DiscussionOpinion

World's #1 Banking CMO: How AI Is Changing Billion Dollar Banks Today | Tim Rutten, CMO @ Backbase

Topline1h 22m

Tim Rutten, CMO of Backbase, discusses how AI is transforming banking operations and go-to-market functions, achieving a 2x pipeline increase with 25% less budget through deterministic AI systems. The conversation explores AI's real limitations in highly regulated industries, the nascent state of AI monetization, and the emerging importance of judgment and taste in an AI-saturated digital landscape.

Summary

Tim Rutten, CMO of the 2.5 billion euro fintech Backbase (powering 120+ banks globally), joins the Top Line podcast to discuss the practical implementation of AI in banking and marketing. The discussion begins with the fundamental challenge facing AI adoption in regulated industries: error tolerance. Highly regulated industries like banking cannot accept 3-5% error rates, forcing implementations to remain heavily guardrailed and mostly deterministic rather than fully agentic. Backbase's approach involves applying AI to only 1% of workflows where reasoning is acceptable, while keeping the remaining 99% deterministic.

The conversation establishes that internal operational use cases—particularly customer care center automation achieving 90% efficiency gains—are driving near-term ROI, while customer-facing AI implementations remain early and limited. Rutten notes that European banks are roughly 10 years ahead of US institutions in digital banking maturity, though both regions face regulatory uncertainty that slows innovation. He emphasizes that banks are shifting budgets from traditional digital transformation initiatives toward AI-native rewrites, representing a significant funding migration.

Rutten shares his team's concrete success with GTMOS, an internally-built AI-native marketing operating system built on Vercel's stack that integrates Salesforce, finance tools, and corporate data. This system has enabled a 2x pipeline increase while reducing his budget by 25%, primarily through workflow automation and agentic capabilities that eliminate manual keyboard work. However, he stresses this represents new capabilities rather than simple labor replacement.

The hosts raise critical concerns about the AI bubble. Sam Jacobs argues that for the massive capital expenditures on AI to be justified, AI must deliver returns across multiple industries beyond the three proven use cases: chat, code, and support. The temporal disconnect between capex spend and revenue generation is identified as the primary bubble risk. Additionally, the business model of frontier labs selling tokens faces existential challenges from open-source and open-weight models, which commoditize intelligence and undermine pricing power.

Rutten counters that sufficient model intelligence already exists for banking; the real needs are smaller, more efficient models with better guardrails and standardized architecture. He references Nicholas Carr's "IT Doesn't Matter" thesis to suggest AI will eventually become commoditized infrastructure like electricity. The discussion explores whether on-prem deployment and model neutrality (banks using multiple models for different workloads) represents the future.

A significant portion of the conversation addresses AI quality and the erosion of judgment. Jacobs expresses concern about ubiquitous "AI slop"—generic Claude artifacts with default styling, generic LinkedIn posts, and low-quality work that floods digital channels because creators lack taste or judgment to refine AI outputs. He worries that without foundational skills, people using AI produce noticeably poor work. Rutten addresses this by describing GTMOS's 22-point quality framework that constrains outputs and prevents slop through deterministic rules and continuous rewriting agents.

The hosts debate whether AI democratizes capability (like Ikea furniture) or debases quality. They settle on the idea that AI accelerates existing talent rather than leveling skill gaps—skilled workers using AI tools produce exceptional work, while unskilled workers produce obvious slop. The conversation notes that anti-AI sentiment is growing, particularly on LinkedIn, while creators are migrating to X and Substack.

For the future, Rutten expresses excitement about real-time marketing pipelines that can generate enterprise-grade landing pages, ABM campaigns, and interactive demos within hours—currently impossible at scale without massive headcount increases. He also predicts autonomous workloads for research and intelligence gathering that operate continuously without human interface.

The bulls vs. bears segment reveals Jacobs is deeply bearish on LinkedIn as a content platform, citing organizational dysfunction, algorithmic opacity, and loss of reach for creators. He's migrating to X for thoughtful discourse and Substack for audience building. Rutten remains bullish on mobile banking apps as the dominant form factor for two more years, with conversational and voice interfaces emerging gradually rather than revolutionary form factor changes. AJ Bruno is bullish on companies using AI to accelerate financial planning, which would enable more sophisticated and frequent incentive plan adjustments.

About this episode

<p>Tim Rutten, CMO of Backbase, the €2.5 billion Amsterdam tech company whose AI-native Banking OS powers 120+ banks worldwide, joins Sam Jacobs, AJ Bruno, and Asad Zaman with the clearest AI ROI numbers we have heard on the show: double the pipeline on 25% less budget. Topics include the custom go-to-market operating system his team built on the Vercel stack, why real bank deployments are 99% deterministic, the 90% cost reduction banks are finding in customer care, and why most enterprise workloads do not need frontier-model intelligence. Plus, where the AI bubble breaks, what happens to judgment and taste when anyone can generate output, and a bulls-versus-bears round on LinkedIn, banking apps, and sales spiffs.</p> <p>Key Takeaways:</p> <p>- Backbase's marketing org runs on GTMOS, a custom go-to-market operating system built on the Vercel stack and wired into Salesforce and the company's finance tools, and the numbers behind it are specific. As Tim Rutten, CMO at Backbase, put it: "Double the pipeline, 25% head cut on my budget... So with 25% less budget, I'm actually still doubling the pipeline." The unlock came from rebuilding how the team works rather than layering tools on top of it.</p> <p>- Shipping AI into a regulated industry means using AI more narrowly than you might expect. On what real bank deployments look like, Tim Rutten said: "the actual implementations are almost 99% deterministic, and there's this one little percent that actually gets them to gain, to reason at the right moment in time." That same architecture caps the bill, because most banking workloads are repetitive and rule-bound. But the same rigor can net fantastic results from AI in broader applications as well.</p> <p>- Backbase treats AI slop as an architecture problem, containing content generation inside hard boundaries. Per Tim Rutten: "You have tone of voice, you have guardrails that we don't have AI slop. There's like 22 rules that are applied. If they don't check out, the agent needs to rewrite and continuously learns to rewrite." The human half of the risk is harder to engineer around, as Asad Zaman, CEO of Sales Talent Agency, framed it: "a lot of people that didn't have a skill are using AI for that thing. And if you didn't have a skill in an area or judgment or taste in that particular area, you have no mechanism to stress test the quality of what is coming out."</p> <p>- On how the AI bubble pops, Sam Jacobs, CEO of Pavilion, put the risk on the business model sitting underneath the capex: "if everybody presumes the way that all these companies make money is selling tokens… and then all of a sudden it turns out selling tokens is a really terrible business. Well, then that's a thing that would compress." Tim Rutten lands in a similar place on timing while staying long on the outcome: "a little bit bearish on timing, massively bullish on the future… it will happen nevertheless, but it's going to be a very rocky ride."</p> <p>Connect with the Hosts & Guests:</p> <p>Host: Sam Jacobs, CEO at Pavilion - https://www.linkedin.com/in/samfjacobs/<br /> Host: AJ Bruno, CEO at QuotaPath - https://www.linkedin.com/in/ajbruno3/<br /> Host: Asad Zaman, CEO at STA - https://www.linkedin.com/in/azaman1/<br /> Guest: Tim Rutten, CMO at Backbase - https://www.linkedin.com/in/timrutten/</p> <p>Topline is more than a YouTube Channel:</p> <p>Subscribe to Topline Newsletter: https://toplinemedia.substack.com/<br /> Tune into Topline Podcast, the #1 podcast for founders, operators, and investors in B2B tech: https://www.joinpavilion.com/topline-podcast<br /> Join the free Topline Slack channel to connect with 600+ revenue leaders to keep the conversation going beyond the podcast: https://www.joinpavilion.com/topline-slack</p> <p>Chapters:<br /> 00:00 Introducing Tim Rutten<br /> 03:11 Inside Backbase And 120 Banks<br /> 09:42 90% Cost Cuts In Call Centers<br /> 12:04 Why The Application Layer Wins<br /> 15:23 Europe, The US, And Regulation<br /> 21:50 Where The AI Bubble Breaks<br /> 27:24 Why Banks Skip Frontier Models<br /> 34:34 Bank Budgets Shift To AI Native<br /> 38:23 Real ROI At Lloyds And JPMorgan<br /> 44:31 Quiz Pro Quo<br /> 49:37 GTMOS, The Go To Market OS<br /> 52:02 Double Pipeline On 25% Less Budget<br /> 53:53 Judgment, Taste, And AI Slop<br /> 1:06:15 Real Time Marketing By Christmas<br /> 1:09:30 Bulls and Bears</p>

Key Insights

  • Banks achieve 90% efficiency gains in customer care centers using AI, but these are internal operational use cases, not customer-facing product innovations.
  • Successful AI implementations in banking are 99% deterministic with guardrails, not probabilistic reasoning—applying full LLM capability across workflows would produce unacceptable error rates.
  • Tim Rutten doubled his marketing pipeline while reducing budget by 25% through GTMOS, an internally-built AI system with 22 quality rules that prevent AI slop through deterministic constraints.
  • Sam Jacobs identifies a critical temporal disconnect: massive AI capex spending is not matched by corresponding revenue growth, creating bubble conditions that require new use cases beyond chat, code, and support.
  • Frontier lab business models (selling expensive tokens) face existential risk from open-source and open-weight models that commoditize intelligence, potentially collapsing the equity valuations built on token sales.
  • European banks are approximately 10 years ahead of US banks in digital maturity, yet both regions face regulatory uncertainty that actually slows AI adoption rather than enabling it.
  • AI accelerates existing talent and taste rather than democratizing capability—skilled writers using Claude produce exceptional work while unskilled users produce obvious slop that lacks judgment.
  • LinkedIn has become the B2B social platform where the least interesting business conversations occur due to algorithmic dysfunction and organizational misalignment, driving creators to X and Substack.
  • Banks will never choose a single AI model; they require model neutrality to use expensive frontier models for high-intelligence tasks and cheaper models for deterministic, repetitive work.
  • The sufficient intelligence for banking already exists today—the real needs are smaller, more efficient models with better infrastructure and guardrails, not fundamentally smarter models.
  • Nicholas Carr's thesis that IT became commoditized as electricity applies to AI: eventually open-source and distributed models will make AI a commodity infrastructure layer rather than differentiator.
  • Real-time marketing pipelines generating enterprise-grade landing pages and ABM campaigns within hours are currently impossible without massive headcount increases, representing the next marketing frontier.

Topics

AI implementation in regulated industries (banking)Error tolerance and guardrailing in high-stakes applicationsBackbase's GTMOS marketing operating systemThe AI bubble and misalignment between capex and revenueOpen-source models vs. frontier lab business modelsAI quality and the importance of judgment/tasteLinkedIn vs. X as content platformsFuture banking form factors and interaction modelsFinancial planning and incentive compensationOperational automation ROI in bankingEuropean vs. US banking digital maturityModel commoditization and electricity analogy

Transcript

Tim Rutten is the CMO at Backbase, 2.5 billion euro fintech powering over 120 banks worldwide. And he's one of the rare go-to-market executives getting massive measurable ROI from AI tooling. Workloads are truly either automated, some of them are agentic, you don't even touch the keyboard anymore. But with 25% less budget, I'm actually still doubling the pipeline. In today's episode, we get Tim to go deep on the AI tooling driving those results. But we don't only focus on what's working. Tim also shares where most organizations are falling short with AI. You don't need Fable. It's way too expensive. You don't need that level of intelligence. You could actually probably do better with ***. Plus, we discuss…

Full transcript available for MurmurCast members

Sign Up to Access

More from Topline

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.