Release v0.89.0
Jul 02, 2026
Partner-wide everything — software, security, compliance, sensitive-data, peripheral and maintenance policies, alert notification channels, routing and escalation, and automations can all now be defined once and applied to every organization — plus scheduled reports that actually email out with branded PDFs, a rebuilt Windows remote-desktop encoder, ticket soft-delete, and AI write tools for billing and ticketing.
Multi-Org Policy Management
- Define a policy once, apply it to every organization. The 'All orgs' owner option now covers nearly the entire configuration surface: software policies, security policies, compliance rule sets, sensitive-data policies, peripheral-control policies, and maintenance windows can all be owned partner-wide instead of per-customer. Existing organization-scoped policies are untouched — partner-wide is a new option, not a migration.
- Partner-wide alert delivery. Notification channels, alert routing rules, and escalation policies can now be shared across all organizations, so your Slack channel, on-call rotation, and escalation chain are configured once instead of per customer. Organization-level rails still work and can override.
- Partner-wide automations. An automation like 'when a device goes offline, run this diagnostic script' can now run across every customer organization from a single definition, with results and alerts always attributed to the device's own organization.
- Partner-wide patching unblocked. The artificial restriction that kept patch features off partner-wide configuration policies is gone — a partner-wide patch policy now resolves and schedules end to end through your update rings.
- Isolation enforced in the database. Every partner-wide capable surface is protected by database-level row security with cross-tenant tests, not just application checks — one customer's data can never leak into another's, even partner-wide.
Reports & Scheduling
- Scheduled reports now actually run and deliver. Daily, weekly, and monthly saved reports execute on schedule in each organization's own timezone and email recipients the results — with branded PDF attachments, an executive-summary cover page, and trend deltas comparing this period to the last.
- Next-run visibility. Report schedules show when they'll fire next, so you can confirm a client report is queued without waiting for it.
Service Desk & Billing
- Ticket soft-delete with restore. Admins can now delete test, spam, and duplicate tickets instead of parking them in Closed forever — and restore them if a delete was a mistake.
- AI can now write to billing and ticketing. The Breeze AI agent (and connected MCP clients) can create and update quotes, invoices, contracts, and tickets — and the previously broken billing read tools were repaired, with a guardrail test so they stay working.
- Quotes, invoices, and contracts feel like one product. The three billing surfaces now share the same status pills, saving behavior, and loading treatments, with full editor parity across them.
Remote Management
- Faster Windows remote desktop. The video encode pipeline was rebuilt — asynchronous encoding, a zero-copy capture path, and a smarter bitrate ramp-up — for noticeably lower latency, especially on busy hosts.
- Adaptive quality survives reconnects. If a remote session drops and auto-reconnects, streaming statistics now carry over, so adaptive bitrate keeps working instead of freezing at the quality it had before the blip.
Reliability & Fixes
- Truthful device reliability scores. Two scoring bugs were fixed: routine Windows software events (like Volume Shadow Copy messages) no longer count as hardware memory or disk errors, and a device's own failure events are no longer double-counted against it. Healthy devices that were stuck with mysteriously low scores will recover as new data accrues.
- A reliability card that explains itself. The device reliability card now shows all five scoring factors with plain-language explanations, so a low score tells you why instead of leaving you guessing.
- Self-hosted UniFi telemetry fix. Deep-telemetry collection for self-hosted UniFi controllers no longer fails with a server error on repeat syncs.
Version 0.89.0 is the multi-org release. Breeze is built for teams that run many customer organizations at once, and this release makes that the default posture of the whole configuration surface instead of a special case.
The headline: define it once, apply it everywhere. Software, security, compliance, sensitive-data, peripheral-control, and maintenance policies; alert notification channels, routing rules, and escalation chains; and automations can all now be owned partner-wide — one definition covering every organization you manage, with per-organization overrides where you need them. Partner-wide patching is unblocked too. Everything existing stays exactly as it was: organization-scoped objects keep working, and “All orgs” is a new choice at creation time, protected end to end by database-level tenant isolation.
Beyond policy, scheduled reports now genuinely deliver themselves — branded PDF attachments, an executive-summary cover, and period-over-period trends land in recipients’ inboxes on schedule. Windows remote desktop got a rebuilt encoder for lower latency, tickets gained soft-delete with restore, the AI agent can now write quotes, invoices, contracts, and tickets, and device reliability scores stopped lying — two scoring bugs that unfairly punished healthy devices are fixed, with a score card that now explains itself. A recommended upgrade for every self-hosted deployment: ten quick idempotent migrations, no new environment variables, no breaking changes.