Product updates
Decagon Experiment Hub isolates tested changes while other agent updates continue
Vendor-reportedChecked: [release]
Editorial attribution: Gosia Madej
Prepared with AI assistance. Editorial attribution does not imply personal product testing or manual review of every story.
Published: · Event date:
Decagon’s October 22 announcement introduces experiment blocks that lock the variants under test while leaving other agent content editable. Results count conversations exposed to the tested step, rather than all traffic.
Vendor-reportedChecked: [release]
What changes for users
For a planned escalation change, define the primary outcome and acceptable regression before allocating traffic. Check how many conversations actually reach the block and how overlapping releases are handled. A significance label alone does not explain practical effect size.
Editorial assessmentChecked: [release]
Decagon already offered A/B testing. Experiment Hub adds blocks holding variant content and traffic allocation, with up to 20 variants sharing a control. A conversation keeps its assigned variant after reaching the block, even if another agent version is published.
Vendor-reportedChecked: [release]
The announcement describes sample-size estimates, confidence intervals and significance verdicts. Teams can check variants in Simulations before live traffic. Gradual ramp-up, metric alerts and automatic stopping for significantly worse variants are vendor-described safeguards, not independently tested protections.
Vendor-reportedChecked: [release]
An exposure-only result answers what happened among conversations reaching that step. It does not establish the same improvement across the entire support queue. Continue tracking overall outcomes and concurrent changes before generalizing a winning variant.
Editorial assessmentChecked: [release]
Checked against sources
AI-assisted source check
How this was checked
No experiment was run and no statistical safeguard was reproduced. The announcement does not establish account-level enablement, edition eligibility or a measured lift for another customer. Source screenshot capture was blocked.
Editorial assessmentChecked: [release]
Source evidence
Original source excerpt. This shows what the source says; it does not establish product performance.
only the versions being compared are locked
Sources
- [release] Decagon: Experiment Hub announcementVendor-reportedPublished: 22 October 2026 · Checked: 22 October 2026
- [product] Decagon: Experiments product overviewVendor-reportedSource publication date not stated · Checked: 22 October 2026