through a synergy of technologies.