LILT AI

Good food. Better conversation. That's how we'd sum up our AI Circle roundtable in NYC with a room full of frontier researchers and builders. Spence's talk on AI benchmarking is worth the read: https://lnkd.in/etEq64S5. Don't miss the next AI Circle X LILT event. AI Circle Albert Chun Ejaaz Ahamadeen Heather Sullivan Morgan Gaybrick

What a week at AI RAISE Paris, which wrapped up yesterday! From a champagne happy hour to an exclusive dinner cohosted by AI Circle and joined by top researchers from leading AI labs, it was one to remember. Our CEO Spence Green led a discussion on AI benchmarking across languages and how LILT is redefining what it means to benchmark agentic performance at the frontier. Heather Sullivan AI Circle Albert Chun Gabrielle Halady Kelly Mapley Louis Cooper #airaise #raise

🚀 We're excited to announce the LILT July 2026 Product Release! This month's release brings greater financial transparency to help you track spending with confidence, plus three powerful new content connectors designed to automate your workflows and accelerate global delivery. ✨ What’s New This Month: 🔹 Enhanced Budget Tracking & Quoting: Get a clearer picture of your multilingual content budget! Dashboards now display in-progress job costs alongside invoiced data, and our quoting engine includes built-in file filters for precise quotes on complex documents. 🔹 NEW Amplience Connector (Beta): Seamlessly translate all your Amplience content—from web pages to promotional banners. Automate source-to-delivery with instant or expert-verified workflows in just a few clicks. 🔹 NEW Inriver Connector (Beta): Eliminate manual exports and reimports! Easily translate product-related content directly within Inriver to launch global campaigns faster and with less manual overhead. 🔹 AEM Connector Enhancements (Beta): You can now translate document assets stored in your AEM Digital Asset Manager (DAM)—including Word, Excel, PowerPoint, text, and HTML files—using the exact same seamless interface as your webpage submissions. Ready to see how LILT can simplify your translations with AI? Dive into the full release details on our blog! 👇 🔗 Read more here: https://lnkd.in/egn4U6BA #LILT #ProductRelease #Localization #Translation #AI #WorkflowAutomation #Amplience #Inriver #AdobeExperienceManager #SaaS #BudgetTracking

Final few spots at our Breakfast next week! Rob Cross will be exploring the surprisingly powerful parallels between music history and the art of localisation. Whether you're in localisation, running a global product, or leading global marketing this room is not one to miss! Join us at the beautiful Aram ܐܪܡ— - A name routed in language history and home in Somerset House by chef Imad Alarnab. ⏰ 8:30am - 10:30am. Request a spot here: https://luma.com/v17ydpyi

Exciting news: our GAIA paper (GAIA-v2-LILT: Multilingual Adaptation of Agent Benchmark beyond Translation) was accepted at the MeLLM Workshop co-located with ACL 2026 in San Diego (July 2–7). The results showed that a rigorous human-led audit of machine-translated GAIA raised agent success by up to +32.7% and shrank the gap to English to as little as 3.1%, proving that much of the multilingual performance gap was actually a benchmark measurement error. We found that functional and cultural alignment errors had the largest impact on agent outcomes. Congratulations to our research team Yunsu Kim, Kaden Uhlig, and Joern Wuebker on these impressive findings.

Paris, July 8th | RAISE Summit - An invitation only evening at the frontier of multilingual AI. As AI shifts from static chatbots to autonomous agents reasoning and acting across multiple steps, we've hit a wall few are talking about: the English-Centric Evaluation Gap. Most multilingual benchmarks today are just translated English datasets, and that translation introduces noise, hallucinations, and "translationese" that can break even the most capable agents. A benchmark can be grammatically flawless and still functionally broken. That's the conversation we're bringing to Paris. AI Circle x LILT AI present: Roundtable Dinner: AI Benchmarking Across Languages An intimate fireside with Spence Green, CEO of LILT, alongside the builders deploying at the frontier. On the table: - The "Fluent yet Broken" Paradox — when perfect grammar hides functional failure - GAIA-v2-LILT — how re-auditing GAIA recovered +20.7 points of "lost" performance, revealing capability gaps that were really measurement gaps - Terminal-Bench & τ³-bench — evaluating agentic coding and multi-turn support in non-English environments -Functional & cultural alignment — what actually breaks when you take English benchmarks global Private Location | Paris, FR July 8th · 7:00 PM Exclusive, RSVP's will be carefully curated. This is the room to be in. RSVP to be considered for an invitation to join us at the table: https://luma.com/7lfhp36n #AI #MachineLearning #AIBenchmarking #MultilingualAI #AIAgents #LLM #RAISESummit #AICircle #LILT

A few weeks ago, we brought together some of the top AI Labs at Oceans in NYC, in partnership with AI Circle. Heather Sullivan and Albert Chun spoke candidly about where LILT is heading in the AI space. Worth a watch.

The countdown to Raise in Paris is on! Come and chat with us at Booth S8, meet LILT CEO, Spence Green and the rest of the team. Be sure to swing by the booth for Champagne Happy Hour on the 8th with AI Circle and Albert Chun. Sign up here: https://luma.com/afxyghon Learn how we: ➡️ Integrate human-governed judgment into your existing pipelines for superior model readiness. ➡️ Provide the precise data needed to deploy sophisticated, localised AI - safely and at scale. ➡️ Combine Adaptive AI with human verification for accurate multilingual content to improve quality and accelerate your time to market Kelly Mapley Louis Cooper Heather Sullivan