For hospital and health system IT, governance, data security, and quality teams, evaluating a new translation solution means looking well beyond speed. Teams need to understand how patient information is protected, how translation quality is maintained, how technology is used responsibly, and how the solution integrates into existing clinical workflows.
Fetch™, The Language Group’s on-demand healthcare translation solution, was built specifically for the demands of hospital discharge. Fetch combines fast, human-reviewed translation with PHI protection and EHR integration to deliver After Visit Summary (AVS) translation in 15 minutes or less.
To answer some of the most common technical, security, and quality questions healthcare organizations ask about Fetch, we spoke with Rolf Hecken, Vice President of Technology at The Language Group (TLG), and Chandler Pitcher, Chief Information Officer (CIO) at Commence, our Fetch technology partner. They explain how Fetch handles PHI, maintains translation quality, achieves rapid turnaround times, addresses AI concerns, supports complex languages, and fits into hospital workflows.
“The advantage is not just speed, although speed is one of the biggest differentiators. The advantage is that Fetch brings speed, security, quality, completeness, and workflow integration together in one solution.”
– Rolf Hecken, Vice President of Technology at The Language Group (TLG)

Q&A: How Fetch Enables Fast, Secure, and Compliant After Visit Summary (AVS) Translation for Hospitals and Health Systems
1. How does Fetch protect PHI throughout the translation process?
Chandler Pitcher, Chief Information Officer (CIO) at Commence: Fetch is designed with PHI protection built into the workflow from the outset.
When PHI is required to process a request, it is stored temporarily in a US-based environment with HIPAA-aligned administrative, physical, and technical safeguards and an independent SOC 2 Type II attestation. The PHI handling process is separate from the translation workflow, meaning PHI is not exposed unnecessarily during translation. To preserve the clarity and understanding of the source text, direct patient identifiers are redacted in a meaningful way that maintains the translatability of sentences in context for the Post Editors.
Once the translated project is delivered, temporary PHI is securely deleted in accordance with Commence’s documented data retention and disposal procedures and the client’s Business Associate Agreement.
Commence executes a Business Associate Agreement with each client. PHI is encrypted in transit and at rest, access is restricted to authorized personnel on a role-based, least-privilege basis, and access to PHI is logged for audit.
For hospital IT, governance, and data security teams, the goal is simple: minimize exposure, control where patient information is stored, keep PHI separate from the translation process, and limit retention of patient data to what is required to complete delivery.
2. What quality measures are in place to ensure translation accuracy?
Rolf Hecken, Vice President of Technology at The Language Group (TLG): Fetch was built to move quickly, but not by removing quality controls.
The workflow uses client-specific translation memory, client-specific neural machine translation, and qualified human linguist review. That combination matters because healthcare systems often have specific terminology, preferred phrasing, and recurring AVS content that should be translated consistently across all facilities.
Translation memory helps maintain consistency by reusing previously approved language where appropriate and contextual accurate. Neural machine translation helps accelerate the first pass. Terminology management and TermQA are of course integrated parts of the process. Then, the output is post-edited by a qualified linguist in accordance with ISO 18587 standards.
Furthermore, specific Language QA teams perform frequent Quality assessments of MT, TM and Human output from a linguistic quality perspective adding Qualitative in addition to the Quantitative measurements we gain during production.
For quality and governance teams, that means Fetch is not simply pushing raw machine translation into the patient record. It is a controlled translation workflow with human reviews built in.
3. How does Fetch achieve such a rapid turnaround time compared with other translation workflows?
Hecken: The speed comes from changing the model.
Traditional document translation workflows are often built around manual project intake, assignment, translation, review, and delivery. That process works for many types of content, but it is not always practical for discharge instructions, where timing is critical.
Fetch fully automated workflows are specifically designed to solve the AVS translation challenge. The request can move quickly through the system, supported by translation memory, neural machine translation, and qualified human post-editing. Our dedicated Post editors are scheduled in shifts and are ready to pick up and review the files as they become available. These redacted files are accessed through a secure online platform, providing the post editors with all the info they require at their fingertips and in visual context.
The result is a much faster process while still maintaining healthcare translation quality standards. Instead of waiting hours or days for translated discharge instructions, Fetch is designed to return a human-reviewed translated AVS in minutes, with 15 minutes or less as the target turnaround.
That speed can make a meaningful difference during discharge because the translation process supports the workflow instead of slowing it down.
4. How does Fetch safeguard against common AI concerns in healthcare, such as hallucinations, bias, and data privacy?
Pitcher: Fetch is not a large language model or open-ended generative AI workflow.
That distinction is important for healthcare organizations. Fetch is designed for translation of defined source content from the AVS or discharge record. It is not generating new clinical advice, adding new instructions, or making open-ended recommendations.
The workflow is also human-reviewed. One hundred percent of the translated output is reviewed by a qualified linguist according to ISO 18587 post-editing standards before delivery.
From a data privacy standpoint, client data is used only to build and maintain that client’s own translation memory and machine translation engine, as permitted under the client’s Business Associate Agreement, and is never used to train models for other clients. PHI is handled separately from the translation workflow, stored temporarily in a secured US-based environment, and securely deleted after project delivery in accordance with documented disposal procedures.
For IT, governance, and quality teams, Fetch helps reduce common AI concerns by avoiding open-ended generative output, limiting the workflow to translation of source content, protecting patient data, and maintaining human review before delivery.
“Fetch is not a large language model or open-ended generative AI workflow… It is not generating new clinical advice, adding new instructions, or making open-ended recommendations.”
– Chandler Pitcher, Chief Information Officer (CIO) at Commence

5. Does Fetch support complex language requirements, including double-byte and right-to-left languages like Chinese and Arabic?
Hecken: Yes. Fetch supports double-byte languages, such as Chinese, and right-to-left languages, such as Arabic.
That matters because healthcare translation is not only about converting words from one language to another. The translated AVS also needs to remain readable and usable for the patient or family. Language direction, characters, formatting, and document structure all affect the final patient-facing output.
Fetch is currently available on demand in certain languages, with additional languages possible by request. For health systems, we can review language coverage, formatting needs, expected volume, and turnaround requirements during implementation.
The goal is to support real hospital workflows while maintaining quality, readability, and speed.
6. Why should hospital IT, governance, data security, and quality teams consider Fetch instead of another translation solution?
Pitcher: Fetch is different because it is live, proven, purpose-built for healthcare, and designed for the pace of hospital discharge.
Many translation workflows still depend on manual handoffs, in-house translator availability, or outside translation requests that can take hours or days. Other emerging solutions may focus on only part of the discharge content, such as provider-edited or free-text fields. Fetch is designed to translate the complete AVS or discharge record, including templated EHR content that may be clinically important for patients and families.
For IT teams, Fetch is built to fit into EHR and AVS workflows, including Epic environments. For governance and data security teams, Fetch is designed around PHI protection under a Business Associate Agreement, minimum-necessary access, temporary storage, controlled and audited workflows, and secure deletion after delivery. For quality teams, Fetch combines translation memory, neural machine translation, and qualified human review in accordance with ISO 18587.
Hecken: The advantage is not just speed, although speed is one of the biggest differentiators. The advantage is that Fetch brings speed, security, quality, completeness, and workflow integration together in one solution. That solution is fully integrated into your EHR system discharge processes and executed by simply clicking ‘print’. Only a minimal addition to the regular discharge process, fast and easy to use.
For hospitals and health systems, that means translated discharge instructions can now be delivered in minutes instead of hours or days, without relying only raw machine translation, exposing PHI unnecessarily, or only partially translated material or even creating a separate manual process for care teams.
On the patient side, the discharge information is always complete, current, and available in both languages. This ensures that critical information like wound care or medication is always available for the patient after leaving the hospital or healthcare facility or on-site language support is not currently available.
Built for the Security, Quality, and Workflow Demands of Healthcare
For healthcare organizations, hospital discharge translation has to work across several priorities at once. Speed matters, but so do translation accuracy, PHI protection, data governance, clinical workflow integration, and the ability to deliver complete information patients can use after discharge.
Fetch brings those requirements together in a solution designed specifically for real-time healthcare translation. Its combination of EHR integration, human-reviewed output, protected PHI workflows, complete AVS translation, and rapid turnaround gives hospital IT, governance, security, and quality teams a practical way to support language access without introducing another disconnected process.
Explore Fetch™ for Your Health System
If your organization is evaluating faster, more secure ways to provide translated discharge information, explore Fetch to learn more about its workflow, security, compliance, and integration capabilities.
Ready to discuss your health system’s requirements? Connect with The Language Group’s healthcare translation team.












