Dagens Vibes — 23. august 2026

The MCP Project released its upcoming roadmap. As an open source project, this is directional, than committing. Exciting stuff: Better agent to agent communication, triggers and events, primitives for progressive discovery, DPoP and much more: https://t.co/f5X1xgKii0

https://x.com/dsp_/status/2091077636608950581

A PE operating partner asked us to build production AI agents inside a portfolio company's billing system, processing real healthcare claims under HIPAA. Two people hand-wrote every rule in their claims engine across 300+ denial codes and payer logic that changes quarterly. Four months later, seven production agents handle it with zero patient data exposure. First month, we didn't touch a model. We mapped their data: where it sits and what's missing, so agents reason from structured facts instead of guessing. I've watched teams skip this step across dozens of engagements. They bolt a model onto the product, watch it hallucinate over unstructured inputs, and decide AI isn't ready for their industry. The data work is what makes it ready. We built an enrichment layer that assembles 34 dynamic variables per claim before any LLM sees it, pre-computed and versioned so the agent receives ranked facts instead of searching for context. Every agent follows one pattern: pre-compute context, strip all patient data before the model sees it, validate output against a strict schema, let deterministic code accept or reject the action. If the output falls outside the allowlist, the system fails closed. Seven agents, each locked to a single workflow like denied claim follow-up or billing reconciliation, each running its own enrichment payload. Then we built the eval harness. Every agent runs against a curated test suite before any update reaches production. When a model provider ships a new version or payer logic changes, the harness catches regression before a single live claim is affected. The flagship agent reconciles denials to the penny: 59 out of 60 on the eval set. Most teams launch an agent and hope it keeps working. We launch one and prove it does on every deployment. We route calls across two model providers. Swapping one changes nothing in the output because the eval harness verifies it. Model integration was the shortest line item in the four-month build. The operating partner now benchmarks the rest of the portfolio against this system. That's the line between a portfolio company running AI and one still running demos.

https://x.com/mardehaym/status/2089648072217243780

🚨BREAKING: the stealth model Ox Alpha is https://t.co/0H6NxhUIyh (Zhipu) everyone else is guessing it from tokenizer vibes and emoji rates. i made the server say its own name out loud. sent it one malformed request and it threw a java stack trace naming its own internal class: https://t.co/lmGA1novUD.ChatCompletionRequest paas/v4/chat = Zhipu's documented route /api/paas/v4/chat/completions thats not a fingerprint guess. thats the server confessing. proven at the serving layer, confidence 0.98.

https://x.com/chetaslua/status/2091086141764354364

i really like this model man, it just gets me. ling 3.0 flash on one dgx spark, official int4, 40.9 tok/s on code, 128k context for balanced work. the spec decode is the mtp layer that ships inside the official checkpoint, 87% draft acceptance, nothing extra to download. these screenshots are ling 3 flash working agentic over telegram, it pulled the pr list off the github api, decided it needed requests, pip installed that itself autonomously, then went and counted the rows in the readme table. 0 hand holding on any of it. been clanking all evening and it has not dropped a tool call yet. if you own a dgx spark you should try it.

https://x.com/sudoingX/status/2091141379363864621

@scaling01 Agree. People are sleeping on using Opus to hill climb. We use it for optimizing CPU and memory, optimizing CI times, improving frame rates, reducing latency, any other kind of problem in the shape of “iterate on X with a profiler and dataset until it hits Y”

https://x.com/bcherny/status/2091308045284585724

Apple Watch -> Grok Bot -> Real Physical Objects

https://x.com/RayFernando1337/status/2091199083365171423

Charlie is doing incredible work on the CLI 🫶 This change is really noticeable!

Charlie Marsh@charliermarsh

In the latest Codex CLI release, I redid the lifecycle to make `codex` startup instant. It's now ~25x faster and immediately responsive. https://t.co/F01ohFN1T1

https://x.com/fcoury/status/2091275968522432974

@ajambrosino thought it was too hard to upload music to twitter, so he just prompted a whole music app hosted on ChatGPT Sites into existence just to share his album https://t.co/vOI40Xmxyc

https://x.com/thsottiaux/status/2091014197144596734