Language Access Cannot Be Automated Away: Why Human Talent — and the Systems That Manage It — Still Matter
By Shelby Edwards, Co-Founder & CXO, Eclipse Scheduling
Carol Velandia's recent piece in MultiLingual Magazine, "Why Language Access Cannot Be Automated Away", put words to a conviction I have carried through my entire career as an interpreter, an LSP operator, and now as a co-founder of a language technology platform. Her central argument is simple and, in my view, unimpeachable: language access is a civil right, not a logistics problem — and civil rights cannot be outsourced to a chatbot.
I want to build on Carol's argument and take it one step further. Because while she rightly defends the irreplaceable role of human linguists, there is a second, quieter truth that deserves the same spotlight: the human professionals who make language access real do not work in a vacuum. Behind every qualified interpreter, translator, captioner, and CART provider stands a language service provider (LSP) or an in-house language access office — and behind those teams must stand a service management system capable of doing the matching, the compliance work, and the trust-building at scale.
Where I Stand With Carol
Carol's framing of AI in high-stakes settings is exactly right. AI is a powerful assistant. It can draft, summarize, transcribe, translate low-risk content, and surface patterns in mountains of data that no human team could process unaided. Used well, it expands access. Used carelessly, it manufactures the illusion of access while quietly stripping people of their rights.
Her courtroom story — a public defender "explaining" a plea deal through Google Translate — is the kind of moment that should stop the industry cold. Not because the technology failed in some exotic way, but because the institution treated language access as a checkbox rather than a safeguard. The tool was not the villain. The absence of a human professional with judgment, ethics, and accountability was.
So let me say it plainly: at Eclipse, we agree. AI belongs inside the workflow, never on top of the person. The four pillars Carol borrows from Juan Santiago — judgment, cultural leadership, agility, and governance — are not "soft skills." They are the load-bearing structure of meaningful language access.
The Part I Want to Amplify: Human Talent Needs Human Coordinators
Carol's article focuses on the linguist in the room. I want to widen the lens to the team that put that linguist in the room in the first place.
In healthcare, that team is a hospital's language services department. In K–12 and higher education, it's an interpreter coordinator or accessibility office. In the courts, it's a court interpreter program manager. In the private market, it's an LSP scheduling coordinator working a queue of requests across dozens of clients and hundreds of providers. Different titles, same job: matching the right human to the right encounter, at the right time, under the right conditions.
That matching is not clerical work. It is the moment where civil rights either get honored or quietly eroded. Consider what a qualified coordinator is weighing on every single assignment:
- Is the provider credentialed for this setting (medical, legal, educational, mental health)?
- Are their certifications current, and have I documented the verification?
- Is there a conflict of interest — a prior relationship, a family connection, an opposing party they have worked for?
- Does the LEP or Deaf consumer have a stated preference (a specific provider, a gender preference for a sensitive exam, a regional dialect, a trauma-informed background)?
- Is the provider's modality and language pair actually right — not "Spanish," but Mexican Spanish vs. Caribbean Spanish vs. K'iche' surfacing mid-encounter?
- Are we honoring Section 1557, Title VI, ADA, FERPA, and any contractual SLAs the client has signed?
- Did we broadcast, confirm, and document the assignment in a way that holds up to audit?
No app handles that on its own. A human professional — supported by software that surfaces the right information at the right moment — does.
Why LSPs and Language Access Offices Are the Real Infrastructure
Carol writes that "language access is infrastructure." I'd extend that: the organizations that staff and govern language access are infrastructure, too. LSPs and in-house language access offices are the connective tissue between civil-rights law on paper and a qualified interpreter walking into an exam room.
When that connective tissue is weak — understaffed coordinators, spreadsheets pretending to be scheduling systems, no credential tracking, no audit trail — three things happen, and I have watched all three happen across the industry:
- Conflicts of interest slip through. A relative gets assigned to interpret an abuse disclosure. An interpreter who worked for the opposing party in a prior case gets dispatched to a hearing. Nobody flagged it because nothing in the system was watching for it.
- Compliance erodes silently. Expired certifications go undetected. Section 1557 documentation requirements get missed. Title VI and ADA obligations turn into "we think we covered that."
- Consumer preferences get steamrolled. The patient who requested a female interpreter for an OB-GYN visit gets a male provider because no one captured the preference in a system. The Deaf student who works best with a specific CART provider gets whoever was available. Trust erodes one assignment at a time.
These are not exotic failure modes. They are everyday failures of operational infrastructure, and they are exactly what Carol means when she warns against treating language access as a logistics problem you can app your way out of.
Where Eclipse Fits: A Service Management System Built for This Reality
This is where I want to be honest about Eclipse's role and where I think purpose-built technology actually earns its place in Carol's framework.
Eclipse is not an AI translator. It is not a replacement for a human interpreter, translator, or captioner. It is an interpreter management system (IMS) — a service management platform that gives LSPs and language access offices the operational backbone to deploy human talent the way Carol describes it: with judgment, cultural leadership, agility, and governance.
Concretely, that means:
Intelligent matching that respects conflicts of interest
Eclipse's scheduling and matching logic considers far more than "who is free at 2 p.m." It surfaces certification status, modality, language pair, prior history with the consumer or opposing party, and any flags a coordinator has set — so the human making the assignment has the right information before they commit a provider to a high-stakes encounter.
Compliance you can defend in an audit
Credential tracking, document expirations, HIPAA-grade access controls, SOC 2 Type II Attested infrastructure, and a full audit trail on every assignment, broadcast, confirmation, and edit. When a regulator, client, or plaintiff's attorney asks "who interpreted this, with what credentials, under what process?" — the answer is one report away, not one frantic email chain away.
Consumer preferences treated as first-class data
The patient's preferred provider, the student's preferred CART writer, the gender preference for a sensitive exam, the dialect that actually matches the consumer's lived language — these belong in a structured field that the matching engine reads on every request, not in a sticky note on a coordinator's monitor.
AI used where it helps, never where it shouldn't
We use AI for intake parsing, smart reminders, no-show prediction, utilization analytics, and reporting. We deliberately do not use AI to make the final assignment decision in high-stakes encounters, to communicate directly with LEP consumers without human oversight, or to generate any output that reaches an end-consumer without a qualified linguist in the loop. That line is non-negotiable.
Built by interpreters, for the people who coordinate them
Eclipse was founded by career interpreters and LSP operators. We built the system we wished we had when we were the ones scrambling at 4:55 p.m. to fill a 5:00 p.m. medical request without compromising on credentials, conflicts, or consumer preference.
Bringing It Together
Carol's argument and mine end up in the same place from two different angles. She defends the interpreter in the room. I want to defend the entire operational chain that puts that interpreter in the room — and to argue that the chain itself deserves real investment, not duct tape.
If you run a language services company, an in-house language access program at a hospital or health system, an interpreter program in a court system, or an accessibility office in a school district, here is the question I'd ask after reading Carol's piece:
When the next high-stakes encounter happens — the K'iche' speaker in a Spanish-language hearing, the Deaf patient in an ED, the immigrant parent at an IEP meeting — does the system behind your interpreter give them the information, the conflict checks, the credential verification, and the consumer-preference signals they need to honor that person's rights? Or are you relying on memory, spreadsheets, and luck?
AI will not fix that answer. Neither will another standalone app. What will fix it is the same answer Carol gives, applied one layer up: human talent, supported by purpose-built infrastructure, governed by people who take responsibility for the outcome.
That is the work LSPs and language access offices do. That is the work Eclipse exists to support.
If you want to see how Eclipse helps LSPs and in-house language access teams operationalize the principles Carol writes about, schedule a demo or explore the platform.
Further Reading
- Carol Velandia — Why Language Access Cannot Be Automated Away (MultiLingual, May 2026)
- HHS Section 1557 — Language Access
- DOJ Title VI Overview
- ADA.gov — Effective Communication
- Eclipse — Platform Features · Pricing · Industries: LSPs
— Shelby Edwards, Co-Founder & CXO, Eclipse Scheduling