How customers use 9fin AI to accelerate their research
- Josh Latham
At 9fin, we’re building tools for debt market professionals. Several of us have worked in big banks and law firms — and have lived the challenges first-hand.
We build tools tailored to our customers’ needs, and our engineers speak with clients on a regular basis to stay ahead of what matters most to them.
That’s why we built 9fin AI, integrating generative AI to help our customers streamline their research. Our latest update adds access to SEC filings, including 10Qs and 10Ks, putting even more critical data at customers’ fingertips.
So how do clients use it? How do we use it at 9fin? And how do we ensure security and confidentiality as the product develops?
We asked our European credit team lead, Josh Latham.
Hi Josh! Tell us a bit about 9fin AI — how does it work?
Let’s say you have a question on a specific issuer.
Just type your question into the search bar, and in seconds 9fin AI will scan our universe of 5,000 earnings call transcripts; bond issuance offering memorandums; SEC filings such as the latest 10Qs and 10Ks; and 9fin’s entire proprietary content library, to find the answer.
Now, imagine sifting through all that without 9fin AI. The time saved is a huge difference from having to search for individual documents, read through the content, and piece together a story manually.
“9fin AI can scour our universe of content for information on anything from springing covenants on a revolving credit facility, to whether a deal has a J.Crew blocker.”
How do our customers use 9fin AI?
There are so many examples I could choose! But here are three I like to highlight.
1. The analyst on the buyside
A company’s earnings have just been published, and the analyst wants to get up to date with the key points, like performance in the quarter, or management guidance. Time is of the essence — they’re stretched thin.
Instead of hours of research, they can ask 9fin AI, “What are the key events from Altice France’s earnings report?”
The model will instantly scan the SEC file, but also the transcript from the earnings call, combing through two different documents at the same time.
What’s more, the analyst can then:
- Access the source documents straight from 9fin AI’s response
- Use our global search tool to dive further into detail if needed
2. The analyst on the LevFin desk
Let’s say a company has just come to market, and you need insights on its business, risks, and competitors. Instead of searching multiple platforms, you can just use 9fin AI.
Run a SWOT analysis on any company in our universe with 9fin AI and get an instant answer.
Beyond asking about figures or inputting yes/no questions, you could ask more complex questions such as “what was management’s tone on TriNet’s Q4 call?” and our model will provide an answer from the transcript.
To see this example in action, check out the short walkthrough below.
3. The senior associate at a law firm
High yield lawyers often have several deals on the go at any given moment. Sifting through documents to find the specific terms or covenants can take hours — and that’s hours they don’t have to spare.
Luckily 9fin AI can do the heavy-lifting, scouring our universe of content for information on anything from springing covenants on a revolving credit facility, to whether a deal has a J.Crew blocker.
9fin AI’s scans through not only 10Ks or 10Qs, but also our bond and loan legal summaries (known as QuickTakes) and offering memorandums, with a link to the source document so lawyers can always delve into the granular details.
“We're in an industry where one mistake could cost millions — or even billions — of dollars … That’s why we’ve worked hard to counter hallucinations in our approach.”
What measures do we have in place to ensure security and confidentiality?
Access to data is tightly controlled — only employees who need it for troubleshooting or product improvements can view it.
Our journalists and analysts are completely separated from this data, so customers can be confident that their queries remain private and aren’t used to train the generative AI model.
Confidentiality is a top priority in this industry, and 9fin AI was built with this in mind.
Accuracy is also a huge priority, especially in this industry — how do we ensure 9fin AI’s responses are reliable?
We're in an industry where one mistake could cost millions — or even billions — of dollars. With stakes this high, our clients are understandably concerned about accuracy. That’s why we’ve worked hard to counter hallucinations in our approach.
1. 9fin AI cites the source material with each response, giving customers a reference to validate answers.
2. We’ve expanded access carefully. Instead of overwhelming the model with our entire universe of content, we’ve strategically phased in access to key resources, like transcripts, offering memorandums, and SEC filings.
3. We rigorously test each expansion before launch, ensuring any issues are resolved before customers use the feature. We plan to keep adding more proprietary content for the model to access, but this careful approach has helped us reduce the risk of hallucinations.
What’s the impact of using 9fin AI?
Our customers have used the feature to ask over 30,000 questions and get insights on over 2,100 companies since we launched 9fin AI in 2023.
Usage of 9fin AI has also tripled since launch, with hundreds of users saving time with the feature each week.
“Manually it would take hours to conduct the research, piece together the individual bits of information, and compile a report — AI can help us achieve much more in a fraction of the time.”
How does our credit team use 9fin AI?
The earnings calendar can be hectic — we can have up to 30 companies reporting in our core coverage in any given week, but we don’t have the time to look through all the documents.
What we can do instead is ask AI: how did they perform this quarter?
We then use excerpts from 9fin AI’s response to produce our own digests, which are weekly wrap-ups of recent earnings. It’s a nice way to balance the speed and impact of AI with our industry expertise, and helps us work more effectively.
That’s also one of the reasons we’re able to get valuable intel to our customers so quickly. Manually it would take hours to conduct the research, piece together the individual bits of information, and compile a report — AI can help us achieve much more in a fraction of the time.
Want to learn more about AI in debt markets? We ran a webinar with 9fin CEO and co-founder Steven Hunter, and Financial Times journalist and author Sunjeet Indap to discuss the topic further. Access the recording here.
Finally — what’s next for 9fin AI?
We want to add even more data to our global search function, so customers can ask industry-specific or market trend questions. Ideally in future evolutions:
- You can ask 9fin AI about the total issuance in European high yield for single-B rated companies in 2024, and the model will use our existing data to extract that information, across multiple companies
- You can ask for the largest cap stacks in Europe and the model will be able to source that information from our database
- 9fin AI will cover more than just text input and output, so users can add images, request graphs, and have 9fin AI complete tasks on their behalf, like creating a slide deck from various pieces of intel
We have big plans for the year ahead — it’ll be exciting to see how we use the latest tech to make 9fin faster and smarter than ever before.
Find out more about 9fin AI in our blog here, and get in touch to see how our AI-powered platform can help you work faster and smarter.