Hila Segal and Roni Kandel on Learning Arc, AI Training, and Learning in the Flow of Work

Enterprise AI Modernization with Rakesh Ravuri

Rakesh Ravuri is helping enterprises modernize legacy systems with AI while preserving the context, governance, and explainability that complex organizations require. As CTO of Publicis Sapient, he leads technology for a digital transformation company helping clients evolve through each major technology shift, from the internet and e-commerce to mobile, cloud, and now AI. Publicis Sapient’s Slingshot platform recently won an AI Excellence Award for its work accelerating software development and legacy modernization. In this episode, Russ and Rakesh explore how Slingshot began as an internal AI tool after the rise of ChatGPT, then evolved into a platform for AI-assisted engineering, modernization, and enterprise transformation. Rakesh explains why Publicis Sapient first built a secure internal chat tool to protect client data, then extended it with APIs, developer plugins, and eventually a modernization workflow. They dive into the legacy technical debt problem, especially large COBOL systems that still power critical business functions in finance, healthcare, telecom, and other enterprise environments. Rakesh explains how Slingshot breaks large codebases into intelligent chunks, extracts business rules, creates specifications, generates new code, and supports modernization without relying on armies of retired COBOL experts. The conversation also covers why context is the key to useful enterprise AI. Rakesh explains Publicis Sapient’s enterprise context graph, which connects strategy, product, engineering, experience, data, code, tests, prompts, and decisions so AI can understand not just what to build, but why it matters. Along the way, Rakesh discusses AI governance, provenance, explainable code, human-in-the-loop review, deterministic testing, regulated environments, reusable enterprise prompts, agentic workflows, and why the future of AI transformation depends on capturing both enterprise knowledge and enterprise behavior. Topics Covered: [00:01] Welcome and intro, Rakesh Ravuri and Publicis Sapient’s AI Excellence Award win [00:38] Publicis Sapient’s background in digital transformation [01:43] AI as the latest transformation trigger [02:33] How Slingshot began as an internal AI tool [02:53] Building a secure internal ChatGPT-style platform [04:10] Creating APIs and early developer plugins [05:13] The legacy technical debt problem [05:52] Using AI to understand millions of lines of COBOL code [06:45] Intelligent chunking and context layers for large codebases [07:55] Moving from code to specification to new code [09:10] Whyhot’s first-principles approach outperformed brute-force code conversion [10:24] Why COBOL modernization has waited decades [13:19] What an enterprise context graph is and why it matters [15:30] Local context versus enterprise context [17:25] Why developers need the business context behind a product decision [18:14] Slingshot as a GPS for modernization [20:00] Explainability, maintainability, and code provenance [21:56] Governance for regulated industries [22:11] Measuring how much code was generated by AI [23:24] Explainable code over working code [24:08] Using context to investigate hallucinations and errors [25:43] Making expert knowledge repeatable [27:15] Building trust through proof-of-concept work [29:10] Guardrails, test cases, and deterministic evaluation [30:53] First conversations CTOs should have about legacy modernization [32:04] How Slingshot differs from coding tools like Copilot and Cursor [35:43] How AI changes teamwork across the software lifecycle [36:11] Shared prompt libraries and enterprise standards [39:56] Capturing enterprise behavior, not just enterprise data [43:59] Final thoughts on AI-driven transformation and modernization

Kenny Thompson on Blending Human Service and AI in Payments and Customer Experience

Kenny Thompson is helping BASYS prove that great customer experience can still be a competitive advantage, even in a highly commoditized payments industry. BASYS works across healthcare, banking, SaaS, distribution, manufacturing, construction materials, media, and other payment-heavy industries, while keeping a strong focus on in-house support and real human service. BASYS was recently recognized for its customer experience work and its ability to blend AI efficiency with human connection. In this episode, Russ and Kenny explore why payments are the lifeblood of so many businesses, especially for SaaS companies, banks, healthcare providers, and small business operators. Kenny explains how BASYS supports customers through complex payment workflows while helping software partners create a more seamless experience for their own users. They dive into how BASYS uses AI, chatbots, and internal support tools without losing the human touch. Kenny shares why the company still answers calls with live people, how its support teams are structured, and why its Kansas City-based model remains central to the company’s identity. The conversation also covers healthcare payment complexity, fragmented systems, customer support standards, partner integrations, Net Promoter Score, company culture, and why BASYS has chosen steady growth and long-term trust over shortcuts. Along the way, Kenny discusses community banks, SaaS partnerships, support escalation, employee hiring, customer retention, and why great service still starts with people, even when AI is helping behind the scenes. Topics Covered: [00:01] Welcome and intro, Kenny Thompson and BASYS [00:54] BASYS’ role in payments across healthcare and other industries [01:20] Why healthcare payment experiences can be clunky and frustrating [02:36] The fragmented nature of hospitals, vendors, and payment systems [03:15] BASYS’ work across payments, distribution, manufacturing, construction, and SaaS [04:19] Blending AI efficiency with live human support [05:03] Why BASYS still answers phone calls with a real person [06:00] Building an in-house support team instead of outsourcing service [07:10] How BASYS integrates payments into software platforms [08:20] Reducing the swivel chair problem in payments workflows [09:32] Why payments are mission critical for SMBs and SaaS users [10:10] Supporting small business owners who rely on payments as their revenue channel [11:27] Why many industries follow each other when technology works [12:17] Why BASYS chose Kansas City-based support over offshore service models [12:50] Tracking Net Promoter Score as a core business metric [14:00] Hiring for customer service quality and cultural fit [15:42] What happens during a typical BASYS support call [16:06] Using AI and internal chatbots to support customer service agents [17:32] Escalation from tier one to tier two support [18:06] How strong onboarding and support reduce customer problems [18:40] Why more processors do not invest this heavily in service [19:33] Maintaining support quality while growing integrations and verticals [20:30] Protecting company culture during growth [22:24] Nonnegotiables for building a service-led company [23:59] How BASYS helps SaaS partners grow revenue and prepare for exits [25:13] Why customer service can differentiate SaaS and payment platforms [26:45] Why Kenny believes human trust still matters in business [27:03] What the payments industry could learn about customer service [29:42] Final thoughts on blending humans, AI, and long-term customer care

Paul Danter on Welocalize Opal, AI Translation, and the Future of Global Content

Paul Danter is helping enterprise teams make global content faster, smarter, and more brand accurate. At Welocalize, Paul works on Opal, an AI powered platform designed to improve how companies translate, adapt, validate, and manage multilingual content at scale. Welocalize recently won an AI Excellence Award for its work helping global brands operationalize AI across language workflows. In this episode, Russ and Paul explore how enterprise localization has changed over the last 20 years, from traditional human translation to neural machine translation to AI powered post editing and quality estimation. Paul explains why translation is no longer just about converting words from one language to another. It is about preserving brand voice, tone, terminology, intent, and business impact across global markets. They dive into Opal and how it helps companies process millions of words while using AI to improve translation quality, route content through the right workflow, and determine when human review is still needed. Paul shares why different content types carry different levels of risk, and why a support article, product launch campaign, and brand tagline should not all be treated the same way. The conversation also covers how AI is creating more content than ever before, why enterprises need governance around multilingual workflows, and how continuous feedback from human linguists can help models improve over time. Along the way, Paul discusses content risk, quality scoring, brand sensitive workflows, reinforcement learning, AI governance, long tail languages, global support content, and why the future of multilingual AI may include living systems that monitor performance and automatically improve content across markets. Topics Covered: [00:01] Welcome and intro, Paul Danter and Welocalize’s AI Excellence Award win [00:53] Welocalize’s background in global language services [01:55] Why enterprise teams need language services to unlock global markets [02:20] How Opal operationalizes AI for translation and brand voice [03:10] Human translation, machine translation, and AI post editing [04:31] Training AI models to sound like a specific brand [05:21] How generative AI changed language quality and automation [07:47] Matching the right workflow to the right content type [09:14] Moving from academic quality scoring to content risk [10:42] Auditing enterprise content and connecting translation workflows [12:42] The role of AI in support content and customer experience [13:20] Why AI governance matters as content volume explodes [14:57] What companies underestimate about multilingual content operations [16:59] How Welocalize measures whether Opal is working [17:25] Faster turnaround times and reduced human editing effort [18:40] What early Opal deployments revealed [20:58] Building trust with enterprise content teams [21:29] Quality testing, certified languages, and human validation [23:40] Why quality estimation matters before human review [25:01] Continuous editing, feedback loops, and model improvement [25:58] Lessons other industries can learn from language AI [28:13] What multilingual AI could look like in five years [29:20] Improving source content before translation begins [30:00] Using performance data to improve localized marketing content [31:25] Advice for founders building AI in brand sensitive workflows [33:05] Language as part of closing the global digital divide [33:40] Final thoughts on Opal, AI, and the Welocalize team

Hila Segal and Roni Kandel on Learning Arc, AI Training, and Learning in the Flow of Work

Hila Segal and Roni Kandel are helping enterprises rethink how employees learn, practice, and apply new skills inside the tools they use every day. At WalkMe, they are building Learning Arc, an AI native learning solution designed to close the gap between training content and real work. WalkMe recently won an award for its work helping organizations deliver learning at the moment employees need it most. In this episode, Russ, Hila, and Roni explore why traditional corporate training often fails to translate into execution. Hila explains how WalkMe’s digital adoption platform has spent years helping employees navigate software workflows, and why customers began asking for a deeper learning experience that could build lasting proficiency, not just guide clicks. They dive into Learning Arc and how it combines AI powered content creation with in app learning delivery. Roni shares how the product helps organizations train employees before they enter sensitive systems, then resurfaces the same learning content when they are actually doing the work. The conversation also covers why learning and software adoption have historically been disconnected, how AI can help L&D teams create and refresh content faster, and why human oversight remains essential. Hila and Roni explain how Learning Arc gives authors visibility into what AI created, what source material was used, and how learning content can be refined to fit business needs. Along the way, they discuss sales training, ERP rollouts, AI literacy, multimodal learning, personalization, learner choice, enterprise software development, and why the future of workplace learning will be more contextual, flexible, and embedded directly into the flow of work. Topics Covered: [00:01] Welcome and intro, Hila Segal, Roni Kandel, WalkMe, and Learning Arc [00:45] WalkMe’s role in digital adoption and enterprise software usage [02:00] Why customers needed deeper skills and lasting proficiency [02:30] What WalkMe Learning Arc is designed to solve [03:36] Moving from traditional training to learning in the flow of work [04:26] Knowing versus doing in sales methodology and workflows [06:12] Why Learning Arc became a standalone product [06:39] Training before access to sensitive systems and processes [07:32] Why the learning and execution gap has lasted so long [08:20] Connecting L&D outcomes to actual work performance [09:25] Why the pace of workplace change is outgrowing traditional training [10:20] Using AI to create and refresh learning content faster [11:48] Early customer use cases and major transformation programs [12:50] Using Learning Arc to support ERP rollouts and onboarding [13:54] Building trust with L&D teams [14:14] How AI can turn months of content work into minutes [15:30] What learning in the flow of work means inside Learning Arc [16:34] Using screen context to surface relevant learning content [17:00] AI literacy training and responsible AI tool access [18:15] Reimagining the learning portal experience [19:50] Giving learners flexible formats like audio, video, and text [21:00] Why different employees need different learning experiences [21:44] Why human oversight remains central to AI generated learning [23:21] How authors control, review, and refine AI created content [24:30] Preventing hallucinations and grounding AI in approved source material [26:19] Balancing personalization and scalability [28:53] What Learning Arc signals about the future of enterprise software [30:45] How AI may change product, development, and software team roles [31:30] Training employees on tools that change constantly [33:14] Why change is central to WalkMe’s mission [35:56] Final thoughts on Learning Arc and the future of workplace training

Robert Klamser on Using AI to Make Chapter 11 Bankruptcy Easier to Understand

Robert Klamser is helping make one of the most complex corners of the legal system easier to navigate. As the leader of Stretto’s AI focused innovation efforts, Robert works on technology that supports professionals, creditors, vendors, employees, courts, and other stakeholders involved in bankruptcy cases. Stretto recently won an AI Excellence Award for Conductor, its AI powered platform built to make Chapter 11 information more accessible, understandable, and useful. In this episode, Russ and Robert explore why bankruptcy is such a document heavy, high stakes, and often confusing process. Robert explains how large Chapter 11 cases can involve thousands of pages of filings, multiple jurisdictions, local rules, federal bankruptcy code, and millions of affected stakeholders who may have never encountered bankruptcy before. They dive into Stretto Conductor and how it helps users ask plain language questions about complex bankruptcy documents. Robert shares how the platform is designed to understand filings in the context of the specific case, the related docket history, the bankruptcy code, and the local rules that may shape the answer. The conversation also covers why general purpose AI is not enough for this kind of work. Robert explains the importance of domain specific AI, grounded answers, citations, legal precision, multilingual access, and making sure the platform provides information without crossing into legal advice. Along the way, Robert discusses creditor communications, call center operations, hallucination concerns, attorney trust, AI adoption in the legal industry, and why the future of legal technology will likely depend on purpose built systems that do one thing extremely well. Topics Covered: [00:01] Welcome and intro, Robert Klamser and Stretto’s AI Excellence Award win [00:17] Stretto’s role in bankruptcy support services and technology [02:11] The communication challenges inside large Chapter 11 cases [03:00] How Conductor helps vendors, creditors, and lawyers understand filings [04:34] What is broken about how information flows in bankruptcy cases [06:31] Why basic case information can be hard for stakeholders to access [07:00] How bankruptcy case websites changed access to documents [09:18] Why documents still need context from the bankruptcy code and local rules [10:54] What Stretto Conductor does differently [11:12] Teaching AI the rules, nuance, and structure of bankruptcy [13:31] Why legal AI must be grounded, intelligent, and precise [15:37] Why purpose built AI matters in bankruptcy law [17:02] Why one document rarely tells the full story in a Chapter 11 case [19:20] Naive retrieval, missing context, and reasoning errors in general AI tools [21:01] Building trust with skeptical legal professionals [21:21] Why every answer must be cited and grounded in the right source [22:38] How users are learning to trust Conductor [23:00] Why Conductor answers questions instead of drafting legal documents [24:21] How large bankruptcies create massive temporary operating structures [24:55] Supporting bankrupt entities while the business keeps operating [27:08] Making newly filed documents understandable within minutes [28:00] Multilingual access for stakeholders in global bankruptcy cases [28:54] How Conductor can reduce pressure on large call centers [31:45] Why lawyers have traditionally been slow to adopt new technology [32:27] How courts, caution, and hallucination concerns affect AI adoption [35:53] Where legal AI may be heading over the next five years [37:00] The need for clearer standards around AI use in legal work [38:25] Why clients may soon expect attorneys to use AI efficiently [39:58] Final thoughts on making bankruptcy easier to understand