Until now, Workly worked with a fixed team of 15 specialist agents. That team covers most of outbound well, but every market has its own sticking points. When replies kept stalling on the same question, the only fix was a manual prompt change.
With this release, Workly closes those gaps itself. It notices the pattern, builds a specialist for it, and tests that specialist before it ever talks to a real prospect.
How it works
1
It spots the gap
The reply analyst groups stalled threads. When the same objection keeps coming back and no agent handles it well, it flags a gap.
2
It builds a specialist
A new agent is written from your knowledge base, your past replies and the lessons already on file.
3
It tests it first
Every new agent runs a sandbox evaluation, a compliance review and the full regression suite. If any check fails, it is thrown away.
4
It joins your team
Only then is it promoted. From that point it handles the replies it was built for.
What you'll see in your workspace
A new Agents tab lists every agent on your team, including the ones Workly created. Each one shows why it was built, its test results and the replies it has handled.
You stay in control
You can pause or remove any agent Workly creates. Agent building can also be turned off in Settings › Agents.
Good to know
Available on all plans. Scale gets the full self-improvement loop.
New agents never change your voice or offer, only how a reply type is handled.
Every agent decision is logged in the decision trace.