Changelog
Follow CrossMind's evolution. New updates every 2 days.
Coding agents now hand work to each other inside one conversation — plus a composer that plans and loops, agents shipped as apps, and local runs that don't touch your quota.
Local agent sessions now recover from the failures that used to kill them — plus a first-class home for your machines and a real skills manager.
A real desktop app, an agent that reaches your local tools, and settings that finally live in one place.
Dashboard Hub loses its clutter, Product Foundation stops crashing, and Google Search & Analytics setup gets simpler.
Content planning organized by time instead of status, every engagement reply in one place, and analytics that fix themselves
Content scheduling rebuilt from scratch, dashboard polish everywhere, and browser automation that doesn't trip alarms
Onboarding now proves what the agent can actually do, drafting and publishing got unified into one flow, and agents can generate and edit their own images
AI-generated content ideas grounded in your strategy, campaigns that live directly on a bet, and multi-account support for your approval queue
Channels are now first-class with per-channel timezones, agent-driven publishing for platforms without OAuth, and one unified analytics view.
An agent-driven content pipeline, strategy and evidence unified into one dashboard hub, and a single workspace for every person and action your agent touches
Review and act on every agent action from one inbox, pick model tiers with predictable billing, and peek into any running session without leaving your workspace
Configure AI abilities with guided forms, watch your dashboard update automatically after every task run, and manage everything about your agent from a redesigned settings panel
Rule-based approval flows, external schedule triggers, and reliability fixes
Agents can now run their own mini-apps on public URLs. Plus: the Ability Marketplace, Agent Identities, bring-your-own API key, and a unified settings layout.
A full admin console for managing your platform, a cleaner model setup, and fixes that keep emails and sessions running smoothly.
Agents now recover from errors and timeouts automatically, Claude Opus 4.7 is live in production, and the interface got a round of focused polish.
Anyone can now start a trial without an invite. Task history is navigable with jump cards. Agents behave more consistently thanks to runtime and context fixes.
Artifact data tables now support a smarter Linear-style list view with inline editing and clickable rows, plus multiple reliability fixes for browser sessions and subagent recovery.
Artifact tool cards render more reliably, pending request panels are consolidated for cleaner UX, and browser takeover lifecycle is more robust.
Onboarding is redesigned with better guidance, six new preset dashboards ship out of the box, and agents recover from errors more gracefully.
Browser takeover is more reliable, social auth works end-to-end, artifacts download properly, and sign-up recovery is cleaner.
Memos now open fresh sessions for clarity, model selection is more consistent across schedules, and the task progress panel has a refined look
Pick the AI model behind your agent or bring your own API key, manage browser tasks directly from schedules, and enjoy more reliable autonomous runs
Social outreach tasks run more reliably across platforms, browser-based tasks recover faster when something goes wrong, and schedule management is cleaner with a new email preview feature
A revamped onboarding experience gives you instant task context, a self-healing browser engine ends stuck sessions, and social auth is now more reliable across all platforms
A new workspace dashboard gives you live task health at a glance, each schedule can now run on its own model and settings, and billing is cleaner from upgrade through checkout
Choose between Claude models on the fly, collaborate with multiple AI personas on the same task, and connect integrations without leaving your workflow
Persona switching is now session-aware with automatic scene routing, and the persona switcher got a full polish pass
New Plus pricing tier, multi-personality agent configuration, reliable session interrupts, and expanded internationalization
New integrations catalog with permission modes, smart email routing, Telegram streaming, and task management improvements
Unlimited skills, automatic skill discovery, mobile improvements, and agent sandboxing
Run multiple copilot sessions simultaneously, get notified when they complete, and see file changes more clearly
Improved browser debugging, stabilized message rendering, and introduced new pricing options
Major stability improvements for OAuth integrations and task scheduling
Building in public
Every update brings CrossMind closer to being the cofounder you need. Want early access?
Get Started