As Dextruss evolves from experimentation into production-grade customer journey orchestration, clarity and specialization become just as important as intelligence.
The v1.0.6 release focuses on removing structural constraints that surface as teams scale: rigid assistant configurations, cluttered admin workflows, and brittle handoffs between automation and human support. This update introduces foundational changes that make it easier to deploy purpose-built agents, manage them safely, and preserve customer context end-to-end.
There's a lot included in this release, so let's walk through what's new in Dextruss v1.0.6.
Multi-Agent Support: From a Single Assistant to Specialized Systems
Until now, most AI implementations assumed a single assistant configuration — one personality, one behavior set, one set of tradeoffs. In real post-sales environments, that model breaks quickly.
Support workflows require different behavior than onboarding.
Security requirements differ from marketing use cases.
Different customer segments need different interactions — at the same time.
With Dextruss v1.0.6, we've introduced true multi-agent support.
A new Agents section allows teams to create and manage multiple distinct agents from a single account. Each agent can be configured independently with its own:
- Behavior and decision logic
- Branding and website deployment settings
- Chat Assistant configuration and example prompts
This makes it possible to deploy specialized agents for different teams, workflows, or customer segments without fragmenting systems or duplicating infrastructure.
The result is specialization without sprawl — agents designed for their purpose, coordinated within one platform.
Streamlined Admin Interface: Less Friction, More Intent
As capabilities expand, structure matters.
In v1.0.6, we reorganized the Admin UI to better reflect how Dextruss is actually used in production environments. These changes focus on reducing cognitive load, clarifying intent, and preventing configuration errors over time.
Key updates include:
- Agents renamed to Functions to more accurately reflect their execution role
- Removal of unnecessary Group layers that added complexity without value
- Separation of global settings (Assistant configuration, example messages, widget settings) into dedicated modules
- New delete confirmation prompts to prevent accidental data loss
These changes are subtle, but intentional. They make it easier for teams to understand what they're configuring, where responsibility lives, and how changes affect the system as a whole.
Smarter Integrations and Stronger Security Controls
Automation only works if identity and context move with the customer.
In v1.0.6, we focused on strengthening the seams where Dextruss connects to external systems and human teams.
Improved Security Controls
- Ability to block anonymous users entirely at the widget level
- More reliable handling of authenticated user identity through the SDK
Better Context Transfer for Zendesk
For Zendesk users, Live Agent handoffs now automatically pass the customer's email address — not just an internal user ID — to Sunshine. This ensures support teams receive full customer context instantly, without additional lookup or manual reconciliation.
These improvements reduce friction at escalation points and ensure automation enhances, rather than obscures, human workflows.
Why This Release Matters
Dextruss v1.0.6 isn't about surface-level features. It's about building the architectural foundation required for autonomous systems to operate safely and effectively in real environments.
This release prioritizes:
- Specialization over one-size-fits-all AI
- Clarity over configuration sprawl
- Continuity of context across systems and teams
As customer journeys grow more complex, the systems that support them must become more intentional — not more complicated.
Looking Ahead
The changes in v1.0.6 lay important groundwork for what's coming next at Dextruss: deeper agent coordination, more autonomous execution across journeys, and tighter alignment between outcomes and system behavior.
If you're interested in testing multi-agent orchestration in a real post-sales environment, we're currently opening early access with select teams.