Alternative Data Weekly #261
Theme: The edge goes to the workflow orchestraters
Special thanks to our sponsor EventVestor.
QUOTES
“Trended and alternative credit data provides the most complete picture of consumers, and TransUnion’s new approach unlocks this vital data in the mortgage lending industry, benefitting homebuyers, lenders and investors” - Satyan Merchant, senior vice president and mortgage business leader at TransUnion.
News
Pods
Charts
Final Thoughts (the hunt)
#1 – Alex Izydorczyk published Skepticism, Early Trends & an Early Leader in AI-for-Hedge Funds Race. October 2025.
My Take: Alex does us all a favor by sharing his effort to track all these startups. He published an admittedly imperfect attempt to place the 100+ firms into 6 categories:
Research Copilots (AI “Analyst” Assistants).
Excel Copilots (Financial Modeling Aides).
“Terminal 2.0” Platforms (Next-Gen Market Terminals).
AI Model Providers (Alpha, Quant, & Forecasting Labs).
Data Extraction Tools.
This is going to evolve but is a great start.
#2 – Didier Lopes published Rethinking the financial interface in the age of data and intelligence abundance. October 2025.
My Take: Centralized models that maintain full control (and have been great businesses!) are being pushed to their limits. The data landscape has simply outgrown the idea of a single vendor controlling the flow.
Open > Controlled
Modular > Monolith
BYOD > Walled Garden
This has me recalling the idea that we are moving from a system of record at the core of everything (CRM, terminal, etc) to a system of action that is at the center of everything.
The key: deeply understand your workflow in the context of what you are trying to accomplish. Be sure your systems have the ability to deliver to you what is needed (preferably without you having to ask for it).
This situation also reminds me somewhat of the old cable bundle, where you had to pay for 500 channels even though you only watched five. The competition for the cable providers came from a billion iPhones, social media, and the 7-10 streaming services that we now pay separately for, but (mostly) use.
#3 – Matt Robinson published AI (Still) Flunks Wall Street’s Analyst Test. October 2025.
My Take: AI has taken the finance world by storm but there is still a long way to go before we turn investment decisions over to AI. It isn’t a question of compute, but more a question of accuracy & validation. A play on the popular quote is that “it won’t be AI that manages your money, it will be a PM that uses AI that will manage your money”. It is simply so much faster & cheaper to get information processed, that it will have to be part of the workflow.
The author reminds us that a good prompt is the key (The Question Is Your Moat).
BONUS: Alan Conroy, Nishant Kumar, Zihao Xu, Andreea Bajenaru, Lukas Olson, Jo Stichbury, Joanna Sych of McKinsey published Unlocking consumable data for generative AI. September 2025. “As enterprises deploy gen AI, one truth is clear: data quality and structure determine impact. Without accurate, validated, and audit-ready data, the promise of generative AI and agents is left unfulfilled.”
manWhat else I am reading:
Jan Aasman published The rise of accountable AI agents: How knowledge graphs solve the autonomy problem. October 2025.
Gurpinder Dhillon of Senzing published The Future of Entity Resolution In the Age of Generative AI. October 2025.
Dan Evans published Intuitus: Thinking Smaller to Discover More. October 2025.
Robin Wigglesworth & Jill R Shah of FT published Hedge funds and high-frequency traders are converging. September 2025.
Julein Brault & Danial Dzulkifly published Jimmy Donaldson, The World’s Top YouTuber, Just Filed a Trademark For a Banking App. October 2025. (for me this really hammers home the growing importance & power of brand, attention, and ultimately trust).
Jonathan Silver of Affinity Solutions published Banking on value: Why optimized commerce technology drives critical bank growth. October 2025.
JP Morgan published 2025 Emerging Technology Trends. October 2025.
Glacier Network Launches Unified Platform for AI and Data Vendor Risk Management. October 2025.
Source: Charlie Ko, Aoife Manley, and Ezra Levine of Gemsen joined Rob Sarnie on his podcast to talk about the future of machine learning in financial services. September 2025.
Unlocking Data Power with GEMSEN: Faster AI, Smarter Insights & Data Privacy (Part 1)
Part 2 found here.
My Take: As I have written about in the past, there is too much data for any human to digest and make sense of. Financial institutions are swimming in data but starving for insights.
Now that we have access to all this data, what do we do? This can all be very expensive and difficult to orchestrate. Gemsen has a solution.
Gemsen provides a statistical abstraction layer that speeds up the time spent training models. You can train on that abstraction in milliseconds. And this, by mathematical nature, is privacy-preserving. The “GEM” is the term used to describe this new ML process.
The amount of compute needed to execute is overwhelming; this solves that.
The rare solution that delivers on the promise of 100x better at 1/100 the cost.
Highlights (22-minute run time & 27-minute run time)
Part 1
Minute 01:00 – background on each person
Minute 06:00 – Charlie provides Gemsen’s background
Minute 10:00 – fundamentally different way of doing ML
Minute 13:00 – the statistical abstraction layer snapshot of massive datasets
Minute 16:00 – privacy at all sizes of dataset
Minute 19:00 – how to build better models faster
Part 2
Minute 00:45 – the Aha! moment
Minute 08:00 – The Mass Challenge & the power of the network
Minute 16:00 – what startup stage is Gemsen & what is the beachhead market
Minute 21:00 – upcoming events for Gemsen
BONUS: Jamie Dimon Interview | Titans and Disruptors. From 9:00 - 14:00 Jamie talks about AI & the use of data.
Source: Plaid published The Fintech Effect 2025.
These awareness numbers seem low to me. If true that only 47% of people are aware of GenAI, we (people reading this) are way out in front of this AI thing.
Source: AWS Outage. Pipeline Problems? What next?
The Hunt.
Since my former employer, ModuleQ, “dissolved” in late August, I have been on the hunt for my next role.
This network has been wonderful, and my optimism for the future is as high as ever. There is a TON of opportunity in the world. The key is to find the best fit for all parties, which comes down to timing, personality fit, role responsibility, and growth stage of the company.
I am focused on finding ambitious, conscientious people who are trying to build a great company, preferably in the data, AI, and/or research space.
Perfect timing for this. That 'system of action' idea is fascinating. Does that mean a real-time workflow overhaul, or just very smart new dashbaords?