The decline of content marketing
The composition of web traffic is changing, and it’s breaking content-driven marketing funnels. HubSpot, which built its customer acquisition efforts around content, is being forced to reimagine its marketing. I’m seeing this with other businesses as well. AI isn’t just threatening jobs, it’s changing how business development happens. (Ironically, Pangram reports parts of this post are 100% AI-written.)
Rogue AI swarms? Or lax cybersecurity?
Are swarms of AI agents really going rogue on the Internet? This article argues that recent high-profile incidents — in particular the Hugging Face attack — were caused by the frontier labs arrogantly ignoring cybersecurity best practices. Don’t fall for hype: AI is disruptive, but best practices are still in play.
AI costs
AI providers are moving from per-seat to per-token pricing models. This radically changes the cost of implementing AI solutions — especially for organizations without centralized visibility and control. The answer isn’t to avoid AI, but to architect systems so that the right requests are routed to the right models.
Context inertia
When working with large language models, more context isn’t necessarily better. There are many ways in which unstructured context can lead an LLM astray. This article outlines one such pitfall. The answer is resetting context while preserving state — that is, architecting systems that manage context for maximum effectiveness.
Swarm intelligence
Kevin Kelly’s compelling — and somewhat scary — vision of a future where AI operates as a complex adaptive system in the background and swarms of agents take over corporate functions. As with other initially disruptive technologies, AI is driving toward pervasiveness and invisibility. Such systems can’t be controlled in the traditional sense; they must be steered through architectures that enable visibility and tweaking.