Ten Big Banks Back the Same AI Platform

WALL STREET RESEARCH

Wall Street Banks Bet on AI Platform Aiera with $25M Round

Ten of Wall Street’s biggest banks are backing Aiera, a financial AI platform, in a rare alliance to build shared AI infrastructure for distributing research.

The $25 million Series B round, announced Monday, also includes Microsoft as a technology partner and expert network firm Third Bridge.

While Aiera didn’t name the participating banks, a spokesperson said clients will gain access to the lenders’ research alongside reports from across the industry. The platform is expected to launch in late 2025 or early 2026.

The investment targets what Aiera calls critical industry challenges around intellectual property protection, content access entitlements, AI accuracy, and citation integrity. I’ve written previously about the absence of AI standards in finance and how this is a barrier to widespread adoption.

To guide product development, Aiera has formed a buy-side advisory council made up of senior leaders from large asset managers.

“Aiera’s technology solutions will be made available to all content creators, content consumers, existing content aggregators, and new and emerging platforms within the financial services industry,” said Gavin Skinner, Aiera's chairman and chief operating officer, who joined the company last fall after two decades at Citi.

In a separate statement, president and CRO Chad Doerge said that Aiera is tackling the "screen space problem" — too many dashboards and different portals.

"We're developing a centralized platform that will contain most of the external content required for institutional investors to conduct comprehensive research workflows," said Doerge. "This unified ecosystem eliminates the need to navigate disparate sources and disjointed dashboards."

Takeaway:

Rather than each bank building competing AI research platforms, traditional rivals are choosing to invest together in shared infrastructure, a step towards industry-driven standards for financial AI.

ICYMI: I demoed Aiera back in January, check it out here.

HEDGING

Ant's Time-Series AI Cuts FX Hedging Costs 60%

Over the past few weeks, I’ve written about how the architecture behind ChatGPT can be applied beyond language. Instead of predicting the next word, you predict the next transaction, payment, or market move.

Stripe trained a model on the language of payments. Netflix is applying the same architecture to refine show recommendations. Startup Unboxed AI built a transformer on grocery store transaction data.

Now, Ant International (part of Ant Group and Alibaba) says it has trained a time-series transformer that slashes FX hedging and liquidity costs. Ant didn’t disclose what data exactly they trained the model on, but the results are pretty striking:

From this South China Morning Post interview with Kelvin Li, who leads Ant’s platform tech unit:

This is the first time I’ve seen a financial services firm talking about reducing hedging costs because AI is that much better than current models.

This is another example of how the transformer architecture—originally developed for language models—is versatile enough to excel in completely different domains.

Takeaway:

If you can reliably reduce risk, companies can reinvest those savings elsewhere. And for banks, better risk models could justify smaller capital cushions, easing regulatory pressure and unlocking balance sheet capacity. In short, AI could make risk cheaper.

If you know of other finance firms using transformers for time series data, please reach out: [email protected]

REGULATION

Banks Adopt AI Tools Despite Policy Gaps

Financial firms are using AI tools, mostly internally, but few have clear policies for the technology, according to the European Securities and Markets Authority.

A survey of 38 industry experts showed 85% of organizations use tools like ChatGPT internally, but only 15% apply them for customer communications due to risk concerns.

Report: Leveraging Large Language Models in Finance: Pathways to Responsible Adoption

Takeaway: Regulation is behind. And regulators promulgate new rules pretty slowly, so it’s going to be some time before there are new rules on the books.

Related: 

• UK finance watchdog teams up with Nvidia to let banks experiment with AI (CNBC)

• BIS, Bank of England Tests AI to Detect Financial Crime in Real-Time Payments (Fintech News)

PERPLEXITY

AI Can Draft ~99% of IPO Documents: Perplexity CEO

Earlier this year, Goldman Sachs CEO said that AI was drafting about 95% of an IPO prospectus. The CEO of Perplexity said that his company can do so at 98% to 99%.

I’ve reached out to Perplexity for more information on whether they’re working with any banks or law firms to use this technology. I haven’t heard back yet.

Also, this story lets me bring back this meme:

AGENTS

JPM Creates Multi-Agent AI System for Research

JPMorgan’s Private Bank has built an AI platform called "Ask D.A.V.I.D." to automate investment research and provide faster answers to client questions about thousands of financial products.

This is the first I've seen a major bank discuss how it's integrating AI across its businesses in great detail.

LLMs alone are very powerful, but they’re probabilistic, which isn’t ideal in finance which demands preciseness. But if you combine LLMs with deterministic systems like a calculator, you get significantly better results.

That’s the stage we’re in now: connection. LLMs aren’t yet standardized across different processes.

JPM is starting to connect different agents.

Ask David — short for "Data Analytics Visualization Insights and Decision-making assistant" — uses multiple AI agents working together to handle different types of queries. The system can process both structured data through database queries and unstructured information like emails, meeting notes, and recordings.

The technology aims to help financial advisors provide real-time answers during client meetings. For example, if a client asks why a particular fund was terminated, the system can quickly retrieve performance data, research findings, and similar fund information that previously required extensive manual research.

The multi-agent system relies on a supervisor agent that delegates tasks to specialized agents handling structured data, unstructured documents, and analytics. It incorporates both short-term and long-term memory to customize user experiences and includes human oversight for accuracy.

Takeaway:

We’re still in the early days of AI adoption in finance. It feels like the internet before the first major web browser. There are a lot of separate systems, but building something smooth and large-scale is still hard. JPM is showing that this won’t be the case for long.

Related:

• AI in Quantitative Investing: JPMorgan’s D.A.V.I.D., BloombergGPT, and NL2SQL Challenges (Quant Journey)

CALENDAR

Upcoming AI + Finance Conferences

I’ve put together a calendar of upcoming AI and finance conferences. Click on the image below to access links and event descriptions. Let me know if I’ve missed any and I’ll add them. Thanks!

WHAT ELSE I’M READING

• Retail Stock Investors Can Now Imitate the Pros With AI Trading Tools (Bloomberg)

• Barclays to roll out Microsoft Co-Pilot to 100,000 staff (Finextra)

• At Amazon, Some Coders Say Their Jobs Have Begun to Resemble Warehouse Work (NYTimes)

• Vista CEO Says AI to Force 60% of SuperReturn Crowd to Seek Work (Bloomberg)

• No, AI Robots Won’t Take All Our Jobs (WSJ Editorial)

• Interview with DTCC’s head of Tech on AI (Traders Magazine)

• Banks play the model field (Evident AI)

• This Stanford professor built a ‘Terminator’ AI fund manager that crushed 93% of human stock pickers. (Fortune)

• Starling rolls out AI-based 'Spending Intelligence' tool (Finextra)

• Paul Tudor Jones Fears AI Job Losses Will Worsen Social Division (Bloomberg)