{"resourceId":"sf-oceia-language-access-technology-pilot-2026","versions":[{"version":"external-33cbf2bf2d006145c848e723dbd9db11dee74f0eb222a757e4d4419d4b464a51","resource":{"id":"sf-oceia-language-access-technology-pilot-2026","title":"San Francisco's immigrant affairs office sets a Nov 2026 to Feb 2027 pilot of up to four AI translation and interpretation vendors with human and community review","organization":"City and County of San Francisco, Office of Civic Engagement and Immigrant Affairs (OCEIA)","sector":"City and county language access","geography":"San Francisco, California, United States","publishedAt":"Undated SF.gov info page; timeline current as of retrieval on 2026-10-29","publicationDate":null,"eventDate":null,"sourceName":"SF.gov","sourceLabel":"SF.gov: Language access technology pilot","sourceUrl":"https://www.sf.gov/language-access-technology-pilot","evidenceClass":"government-evaluation","outcomeClass":"emerging","topics":["accessibility-workforce","governance-procurement","knowledge-work"],"finding":"OCEIA will run a three-month pilot of translation and interpretation tools using only public materials in low-risk, controlled settings. City language specialists, bilingual staff and community-based organizations will review output for accuracy, cultural appropriateness and ease of use. The city plans to finalize short-term contracts with up to four vendors from its Emerging Technologies vendor pool in early November 2026, test from November 2026 to February 2027, and decide next steps in spring 2027.","sledRelevance":"Newly relevant because vendor contracts are due to be finalized in early November 2026. It is a concrete model for testing AI language access under human review before a long-term contract, using a pre-qualified vendor pool.","evidence":"Program design and timeline only; no results yet. Dates are stated as subject to change. The page gives no evaluation metrics, vendor names or budget.","architectureImplications":"Scoping to public, low-risk materials keeps resident data out of the pilot; comparing multiple vendors implies a common test corpus.","governanceImplications":"Human and community review is built into the evaluation; decisions wait for staff and community input.","securityPrivacyImplications":"Using only public materials avoids exposing personal information during testing; production use with resident data would need separate review.","caveats":"No outcomes; undated page; schedule may slip; evaluation criteria not published in detail.","streamIds":["citizen-engagement"],"roles":{"sales":"This is a buyer showing how it wants to evaluate AI translation: short contracts, several vendors, public materials only, and human and community reviewers deciding quality. Sellers should expect side-by-side tests and cultural-appropriateness review, not a single demo. The opening is evaluation support and review tooling, not a citywide rollout; results do not exist yet.","engineering":"Engineering fit depends on handling a shared test corpus of public documents and supporting reviewer workflows for scoring accuracy and cultural appropriateness. Because the pilot excludes resident data, privacy controls are not yet tested, so a later production phase will need data handling, glossary management and integration with existing interpretation services. Be ready to show per-language quality evidence.","delivery":"Delivery work is a time-boxed evaluation: a document set, reviewer rubrics, scoring by language specialists and community organizations, and a recommendation by spring 2027. Ownership sits with OCEIA. Proposed acceptance criteria should be per language and per document type, with reviewer agreement measured. This is interpretation of a planned pilot, not observed results."},"roleDetails":{"sales":{"customerProblem":"Cities must deliver consistent, high-quality translation and interpretation at volume without losing human review or community trust.","stakeholders":["Language access or immigrant affairs office","Department language liaisons","Community-based organizations","Procurement"],"discoveryQuestions":["Which document types and languages will the pilot cover?","How will reviewers score cultural appropriateness?","What evidence would justify a long-term contract?"],"valueHypothesis":"Machine translation with human review may raise consistency and speed for public materials; this pilot is designed to test that.","potentialEngagement":"Support a structured multi-vendor evaluation with a shared test corpus and reviewer rubric.","unsupportedClaims":["Any accuracy or cost results","Suitability for vital or legal documents with resident data","Replacement of human interpreters"]},"engineering":{"fit":"Fits translation and interpretation tools that can be tested on public content in controlled settings.","architectureIntegration":"Shared test corpus, reviewer scoring workflow, glossary and terminology management; later integration with interpretation scheduling.","prerequisites":["Representative public document set","Reviewer rubric","Vendor-pool contract vehicle"],"constraints":["Public materials only during pilot","Up to four vendors","Short three-month window"],"security":"No resident data in pilot; a production phase would require data-handling and retention review.","proofOfValue":["Per-language reviewer scores against human translations","Time-to-publish comparison","Reviewer agreement rates"]},"delivery":{"work":"Run a three-month multi-vendor language-technology evaluation with human and community review.","dependencies":["Vendor contracts finalized early November 2026","Availability of language specialists and community reviewers"],"ownership":"OCEIA","skillsAdoption":"Reviewers need a common rubric and training on post-editing and scoring.","governanceCheckpoints":["Contract finalization","Mid-pilot review","Spring 2027 decision with staff and community input"],"acceptanceCriteria":["Proposed: documented quality scores per language and vendor","Proposed: reviewer sign-off before any public use"],"risks":["Schedule slip","Rubric inconsistency across reviewers","Pilot results not transferable to sensitive documents"]}},"retrievedAt":"2026-10-30T03:09:17Z","enrichedAt":"2026-10-30T03:17:06Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Keeps bilingual staff and language specialists as reviewers rather than replacing them; community organizations take part in quality review.","procurementImplications":"Uses an emerging-technology vendor pool for short pilot contracts before any long-term commitment.","operatingModelImplications":"Positions AI as support to an existing language-access program, with a planned decision gate.","sourceVerification":{"openedUrl":"https://www.sf.gov/language-access-technology-pilot","referenceExcerpt":"We will test several tools using only public materials and low-risk, controlled settings.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}