Boostlingo has moved its AI Interpreter into general availability for 11 languages after a beta program that handled hundreds of thousands of minutes. The consequential feature for language-service providers is not the synthetic voice. It is the ability to decide, language by language, when an automated session may be offered and when a human interpreter remains the route of choice.

General availability arrives with language-level controls

Boostlingo announced general availability on August 3 for Arabic, Cantonese, Dari, French, Haitian Creole, Korean, Mandarin, Portuguese, Russian, Spanish and Vietnamese. Thirteen additional languages remain in beta. The service is available through mobile and web applications, telephone access, SIP connections and software development kits for iOS and Android. Boostlingo says it is live in the United States, Canada, Australia, the European Union and the United Kingdom.

For LSP administrators, the platform supports routing policies that determine which languages can be handled by AI. That makes deployment a governance decision rather than a universal switch. An agency could enable automated interpreting for a high-volume, lower-risk use case while retaining human-only routing for another language, client or context.

Quality claims require context

Boostlingo says its system outperformed a leading open-source model by 17 percent in accuracy testing. The company has not identified the comparison model or published a complete independent methodology in its announcement, so the figure should be treated as a vendor-reported result, not a market benchmark.

Its more useful contribution may be Assure, a quality framework that combines automated testing with human review of staged medical conversations. The evaluation covers more than 40 submetrics across accuracy, professionalism, conversational flow and technical quality. This is a more credible direction than relying on one aggregate score, because interpreting performance can fail through omissions, turn-taking problems, latency or an inability to manage uncertainty even when much of the language is correct.

Buyers should still ask how results vary by language pair, accent, audio quality, domain and speaking style. They should also ask how the platform identifies a failed interaction, how a user reaches a human interpreter, what is retained, and whether the service is approved for the buyer’s intended clinical, legal or public-service setting.

The market is interested, but adoption is early

Boostlingo’s 2026 State of Interpreting Technology survey gathered responses from more than 370 stakeholders. The company reports that 16.8 percent were using AI interpreting and another 15.9 percent were evaluating it. At the same time, 46.8 percent said on-site interpreting remained their standard choice for high-risk encounters.

The same survey found that organizations use an average of 2.3 interpreting modalities. Phone and video were the most common pairing. Half of respondents reported at least one encounter with a limited-English-proficient person in which an interpreter was needed but unavailable. Inconsistent quality was described as having a moderate or severe operational impact by 53.8 percent.

These findings come from a vendor-sponsored survey and should be read as directional rather than representative of the entire North American market. They nevertheless show why automated interpreting is gaining attention. Availability, cost and speed are persistent problems, but high-risk work continues to demand a stronger level of assurance.

A practical test for LSPs

The immediate commercial opportunity is not to label AI interpreting either acceptable or unacceptable. It is to design a defensible service policy. That policy should define eligible scenarios, language-level evidence, escalation paths, client consent, data handling and ongoing review. Agencies that can explain those controls may be better positioned than providers selling automation as an undifferentiated replacement for human work.

Sources