Rented Competence
Why can't we just use this? I've heard that question more in the last six months than in the previous ten years. In the era of GenAI, it's a fair
Why can't we just use this? I've heard that question more in the last six months than in the previous ten years. In the era of GenAI, it's a fair question. The AI-built site or app or chatbot works. It was demoed to a client in a pitch and everyone
AI's effect on the software ecosystem
If your day job isn't focused on software or application development, you're aware that AI has had a dramatic impact, but you might not realize the extent of its ripple
If your day job isn't focused on software or application development, you're aware that AI has had a dramatic impact, but you might not realize the extent of its ripple effect on the ecosystem. It's not just code production that's transforming; it's the tools themselves, too. Tools, like the source
Infinite Toast LABs
Working on a new game in the 'Infinite Toast' series. I was going to post a WIP game screen, but with all the REA going on, I'm only posting the
Working on a new game in the 'Infinite Toast' series. I was going to post a WIP game screen, but with all the REA going on, I'm only posting the icon for my LAB game research app. Stay tuned.
You need a data plan
If you’re focusing your resources on building your agentic layer and your plan is to just connect it to your existing data as-is, you’re doing it wrong. Does your agent
If you’re focusing your resources on building your agentic layer and your plan is to just connect it to your existing data as-is, you’re doing it wrong. Does your agent know where the data is and how to access it with the right permissions? Your data is sitting across multiple
AI Trends to Watch
As your team builds your proprietary agentic runtime, two trends to watch: 1. Model macro-ization and micro-ization: Google DeepMind just released Genie 4, a world model capable of generating photorealistic,
As your team builds your proprietary agentic runtime, two trends to watch: 1. Model macro-ization and micro-ization: Google DeepMind just released Genie 4, a world model capable of generating photorealistic, interactive environments with native audio and conversational characters for games and robotics. It's a shift toward using world models for
AI voice innovation: Dessa
I'd usually wait until a weekday to post, but I really think you should hear the latest in AI voice innovation, Dessa, from Block: https://block.xyz/dessa
I'd usually wait until a weekday to post, but I really think you should hear the latest in AI voice innovation, Dessa, from Block: https://block.xyz/dessa
Claude Duet joins the autonomous agent space
Anthropic has launched Claude Duet, its standalone, autonomous agent for iOS. Duet joins the growing persistent agent space, proactively monitoring apps, prioritizing tasks, and integrating with commerce services like Stripe
Anthropic has launched Claude Duet, its standalone, autonomous agent for iOS. Duet joins the growing persistent agent space, proactively monitoring apps, prioritizing tasks, and integrating with commerce services like Stripe to execute purchases on behalf of its user. This move aligns Anthropic to the broader AI ecosystem trajectory, especially enterprise-targeted
RSI coming to a harness near you…
Is your team still building a proprietary agentic harness or runtime? Does it include RSI? If not, add it to your ever-growing backlog. Google Research just open-sourced its 'Harness2' recursive
Is your team still building a proprietary agentic harness or runtime? Does it include RSI? If not, add it to your ever-growing backlog. Google Research just open-sourced its 'Harness2' recursive self-improving agent harness, enabling agents to encounter previously unseen tasks, propose, probe, refine, and improve itself, all while keeping model
System One Extraction Models
Heads-up: Another new System One model, this time from Cloudflare. As announced on their blog, Clairic is a System One extraction model, built to extract values directly from unstructured inputs.
Heads-up: Another new System One model, this time from Cloudflare. As announced on their blog, Clairic is a System One extraction model, built to extract values directly from unstructured inputs. Check it out: https://lnkd.in/gdCVfYTJ
Perplexity on your Mac?
Perplexity Computer can now run fully local on your Mac. Work with your native Mac apps and local files, and run tasks without sending sensitive data to the cloud. The
Perplexity Computer can now run fully local on your Mac. Work with your native Mac apps and local files, and run tasks without sending sensitive data to the cloud. The model and the agent both run on your Mac. Local tasks use PPLX Pro, our post-trained version of Qwen 3.8,
AI Infrastructure Consolidation
More signs that the AI infrastructure sector's consolidation of independent neocloud providers are starting to crop up. One example: CoreWeave's bonds entered distressed territory last week, triggering a stock decline,
More signs that the AI infrastructure sector's consolidation of independent neocloud providers are starting to crop up. One example: CoreWeave's bonds entered distressed territory last week, triggering a stock decline, and raising concerns about whether these neocloud buildouts represent sustainable investments.
DHS - DeepSeek
Heads-up: DHS is ordering the removal of DeepSeek products and third-party services from civilian agency information technology citing multiple threat vectors to systems. Read the order, here: https://lnkd.in/gvtH5TvG
Heads-up: DHS is ordering the removal of DeepSeek products and third-party services from civilian agency information technology citing multiple threat vectors to systems. Read the order, here: https://lnkd.in/gvtH5TvG
From tokenmaxxing to self-replication
We had tokenmaxxing, then the tokenpocalypse, and then the token famine. This drove price wars, interest and competition in open source/weight models, new model types (e.g, decision, CDM, etc.), and
We had tokenmaxxing, then the tokenpocalypse, and then the token famine. This drove price wars, interest and competition in open source/weight models, new model types (e.g, decision, CDM, etc.), and the pursuit of locally-runnable models. As innovation, including LLM gateways like OpenRouter, continues to drive capability, capacity, and cost efficiency
Pay attention to the AI company shakeout
Pay attention: as consumer-play AI companies like Character.ai (who filed for bankruptcy after ~5 years) are winnowed out, watch for AI startups to pivot toward the enterprise, or abandon their
Pay attention: as consumer-play AI companies like Character.ai (who filed for bankruptcy after ~5 years) are winnowed out, watch for AI startups to pivot toward the enterprise, or abandon their own stacks in favor of companies chasing the enterprise like
Another new model architecture just dropped…
Another new model architecture just dropped. According to ScaleDown, Constrained Decoding Models (CDM) are task-specific, purpose-built language models, each trained to do one job: compression, summarization, extraction, and classification. Check
Another new model architecture just dropped. According to ScaleDown, Constrained Decoding Models (CDM) are task-specific, purpose-built language models, each trained to do one job: compression, summarization, extraction, and classification. Check out the post from founder Neal Patel: https://lnkd.in/gbDE-yWn
You Better Move Quickly…
If you want the agentic runtime your team is developing to stay at parity with those offered by AI companies, you'll need to move quickly. Always-on, user-dedicated autonomous agents (Spark,
If you want the agentic runtime your team is developing to stay at parity with those offered by AI companies, you'll need to move quickly. Always-on, user-dedicated autonomous agents (Spark, Muse, Dots, Grok Bot, etc.), dynamic work analysis with model autorouting (monomodal, multimodal, omnimodal; decisioning, etc.), cost management (frontier, flagship,
Your Agentic Runtime is Already Antiquated…
The enterprise agentic runtime your team is building was likely built on a model of how people work in the enterprise: handling individual tasks. That evolved to serial chained tasks,
The enterprise agentic runtime your team is building was likely built on a model of how people work in the enterprise: handling individual tasks. That evolved to serial chained tasks, and then, maybe, to semi-autonomous workflows where the agent decides what to do based on the input and training of
Hello World!
Hello? Is this thing on?
Hello? Is this thing on?