Smarter goals, sharper KPI reporting, better agent teamwork, and more reliable campaigns
Turn agent conversations into owned goal drafts, get accurate KPI figures you can trust, coordinate teams across shared goals, and recover from campaign failures without losing your place.
What's new in Hub
This update covers four improvements across goal management, KPI reporting, agent teamwork, and campaign reliability.
Create goals straight from an agent conversation
Managing goals and agents used to mean juggling several competing panels. The workspace has been simplified so the most important information — the outcome, the owner, and the next action — stands out immediately, with advanced options tucked away unless you need them.
More usefully, admin conversations with your agents can now turn that discussion into an actionable goal. An explicit Create draft goal action creates an owned draft complete with success criteria and links to the relevant KPI evidence the agent has already gathered. The draft is created safely and won't create duplicates if you run it more than once, and it doesn't silently reassign agents or change KPI targets in the process.
Where KPI data genuinely isn't available, Hub now says so — rather than showing a misleading zero.
KPI numbers you can actually trust
A bug could cause the live KPI total to be capped at a lower value than the snapshots used for historical reporting — meaning your live figure and your history chart told different stories. That's now fixed: the live scan runs completely up to a defined safety budget so both numbers are consistent.
Archived records are now excluded consistently from KPI calculations, and an explicit timezone setting with proper daylight-saving-time awareness means scheduled period boundaries match what you'd expect — not whatever the server clock happened to say.
If a read fails for any reason, the resulting value is carried through formulas as "unavailable" rather than silently becoming zero. Existing snapshots are unaffected.
On the interface side, the history chart's actual line now uses a solid theme colour for clarity. Goals and history panels expose a Retry state so you can recover from a failed load without losing any draft you were working on. The KPI editor has also been updated to use compact, consistent field components throughout.
Agents that work together across shared goals
Agents now expose their full remit — objectives, constraints, shared Hub Goal responsibilities, and the accountable owner — in a common workspace called Hub Compact. Leadership agents can use named teams, pull in KPI evidence, and propose work for admin approval, all from one place.
The update also adds:
- Draft creation with reusable versioned templates, so recurring work doesn't start from scratch each time
- Configuration previews before a template goes live
- Role-labelled directory filtering so you can find agents by what they do, not just what they're named
- Visible objective details alongside approval and result controls, so reviewers have the context they need without navigating elsewhere
Source freshness is now explicitly marked as unknown where the underlying data doesn't provide a timestamp, rather than leaving it ambiguous.
Campaigns that recover gracefully and link records correctly
If a campaign encountered a failure, it could previously show empty data or an incorrect setup prompt — making it hard to tell whether you had a configuration problem or a transient error. Recoverable failures now show a clear, actionable message instead.
Campaign detail pages can now link existing CRM records through declared relations. Concurrent multi-record membership is preserved during linking, and you'll see a warning before a singular relation is replaced — so you're never surprised by a record being silently swapped out.
On the data integrity side, the marketing webhook now rejects unknown, malformed, or ambiguously owned topics rather than falling back to a global consent change. Generic updates can no longer accidentally opt contacts back in, and missing webhook authentication now fails closed rather than passing through.
More improvements are in progress. We'll keep you updated as they roll out.