holaOS launches Beta 0.1 with Multi Workspaces, Sub Agents, and Dashboard

• ai_for_success's launch thread says holaOS Beta 0.1 shipped three core pieces: Multi Workspaces, Sub Agents, and a Dashboard.
• According to ai_for_success's Kickoff flow post, holaOS now starts a workstream with a structured intake of goal, context, sources, rules, and output format, then carries review corrections into later runs.
• testingcatalog's release post framed Beta 0.1 as a management layer on top of holaOS's earlier Agent Computer foundation, with each workstream keeping its own isolated context, memory, and agent.
• In testingcatalog's dashboard screenshot, the product already looks less like a chat tab and more like a persistent workspace with stored metrics, app navigation, and an active assistant pane.

You can jump straight to the GitHub repo from ai_for_success's try-it post, browse the releases page from testingcatalog's follow-up, and see a live dashboard UI in testingcatalog's screenshot post. One early user, in chetaslua's hands-on post, said a red-teaming pipeline carried memory over from the previous run.

Beta 0.1 adds three product-level features on top of the agent-computer base:

1. Multi Workspaces: separate context, rules, files, tools, and history for each workstream, per ai_for_success's feature list.
2. Sub Agents: parallel delegation under a single coordinator, per ai_for_success's feature list.
3. Dashboard: a control surface for active runs, review steps, and next actions, per ai_for_success's feature list.

The release framing in testingcatalog's release post is the useful part: holaOS is positioning these as an orchestration layer for ongoing work, not as a better one-shot chat.

The new Kickoff flow is the clearest process detail in the evidence. It collects five inputs up front:

• goal
• context
• sources
• rules
• output format

According to ai_for_success's Kickoff flow post, holaOS uses that intake to generate a first output immediately, then persists human corrections so the next run starts from the revised version instead of from zero.

The architectural split is simple in testingcatalog's architecture post: the agent holds state, the workspace holds context. That matches the product copy around Multi Workspaces, but it also points at the system's main bet, which is that persistent work needs a stable container around the agent.

The strongest usage claim in the tweet pool comes from chetaslua's hands-on post, which said a red-teaming pipeline remembered the last run. The same post also discloses it was a paid post, via chetaslua's paid post label.

The screenshot in testingcatalog's dashboard screenshot adds details the launch threads skip. The dashboard is shown reading from twitter_posts + twitter_post_metrics via WORKSPACE_DB_PATH, rendering stored aggregate snapshots over time, and pairing that with a right-side assistant pane inside the same workspace.

That makes Beta 0.1 look less like a memory checkbox and more like a persistent project surface with data sources, generated artifacts, and an always-on operator view.