NVIDIA AI Podcast – Episode 263

Alembic and the Future of AI in Marketing
Guest: Thomas Puig (Founder & CEO, Alembic)
Host: Noah Kravitz
Date: July 2, 2025

Episode Overview

In this episode, host Noah Kravitz sits down with Thomas Puig, founder and CEO of Alembic, to explore the transformative role of AI and advanced analytics in marketing. Puig shares the origin story of Alembic, discusses the unique technical innovations powering their marketing intelligence platform, and analyzes the changing landscape of data, creativity, and brand engagement in the AI era. The conversation journeys from technical deep dives into signal processing and causal inference, to broader philosophical and practical questions about human creativity, personalization, and the future of marketing work.

Key Discussion Points and Insights

1. Alembic’s Origin and Philosophy

[01:15 - 05:37]

• Alembic was founded by Puig and two co-founders from diverse backgrounds—quantitative economics, creative direction for global brands, and infrastructure engineering (ex-Twitter).
• The company’s mission: reclaim storytelling and creativity in marketing, which has become overly obsessed with incremental digital metrics.
• Inspiration from pandemic-era advancements in large-scale, causal modeling: "A lot of people think mRNA is the only big tech to come out of the pandemic. The other thing to come out was a lot of incredible math." – Thomas Puig [03:06]
• Alembic runs its own physical hardware cluster for privacy and performance.

2. Data, Private Datasets, and the Next Frontier

[04:19 - 08:16]

• The value and profit in the coming decade will follow proprietary information, not public datasets:
  • "All the alpha, all the profit that will exist in corporations in the next while will all come from private data sets." – Thomas Puig [04:27]
• LLMs trained on similar public data are converging, suggesting that unique value will be driven by proprietary and lived-experience data (e.g., Disney as a data-rich entity).
• Comparing the rise of AI in marketing to developments in finance (Renaissance Technologies, Palantir) and noting the increasing role of major enterprises due to capital and hardware resources.

3. Alembic’s Approach: Privacy, Data Ingestion, and Signal Processing

[08:16 - 15:03]

• Alembic only works with anonymous, aggregate data—no PII, modeled after biomedical standards.
• Handles enormous quantities of disparate marketing data (e.g., 100 billion rows in three days).
• The challenge: In marketing, "When you have that much data, you don’t even know the question to ask to be able to analyze it." – Thomas Puig [09:27]
• Technical innovation: Using NVIDIA-hardware-custom spiking neural networks (SNNs) for signal processing and anomaly detection—modeled after human neural signaling.
  • "We actually use spiking neural networks to do it...a digital twin of the human brain to where it can actually process signals like neuron spikes, just like your brain would." [11:19]
• Alembic built its own simulator for neuromorphic computing on NVIDIA chips.

4. Causal AI and Insights Delivery

[12:33 - 17:32]

• SNNs enable detection of outliers without time history—solving for short-lived, high-impact marketing events (e.g., Olympic sponsorship case study with Delta).
• Causal modeling: Maps the chain of events (e.g., seeing a medal ceremony, searching for a flight, booking a ticket) to link ad spend to actual outcomes.
• Delivery of insights is inspired by intelligence briefings, not dashboards:
  • "Nobody wakes up in the morning, goes, you know what I want in life? Another dashboard...We provide [clients] literal intelligence briefings." [14:10]

5. Changing Creativity and Democratization

[17:34 - 21:10]

• AI is shifting the creative landscape: Curators are becoming creators, expanding from the old 1-9-90 rule (creators, curators, consumers).
  • "Suddenly, with AI, the curators are actually creators...You've 10x the footprint of people who are wanting to build things." [18:07]
• Analogies to music’s democratization with home recording and software’s parallel evolution into mass-custom, rather than bespoke, production.
• Cautions about maintaining access to tools and the risk of concentration.

6. AI, Personalization, and the Value of Shared Experiences

[21:10 - 24:57]

• AI allows for deep personalization, but Puig pushes back on overemphasis:
  • "Being able to really build these experiences that people have and to be able to bring people closer together, those are the marketers that are going to win." [22:21]
• LLMs by design produce the 'expected' (the median), but greatness in marketing/art comes from the unique and the unexpected.
• The role of personalization: valuable after brand connection is formed, but shared human experiences remain the premium.

7. Role of Human Creativity in the AI Era

[25:46 - 31:59]

• Discussion about whether AI turns creative work into more of a team leader/curator’s job ("Andy Warhol did not paint all his pieces. How's it different?" [27:15]).
• Concern that rapid iteration and reduced cost of 'failure' may make it harder for creatives to develop true mastery:
  • "The interesting conversation about this most is what happens to us as humans when we don’t fail enough to improve?" [29:28]

8. The Last Mile: Actionable Insights and Human-Readable Intelligence

[32:04 - 34:27]

• The final challenge: Translate technical and statistical outputs into actionable, understandable, and valuable intelligence for humans of differing specialties.
• Strong data and intelligence empower marketers and creatives; clarity in ROI will fund more creative risk-taking:
  • "If companies see that somebody who's creative is making me $2 for every dollar I spend, I'm going to spend as many dollars with that human as possible." [34:01]

Notable Quotes & Memorable Moments

• On Private Datasets:
  • "Follow the information. You will find the money." – Thomas Puig [04:23]
• On SNNs & AI:
  • "A spiking neural network is meant to be a digital twin of the human brain to where it can actually process signals like neuron spikes, just like your brain would." – Thomas Puig [11:19]
• On the Limitations of Personalization:
  • "But what great artist, what great marketer wakes up in the morning, goes, you know what? I want to be the median." – Thomas Puig [23:32]
• On Democratization of Creativity:
  • "I think about the number of musicians who never would have had anyone hear them at all if a four-track didn't exist." – Thomas Puig [20:07]
• On the Danger of Losing Failure:
  • "What happens to us as humans when we don't fail enough to improve?" – Thomas Puig [29:28]

Timestamps for Important Segments

• [01:15] – Alembic’s origin story and founding team
• [03:06] – Inspiration from pandemic-era math and causal modeling
• [04:19] – The importance of proprietary data for business value
• [08:16] – Alembic’s stance on data privacy and scale of ingestion
• [09:27] – The problem of not knowing what questions to ask
• [10:40] – Use of custom spiking neural networks for signal processing
• [12:33] – SNNs and causal chains in marketing moments (Delta/Olympics example)
• [14:10] – Intelligence briefings vs. dashboards for marketing analytics
• [18:07] – Expansion of creator class via AI
• [22:21] – Value of communal experiences vs. hyper-personalization
• [29:28] – The necessity of failure for creative growth
• [32:18] – Translating insights to action: meeting people where they are
• [34:01] – Data-driven advocacy for creative investment

Practical Advice & Resources

[35:32 - 38:14]

• Beginner Marketers and AI:

  • Start with ChatGPT, but approach it as a collaborative planner, not just a declarative tool.
  • "Plan before you create...The number one thing that people forget to do is tell the system, 'You may ask clarifying questions.'" – Thomas Puig [35:48]
  • For complex prompts, bookend instructions at the top and bottom.
  • Follow voices you respect in marketing and tech to guide your learning.

• Advanced Users:

  • Curate technical content streams (e.g., from prominent data scientists) for continuous learning.
  • Seek methodologies across disciplines; don't get wedded to one class of models or algorithms.

Closing Details

• Website: getalembic.com
• Alembic participates in major industry conferences (Gartner, NVIDIA, Forrester).
• Puig encourages curiosity, breadth, hands-on exploration, and finding your unique "voice" in the era of AI-powered marketing.

Tone:
Engaged, technical, yet accessible and often philosophical—balancing practical technical details with broader questions about creativity, meaning, and the human dimension of marketing in the AI age.