See how AI handles these workflows

Published examples from other companies

These are case studies published by other companies, showing that these approaches are already in use elsewhere. The figures are as disclosed by each source and not independently audited. These companies have no relationship with SY Technology and are not our clients. Your results are estimated after a diagnosis, based on your own situation.

• Klarna (Swedish payments company). Announced in February 2024: in its first month, its AI assistant handled 2.3 million conversations, about two-thirds of customer service chats, and cut average resolution time from 11 minutes to under 2. In May 2025, its CEO told Bloomberg that cost-cutting in customer service had gone too far and that customers should always be able to choose a real person.

  What we borrow：Let the AI answer routine questions first; discounts, complaints and anyone asking for a person go to a human.

  Source：Klarna press release, Feb 2024, Bloomberg, May 2025 · as disclosed by the source

• Telstra (Australian telecoms company). Published in April 2026: its internal AI assistant AskTelstra helps 8,000+ front-line staff find answers faster, cutting average customer call time by over a minute.

  What we borrow：Answers come only from the company's own documents, so staff stop chasing people for answers.

  Source：Telstra article, Apr 2026 · as disclosed by the source

• Medtronic (US medical device company). Uses AI to pull material information and part numbers from engineering drawings (built by IBM Consulting on Microsoft generative AI). Published in 2024: up to 90% accuracy on key material information and over 85% on part numbers.

  What we borrow：AI reading documents isn't 100% accurate, so anything uncertain must be confirmed by a person.

  Source：IBM case study, 2024 · as disclosed by the source

• A brick-and-mortar retailer in shopping malls (China, anonymous). A case compiled by Datawhale: AI automatically checks outlet till records, mall settlement statements and paper slips, and staff only handle the differences the AI flags. Reconciliation went from one or two person-days per cycle to a few hours.

  What we borrow：Let the system match first; people only look at what doesn't match.

  Source：Datawhale “FDE Cases 100” CASE-015 · as disclosed by the source

• A custom-manufacturing company (China, anonymous). A case compiled by Datawhale: AI reads drawings into structured data and generates quotes from the company's materials, process and past-pricing rules; people calibrate before anything goes out. Quoting workload fell by at least 50%.

  What we borrow：The system drafts the numbers; a person checks before the customer sees them.

  Source：Datawhale “FDE Cases 100” CASE-019 · as disclosed by the source

Notes

• Company names, people, places and amounts in the demos are fictional demo data, used only to explain the workflow.
• The demos show what it looks like once built around a client's workflow, not off-the-shelf software. Which systems can be connected, and how, is confirmed one by one at diagnosis.
• The real system recordings come from a system we built for a licensed moneylending company, shown with demo data.
• Company names, figures and trademarks belong to their respective sources and are shown for reference only. Case research drew on the open-source FDE-case-library (MIT License, Copyright (c) 2026 guminghao-avi contributors); the full licence is below. For Datawhale cases, Datawhale's own pages are the authority.

FDE-case-library (https://github.com/guminghao-avi/FDE-case-library)

MIT License

Copyright (c) 2026 guminghao-avi contributors

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.