Hello,

We would like to follow up with you regarding the slowdown experienced on the Swiftask platform on July 23, 2026, between 2:00 PM and 9:00 PM (UTC), which affected all our users.

What happened

After a thorough investigation by our technical teams, we identified the source of the issue: in certain configurations within the multi-agent module, an agent and its sub-agents could trigger a loop by calling each other repeatedly. These repetitive executions gradually saturated our database connections, which slowed down the entire platform.

We want to reassure you: no data was compromised, lost, or exposed. This incident was strictly performance-related.

What we have done

As soon as the cause was identified, our teams worked to deploy fixes, which were put into production on July 23 at 9:00 PM (UTC). These improvements include:

• Implementing an anti-loop safeguard, which automatically detects and interrupts recursive calls between an agent and its sub-agents;

• Optimizing our caching system to reduce the load on our database;

• Patching our parallel processing management for better load distribution.

What we continue to do

We are actively continuing our work to strengthen the platform's reliability over the long term:

• At the application level: reinforcing early detection mechanisms for agent loops;

• At the infrastructure level: reinforcing our capacities and enhancing monitoring of database connections to detect such situations before they impact your experience.

We are fully committed to ensuring a stable and high-performance service, and we will keep you updated on the progress of these improvements.

Thank you for your trust and patience.

Sincerely,

The Swiftask Team