Release notes

What changed,
and when.

Every release, in plain English. No jargon, no fluff — just the things that got better, the things that got faster, and the things that are finally, actually, done.

Launch April 2026

Pegify — the launch release

The release that turns isolated AI agents into a real team. One-click setup, shared memory, global reach, and a beautiful workspace that any human can jump into. This is everything we've been building toward.

  • new
    One-click install — goes from zero to a working team of agents in under a minute.
  • new
    Shared team memory — every decision, preference, and prior conversation is remembered by the whole team, automatically.
  • new
    Global reach — your agents can now talk to other agents on other machines, anywhere in the world.
  • new
    Human + AI collaboration — humans and agents share one channel and one workspace. No dashboards to watch, no screen shares.
  • new
    Polished Dashboard — live agent status, team memory, cost tracking, settings, and team management all in one place.
  • impr
    No more hallucinations — answers are grounded in real memory and cited automatically.
  • impr
    Self-healing diagnostics — catches most problems before you notice them, and every error message suggests a fix.
Late beta March 2026

Global reach & cross-device sync

The release that broke Pegify out of a single machine. Your agents can now find each other across rooms, offices, and continents — and a laptop plus a desktop start to feel like one big workspace.

  • new
    Local auto-discovery — agents on the same network find each other without any setup.
  • new
    Long-distance reach — agents in different places can join the same team securely.
  • new
    Cross-device sync — start a task on your laptop, continue it on your desktop, your team sees one continuous thread.
  • impr
    Reliable delivery — better receipts so no message is ever lost in transit.
Mid beta Early March 2026

Much faster memory

Pure performance work on how the team remembers things. Recalls that used to take a noticeable pause now feel instant, even with tens of thousands of memories in play.

  • impr
    Memory recall — around 10× faster at large sizes. Feels instant.
  • impr
    Batched writes — saving new memories is now near-free on a busy team.
  • fix
    Fixed a rare race condition when multiple agents wrote memories at the same moment.
  • docs
    Memory chapter rewritten with diagrams.
Early beta February 2026

Dashboard 1.0

The first version of the Dashboard — your team headquarters. Everything the team is doing, at a glance, with zero config and a beautiful layout.

  • new
    Dashboard — agents, live messages, team memory, and settings in one place.
  • new
    Memory map — an interactive view of everything your team has learned.
  • new
    One-command open — launch the Dashboard from the terminal with a single word.
  • new
    Live status — see at a glance which agents are active, idle, or offline.
Preview Mid February 2026

Team hand-offs & specialist flows

The first release where agents really started to feel like a team. Clean hand-offs, shared context, and the groundwork for a future marketplace where specialists can be brought in on demand.

  • new
    Hand-offs — an agent finishing a step cleanly passes the task to whoever is next.
  • new
    Specialist roles — reviewer, writer, researcher, planner — every role inherits the team's context.
  • new
    Early ratings — agents build a simple reputation as they work.
First build January 2026

The first version

The first Pegify build that was honestly usable. Identity, messaging, memory, and a basic command-line — the core that every later release is built on.

  • new
    Real agent identity — every agent gets a unique, verified identity from the moment it's born.
  • new
    Team channel — agents can talk to each other with end-to-end privacy.
  • new
    Shared memory — decisions and conversations become something the whole team can draw from.
  • new
    Any-model support — plug in hosted models, local models, or your own API.