# 215 Humanless LSP as a Fun Weekend Project

Introduction and Catching Up

In this section, the hosts introduce the podcast and discuss their recent experiences at events.

Hosts' Recent Experiences at Events

• The hosts reflect on attending Slaton London, highlighting the conference's positive aspects such as great speakers and relevant content.

• They mention upcoming events like Slaton California Meno Park on September 5th and encourage registration for Slater con remote.

Discussion on Industry Topics

This part covers various industry topics discussed in upcoming conferences.

Industry Topics at Conferences

• The agenda for Slater con remote includes discussions on new ISO standards, language shops in the age of AI, preparing for right-to-left projects, and more.

• Intento and e2f will present their State of Mt report at Slaton remote, focusing on machine translation engines and LLMs.

PR Guide on Language AI for Consumers

The hosts talk about a PR guide focusing on consumer-level AI applications impacting B2B client expectations.

Consumer-Level AI Applications

• The guide discusses how consumer-level AI applications influence B2B client expectations, providing 10 use cases like website translation and transcribing podcasts.

• Subscribers can access the guide freely or purchase it for $250 to explore different consumer use cases of language AI.

Andrew Ang's Project & Industry Updates

Andrew Ang's project on agentic machine translation is discussed along with updates from Apple, RWS results, CEO changes, deals news, and mergers/acquisitions.

Agentic Machine Translation & Industry Updates

• Andrew Ang explores agentic machine translation as a virtual LSP concept using different LLMS for tasks to enhance traditional neural machine translation methods.

An Open Source Demonstration

In this section, the speaker discusses an open-source demonstration project that involves using an agentic workflow for translation and reflects on its performance compared to leading commercial providers.

Agentic Workflow for Translation

• The demonstration involves prompting a language model (LLM) to translate between languages, providing constructive suggestions for refinement.

• The software is considered immature, with hopes that the open-source community can enhance agentic translation capabilities.

• Reference to a Chinese tech company utilizing virtual LLM-based language service providers (LSPs) with various roles like project managers.

Reactions and Discussions

This part delves into the reactions and discussions surrounding the concept of agentic machine translation within the language industry and AI communities.

Industry Reactions

• Varied responses from individuals in the language industry and enterprises regarding the potential of agentic machine translation.

• Comments range from revolutionary praise to concerns about cost implications and questioning the necessity of an agentic approach.

Expert Perspectives on Agentic Translation

Insights are shared regarding leveraging new models in localization workflows, contrasting perspectives from different experts in AI and localization industries.

Leveraging New Models

• Contrasting viewpoints between experts familiar with localization workflows (e.g., Phase representative Gar) and those more focused on foundational AI technologies (e.g., Andrew).

• Discussion on understanding nuances in localization workflows beyond AI comprehension, highlighting challenges in implementing detailed tasks through LLM prompts.

Evaluation of Agentic Machine Translation

Evaluation of agentic machine translation as a potential innovation or experimental endeavor within the realm of translation technology.

Innovation Assessment

• Contemplation on whether agentic machine translation represents a revolutionary advancement or merely a playful weekend project.

Detailed Discussion on App Localization and RWS Financial Results

In this section, the discussion revolves around app localization using machine translation APIs and an analysis of RWS's financial results.

App Localization and Translation

• The conversation delves into translating system UI and longer text batches through app UI. There is a query about whether developers should use a machine translation API or integrate Apple's translation API for localizing app UI.

• Uncertainty persists regarding utilizing a machine translation API or integrating Apple's translation API for batch translations. Speculation leans towards the latter, indicating that app localization may not be the primary focus of a translation API.

RWS Financial Results Analysis

• RWS published their half-year results with shares up by 10%. A brief overview of the revenue decline by 4% to £350 million compared to the previous year, along with a 16% decrease in adjusted profit before tax.

• Despite revenue and profit declines, investors seem positive, leading to a stock increase. The discussion touches upon top-line numbers such as revenue decrease and profit fall.

• The stock's rise is attributed to traction gained from new AI service offerings by RWS. Approximately 25% of service revenues stem from AI-centric solutions like training AI and post-editing machine translations.

RWS Product Offerings and Market Segments Analysis

This section focuses on RWS's product offerings, including linguistic AI solutions like Evolve and digital self-service platforms like Hi.

RWS Product Offerings

• RWS introduces new products: Evolve, described as a linguistic AI solution for enterprises; Hi, labeled as a digital self-service platform aimed at small to medium-sized businesses. These offerings contribute to RWS gaining traction in the market.

Market Segments Analysis

• Insights into RWS divisions reveal growth in Language Services and IP Services while Language & Content Technology Revenue contracts but sees an increase in SAS license Revenue. Challenges persist in regulated industries like life sciences and financial services.

Outlook and Strategic Initiatives of RWS

This segment discusses the outlook for RWS, anticipating stronger performance driven by new business wins, recovery in high-margin segments, ongoing investments in AI, strategic initiatives supporting long-term growth despite macroeconomic uncertainties.

Outlook Highlights

• Expectations are optimistic for a robust second half with anticipated growth fueled by new business wins, margin recoveries, ongoing investments in AI technologies alongside strategic initiatives supporting long-term expansion amidst macroeconomic uncertainties. Confidence prevails despite potential market impacts.

Transition Management at RWS

Delving into challenges faced by companies transitioning from traditional profitable services to newer AI-driven solutions while managing investor expectations positively due to effective transition strategies implemented by companies like RWS.

Transition Management Insights

DeepL Funding and Valuation

The discussion revolves around DeepL's recent funding round, the valuation of the company, and its plans for expansion.

DeepL's Funding Round and Valuation

• DeepL raised $300 million in funding with a valuation of $2 billion, achieving Unicorn status.

• Dubbed as a "double unicorn," sparking humor about having two horns or heads.

• Index Ventures led the investment round, joined by other investors like Iconic Growth and existing investors VP Atomico and W.

• Funds will fuel DeepL's expansion into the US, focusing on AI models like translation.

• Despite limited financial details, DeepL boasts 100,000+ customers across 32 languages.

Unbubble Tower llm Launch

Unbubble introduces Tower llm with claims of superior machine translation quality compared to industry giants.

Unbubble Tower llm Announcement

• Unbubble launches Tower llm offering top-tier machine translation trained on high-quality multilingual data.

• Claims superiority over GPT models and Google Translate in press release, igniting competition in the AI translation space.

Keywords Studio Potential Sale

Keywords Studio considers a significant sale following an offer from a private equity firm.

Keywords Studio Sale Consideration

• Keywords Studio contemplates a sale to a Swedish private equity firm for £2.2 billion or approximately $2.8 billion USD.

Private Equity and Listed Companies

The discussion revolves around the contrast between private equity companies and listed companies in the language industry, highlighting the benefits of transparency and financial insights provided by listed companies.

Private Equity vs. Listed Companies

• Private equity ownership is sarcastically mentioned, emphasizing the appreciation for listed companies that offer regular information and disclosures.

• Contrasts are drawn between private equity firms being closed boxes versus listed companies allowing access to financial details.

• Mention of turbulent times in media localization in 2023 due to strikes in Hollywood, with Zoo's positive financial results indicating a demand resurgence.

Financial Performance of Language Service Providers

This segment delves into the financial performance of language service providers, focusing on specific companies like Zoo Italia and AMN Healthcare Interpretation Services.

Zoo Italia and AMN Healthcare Interpretation Services

• Zoo Italia experienced a sharp uptick in March 2024 revenue, signaling a recovery in media localization demand.

• AMN Healthcare Interpretation Services saw significant growth with a 16% increase year-on-year in language services business.

• Despite flat revenues overall, AMN's language services segment performed exceptionally well, prompting additional investments towards key areas of growth.

Impact of Managed Services on Revenue

The focus shifts to managed services within the language industry and its impact on revenue generation for businesses.

Managed Services Influence

• Managed services designed to manage internal translation processes reflect an industry trend towards cost-efficient translation management.