Democratizing AI creation Build your own AI agent for work, life and play
Joined November 2020
- Tweets925
- Following449
- Followers25.2K
- Likes159
Pinned Tweet
#38 | SoftBank's $50B Malaysia AI bet, SK's mega data center push, Sakana's homegrown model, and more… linkedin.com/pulse/38-softba…
Democratizing AI creation doesn’t mean everyone becomes an engineer. It means the person who knows the problem best can build the solution. The next wave of useful agents will come from people who know their work, not just people who know code.
Agents are moving from experiments to infrastructure. New agent frameworks and sandboxes keep landing, along with open standards for how agents share skills and tools. Open standards mean you don’t have to start from zero.
Practical beats impressive. An agent that reliably handles your weekly report, inbox triage, or research digest is worth more than a demo that goes viral once. What’s the dullest task you’d happily hand off?
The best AI agent you’ll ever use is probably one you built for a problem only you have. A tool made for everyone solves nothing in particular. What would you build if it took an afternoon?
Meta is raising $95B and Amazon says its 2026 cloud capacity is sold out. The infrastructure race is enormous, and it pushes the cost of intelligence down for everyone building on top. The scarce thing is shifting from compute to good ideas.
Moonshot says Kimi K3.1 Pro scores 65.7% on Terminal-Bench 4.0, within a point of the top closed model, with open weights planned for early November. Company-reported, but the direction is clear: frontier-level capability is becoming something anyone can build on.
SK Telecom now gives each AI agent an employee ID, a role, and a human supervisor, and can suspend underperformers. We may end up managing agents the way we manage teammates. Would you rather have a manager for your agents or an agent for your manager?
Anthropic launched Claude Duet, a proactive agent that sends a morning brief, works in the background, and checks with you before acting.
“Checks with you before acting” is the feature to watch. Useful agents earn trust in small steps.
#16 | PROOF OF WORK: You Don't Have to Quit to Level Up linkedin.com/pulse/16-proof-…
The Energy Layer | Issue 17: The Flexibility Market Just Got Teeth linkedin.com/pulse/energy-la…
A security report found 76% of enabled AI agents hadn’t run in 60 days but still held live credentials. Building agents is getting easy. Retiring them is the part nobody plans for. Every agent you create should come with an off switch.
Notion’s agents can now run a single task for 24+ hours, up from five minutes. That is the real shift in agents: from “answer my prompt” to “go work while I sleep.” The builders who learn to scope long-running work well will pull ahead.
#37 | China's $17B data center hinterland, Anthropic's Osaka bet, Singapore tests the agents, and more… linkedin.com/pulse/37-chinas…
Delight retweeted
What does it take to prepare your organization for AI agents?
Our most-liked webinar explores AI agent adoption, early implementation lessons and the technology foundations needed for success.
Watch the on-demand session 👉 gtnr.it/3SbP6ve
The Energy Layer | Issue 16: Who Gets the Megawatts? linkedin.com/pulse/energy-la…
#14 | PROOF OF WORK: Nobody Promoted You to Manager. You're Managing Anyway. linkedin.com/pulse/14-proof-…
#36 | Manus raises $500M after Meta split, CoreWeave lands in India, AirTrunk's $30B Japan bet, MAS sets AI rules for finance, and more… linkedin.com/pulse/36-manus-…
#35 | DeepSeek's $12B raise, Moonshot eyes Hong Kong, Japan's AI action plan, Meta's Southeast Asia compute deal, and more… linkedin.com/pulse/35-deepse…
Delight retweeted
AI capabilities are advancing exponentially by some measures. Infrastructure builds more linearly. Organizations can take years to change.
The result: bottlenecks, risks, and opportunities.
New MGI research maps the AI economy as an interconnected system: mck.co/aieconomy
