Maximilian Eber is a co-founder and Chief Product & Technology Officer of Taktile, a platform for AI-driven decision automation in financial services [1]. He holds a Ph.D. from Harvard University and took Taktile through Y Combinator's Summer 2020 batch, having co-founded the company in March 2020 after several years working on machine learning and causal inference at QuantCo [1]. His professional presence is documented in part through Taktile's video content and his personal profile on X [2][3][4].
As part of his role at Taktile, Eber hosts a video series examining AI adoption in financial services, through which he engages industry practitioners on questions of credit risk and underwriting [3]. In one such conversation with a Chime product director, Eber probed how the company reconciles product growth with lending risk, drawing out the point that in credit products the constraint is not user acquisition but repayment, and that Chime's strategy centers on cultivating primary account relationships to obtain continuous cash-flow visibility rather than the single-point-in-time snapshots typical of bureau data [3]. The exchange also explored how a primary banking relationship allows a lender to align repayment timing with a customer's actual deposit schedule, rather than relying on third-party pulls of uncertain timing [3].
Eber has articulated views on the near-term trajectory of artificial intelligence relevant to financial services applications. He has argued that solving the problem of model hallucination would open up use cases requiring very high reliability, on the order of 99.9 percent accuracy, which current systems cannot support [4]. He has also pointed to reductions in inference cost and latency as a separate, and likely, development that would make real-time applications feasible, citing credit card fraud detection at the point of a transaction as an example of a use case presently out of reach because there is insufficient time to call a model during the authorization window [4].
Founded
Insights & ideas
Maximilian Eber's recurring theme is that reliability, not raw capability, is what determines whether AI can be trusted in financial services. He argues that eradicating hallucinations would unlock an entire class of use cases currently blocked because they demand 99.9% accuracy rather than 98%, and he sees active research on why models hallucinate as a promising path toward that threshold [2]. He also emphasizes that falling inference cost and latency will be a second major unlock, since time-sensitive scenarios like credit card fraud detection at the moment of a swipe are currently out of reach for AI but become attackable once models get faster and cheaper [2].
In discussing credit products with industry guests, Eber frames the central challenge of fintech underwriting as aligning product, risk, and marketing around a shared definition of lifetime value, since cheap-to-acquire leads are often the ones risk teams reject, and only by capturing repayment behavior in that value definition can the functions work together [1].
Education
Harvard University · Doctor of Philosophy (Ph.D.)2012 - 2016
Y Combinator · S202020 - 2020
Media & appearances
- Chime’s AI-Driven Credit Strategy ft. Baishi Wu - YouTubeYouTube
Maximilian Eber, as host and co-founder of Taktile, interviews Baishi Wu about Chime's credit strategy. The discussion covers how Chime aligns product and risk management, the role of primary account relationships in underwriting, and how AI and automation can improve compliance and policy development in financial services.
- What will the next generation of AI look like? - YouTubeYouTube
Maximilian Eber discusses two key developments he expects in AI: solving hallucinations to enable high-reliability use cases requiring 99.9% accuracy, and reducing inference cost and time to make real-time applications like credit card fraud detection feasible for AI systems.
- 201303-25 005 Conficamp Interview Maximilian EbertKonfiweb
This page shows public professional information only, each fact cited. Is this you? Corrections or removal within 24 hours, no questions asked.