Sources: Salesforce staff worry Anthropic's Claude Tag could cannibalize Slackbot and give the AI firm more leverage over the enterprise software industry
THE SO WHAT
Control over the in-Slack assistant layer is becoming a strategic chokepoint—if Claude Tag owns the UX, Slack risks becoming just a conduit. If you're building on Slack, assume the assistant surface may be intermediated by third-party models and design for portability across vendors.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIA profile of Jacob Andreou, the 33-year-old former Snap exec leading Microsoft's consolidated Copilot team efforts to catch up with OpenAI and Anthropic
Microsoft putting a 33‑year‑old consumer product operator over a consolidated Copilot org is a bet that AI distribution and UX coherence matter as much as raw model capability. If you're building on Copilot, expect faster, more opinionated product integration decisions—and less tolerance for fragmented AI experiments across Microsoft surfaces.
Applied AIResearchers say Z.ai's GLM-5.2 matches latest US models at finding security bugs, as critics question the US' lax approach in restricting Chinese open models
If GLM‑5.2 can match top U.S. models at vulnerability discovery, then model capability is now a shared global baseline, not a Western advantage. Security teams should assume adversaries have access to near‑frontier bug‑hunting models—raise your bar on code review, red‑teaming, and patch velocity accordingly.
Applied AISources: Google told Meta around March it couldn't offer all the Gemini capacity Meta wanted to buy, disrupting and delaying some of Meta's internal AI projects
When Meta can't buy all the Gemini capacity it wants, you know GPU scarcity has moved from vendor marketing line to real execution risk. Any team with material AI roadmaps should be locking in multi‑vendor capacity and on‑prem/hybrid contingencies now—not when your launch window is already slipping.
Wayfinder Router: deterministic routing of queries between local and hosted LLM
Deterministic routing between local and hosted LLMs is the kind of plumbing that makes hybrid AI architectures operational instead of aspirational. If your cost or latency curves are starting to bite, evaluate routers like this as a way to reserve frontier models for only the queries that truly need them.