Salesforce unveils Slackforce at Dreamforce 2026, letting users query and complete nearly all Salesforce tasks inside Slack through Surface-generated interactive interfaces, with two-way Slackbot voice due later this year.
Anthropic launches a beta redesign of Claude Code Projects that uses conversation to coordinate parallel coding threads, with each cloud session working on its own repository branch while shared memory preserves project decisions.
Google releases Gemini 3.8 Live and Extended Thinking for developers to build real-time voice agents that execute tasks mid-conversation, adding asynchronous function calling, live visual context and support for more than 97 languages.
Salesforce launches Koa, a business reasoning model built on Nvidia Nemotron 3 Super for long-running AI agents, alongside AIforce, Claudeforce’s 37 pre-built sales skills, and Headless 360 API access at Dreamforce 2026.
Gitlawb launches OpenClaude, an open-source terminal coding agent that works with cloud APIs and local AI backends, including Gemini, Ollama, Bedrock and OpenAI-compatible providers, while its --bg mode runs prompts in detached sessions with logs stored in ~/.openclaude/bg-sessions/.
Chat On Steroids now gives ChatGPT local MCP access to edit approved files, run tests, retain terminal sessions and control desktop tools through Chrome, while supporting multi-agent workflows on Windows, macOS and Linux.
Google releases ARTEMIS, an open-source platform that converts natural-language requests into Android automation, linking real devices to Antigravity, Codex, Claude Code and Windsurf via MCP, with a Pro mode designed for workflows of more than 100 steps.
shadcn-ui launches @shadcn/lint, an agent-first Tailwind linter for ESLint and Oxlint that guides coding agents to fix component, variant and theme violations in one correction round, with support for Tailwind v4 and Node.js 20.19+.
DreambigOu releases ELI5, a Claude Code skill that tailors explanations for children, managers, engineers and parents, with its evaluation reporting an 83.3% pass rate versus 41.6% without the skill.
LLMTraceFX ships a deterministic KV-cache auditor in merged PR #74 that independently verifies cache attestations and flags evictions, confirming token-identical cached outputs across 10 synthetic tests while showing an exact duplicate reuses 8 of 9 tokens.