Urgent.News

What's breaking now, across thousands of outlets.

Tech

Connected steps are not a workflow

Take a task that crosses two apps. A customer asks for a refund. The request lives in the support desk. The order lives in the billing system. End to end, the agent reads both, decides against the refund rules, issues the refund, and confirms back to the customer. The rules are in neither API. Refunds under fifty dollars are fine when the order is paid and never refunded before. Over that, a…

A customer requests a refund, and the request exists within the support desk while the order resides in the billing system. The agent, upon reading both, determines that the refund is not warranted according to the rules, proceeds to issue the refund, and informs the customer. These rules are not defined within any API, so refunds under fifty dollars proceed smoothly, but anything above that threshold requires a manager's approval.

Testimonials are held for human approval before being released. Escalations occur after two interactions with the customer. However, these interconnected steps are not a true workflow. The agent reads the order in the billing system, then transfers this information to the support desk, where it becomes a snapshot of a moment that has already occurred.

The billing system allows for refunds under one permission set, while the support desk adds a customer note under a different set. Crucially, there is no confirmation that the refund has actually been processed before the note is sent out. This showcases that connected steps alone do not constitute a workflow. The agent carries out every action permitted by the apps, yet the outcome is incorrect because the rules and handoffs exist in the gaps between the systems.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

BERT's tokenizer bottleneck was hiding before WordPiece

BERT's tokenizer bottleneck was hiding before WordPiece Normalization consumed 88% of a warm English encoding pass with bert-base-uncased . WordPiece, the algorithm that chooses tokens, took 8%.

  • BERT tokenizer bottleneck hidden before WordPiece
  • WordPiece accounted for 88% of warm English encoding pass
  • Proposed branch improved BERT encoding 50x

More from Friday 16 October →