HUGE Cursor Update and DeepSeek v4 Catches Up To Opus 4.8?
TL;DR
Codex and Cursor are converging on the same idea: OpenAI's new Codex Sites and Cursor's new Canvases both let teams generate internal web apps tied to company data like Gmail, Slack, and databases.
Codex now looks a lot more like a vibe-coding platform: Riley shows a Kanban board built with a Convex database plus a custom /con skill, so tasks can be added from any chat by any connected agent.
DeepSeek V4 is becoming a real economic threat: Lindy founder Flo Crivello says they moved 100 percent of traffic to DeepSeek V4, saving millions while seeing better performance on some core tasks than Anthropic models.
Model pricing is becoming the whole story for consumer agents: Riley says DeepSeek V4 Pro costs about $1.30 per 1 million tokens versus $30 for Opus 4.8 and $35 for GPT-5.5, which he argues makes premium agents too expensive for everyday users.
Apple may be turning iMessage into an agent platform: poke.com is now an officially approved iMessage agent, and Riley expects WWDC could introduce a broader approval flow for companies to ship agents inside Messages.
Hermes and Microsoft both signal the stack is widening: Hermes launched a desktop app as an open alternative to Claude-style super apps, and Microsoft introduced an always-on OpenClaw companion, new hardware, and fresh AI models Riley plans to test.
The Breakdown
DeepSeek V4 may already be good enough to replace Anthropic and OpenAI on real agent workloads at roughly 23 to 27 times lower cost, while Cursor and Codex both take a big step toward becoming full AI app platforms with internal site-building features. Riley Brown also flags Apple turning iMessage into a possible consumer agent hub, Hermes shipping a desktop app, and Microsoft finally making a serious AI push.
Was This Useful?
Share
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
The Art of Tasteful Prompting
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.

Playbook
The Codex /goal Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.