# Morning AI TL;DR — 2026-05-23

- **Microsoft open-sourced RAMPART and Clarity for agent safety testing** — RAMPART turns adversarial/benign agent scenarios into repeatable CI tests, while Clarity targets safer agent development workflows; useful for teams shipping tool-using agents. [Source](https://www.microsoft.com/en-us/security/blog/2026/05/20/introducing-rampart-and-clarity-open-source-tools-to-bring-safety-into-agent-development-workflow/)
- **Google pushed Gemini 3.5 Flash into mainstream surfaces** — Google says 3.5 Flash is now the default model for the Gemini app and AI Mode in Search globally, raising the baseline for fast agent/coding-style interactions. [Source](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/)
- **Google’s developer story is now explicitly agent-first** — I/O developer updates centered on Antigravity, Managed Agents in the Gemini API, and AI Studio expansions, signaling a move from chat APIs to managed agent deployment surfaces. [Source](https://blog.google/innovation-and-ai/technology/developers-tools/google-io-2026-developer-highlights/)
- **Mistral previewed Medium 3.5 plus remote coding agents in Vibe** — Mistral says Medium 3.5 is the new default for Vibe and Le Chat, paired with long-running cloud coding agents and a Work mode for complex tasks. [Source](https://mistral.ai/news/vibe-remote-agents-mistral-medium-3-5)
- **Salesforce’s Agentforce Coworker surfaced as a beta developer doc/product signal** — Salesforce docs describe Agentforce Coworker as a beta feature tied to search/result workflows, suggesting Agentforce is moving closer to an always-available CRM action layer. [Source](https://developer.salesforce.com/docs/data/agentforce-coworker/guide/agentforce-coworker-a-home.html)
- **Formula 1 expanded Agentforce for fan engagement** — F1 and Salesforce are adding an Agentforce-powered companion that answers questions from trusted F1 sources about 2026 technical regulations, a concrete consumer-facing Salesforce AI deployment. [Source](https://investor.salesforce.com/news/news-details/2026/Formula-1-and-Salesforce-Deepen-Partnership-Expanding-Agentforce-to-Grow-Fan-Connection-Worldwide/default.aspx)
- **GitHub is leaning into enterprise AI coding-agent positioning** — GitHub highlighted Gartner recognition for enterprise AI coding agents, reinforcing that governance, shared platform controls, and agent reliability are now key differentiators beyond autocomplete. [Source](https://github.blog/ai-and-ml/github-copilot/github-recognized-as-a-leader-in-the-gartner-magic-quadrant-for-enterprise-ai-coding-agents-for-the-third-year-in-a-row/)

**Worth watching:** The practical edge is shifting toward agent safety/evals, governance, and CRM/workflow integration—not just bigger model benchmarks—so builders should invest in test harnesses and deployment controls now.