Claira co-founder, Eric Chang, and co-founder and advisor, Joe Squeri, answer SCI's questions

Q: Claira, the document intelligence fintech, recently announced its collaboration with Citi to help boost digital transformation within securitisation and develop next-generation data analysis solutions for CLOs (SCI 29 June). Tell us more about your vision for the company.
JS: The digitisation of structured credit documents has been something I’ve been trying to solve my entire career, and I’ve suffered through the various pain points. When I joined Exos Financial, a next-gen B2B platform for institutional finance,

we went through that same process in 2018 to find a data science-based product that could digitise our financial documents. We tried a dozen different legal AI products, of which none could draw complex insights from structured credit documents, even after six to nine months training them.

EC: The information in documents is really hard to access, and firms can spend months negotiating 400 pages, and going through all the conditions and revisions they want to account for to make sure that’s incorporated and is really a key part to a security or a deal. As a trading analyst, you’ll probably spend anywhere from 30 minutes to a few hours trying to figure out a deal and typing it into your models.

With Claira, you can get that information within minutes. Clients will get a data feed, and they can incorporate that into their models, and will end up with something that’s much more accurate, and more reflective of the security that you are transacting.

JS: And once we developed this for Exos Financial as a solution for our credit agreements, we saw that what we had was something really innovative which we spun out into Claira. We then started incubating Claira as a fintech, and now it is really starting to solve pain points for other clients, and this partnership with Citi is a great example of that.

Q: Why is the speed of processing documentation so important? Why is it something firms are focusing on more at the moment?
EC: I think the speed of the information is the main point because these are live trades - and the faster we react, the greater the advantage. I think this shift is also reflective of the broad changes that are occurring within the industry as we see more digitisation and streamlining of the markets. Our partners at Citi were also involved in the Octaura Holdings announcement (SCI 14 June), and you can see this across the free markets that where traditionally things had been done over the phone or via email and you had time, that’s no longer available.

JS: Claira’s competitors generally have analysts with advanced degrees, who are manually reading through documents and extracting information into Excel. So, not only is this an opportunity to get things done faster, but human accuracy is only 85% accurate, whereas Claira is close to 100%.

Q: What is Claira able to offer clients that other firms cannot?
EC: First and foremost, I would say it is our approach to the tech-solution and the actual depth of AI analytics that we are able to provide. Most of the firms that are looking to digitise in the document space may take what we would call a ‘Swiss Army knife’ approach to the technology – where there is a lot of labelling and tagging, and the technology can be applied to almost anything because you have to feed it the data.

What this approach doesn’t offer is specific solutions that are particularly skilled or effective in solving any one thing really well. Whereas, at Claira, we’ve taken a computational linguistic approach - it is a bottoms-up proprietary method of understanding legal and financial language, and we’ve built a model that represents a document as a whole. By doing that, we can present results to our clients with the detail that a portfolio manager or trader needs when it comes to the conditions and the provisions that are outlined in the documents.

JS: And Claira can do all of this in minutes, as opposed to hours - which, again, better caters to the recent market structure changes we are seeing in credit that are increasing the speed of trading.

EC: There’s also a scalability that these approaches cannot offer because, unlike those, our models are pre-trained. As a CLO manager or investor, you can load the documents you need, and get the AI analytics that you want in a very streamlined fashion.

The integration of this technology into clients’ investment process is even more streamlined now, given the greater digitisation seen across the market. Most clients now have quantitative models and document databases already established, and we can just plug directly into their existing risk and portfolio management systems and so on.

Another key part of our solution is that there’s also room to adjust and change, as we are able to configure the model for what they want from them. If a client wants something new, that would only be a couple of weeks of work for us to make happen, where these other Swiss Army knife approaches would need another nine months of tagging and labelling again on the bank’s behalf. This is not a separate process a client has to download into excel; it is well engrained into their investment process, which also helps to really speed things up.

Even with all this scalability, Claira can help with transparency too. These are often US$100m deals being analysed, and with Claira everything can be traced back to the context of the document itself. So, as an analyst, you are still in the loop, and can trace back and dive in a little bit further into the document without having to read the whole 400-page agreement again.

Q: What securitised products is Claira able to work with?
EC: Our work with CLOs is really a product of our partnership with Citi, and prior to this we were already covering different areas, including commercial real estate loans and people with debt=backed securities. But wherever there are complex terms and conditions, Claira can come in and help.

JS: And even taking the work that’s being done in a post-signature way to helping benchmarks for pre-signature negotiations as well. Whenever you’re in negotiations, lawyers or business heads need to remember what they’ve agreed on for a private client, and all of that is based on either notes they’ve taken, observations they’ve written down, or just gut feel. Claira allows for the digitisation of all your historical documents, which means clients can create benchmarks for comparison for future negotiations - or on a pre-signature basis - which ultimately allows for people to take a more data-driven approach than this gut-feel analysis.

Q: How does natural language understanding (NLU) differ to existing natural language processing (NLP) services?
EC: From a technology standpoint, what we have built is proprietary, and so my hope is that everyone is going to take up on it. What we hear from clients is that the current batch of NLP solutions are extremely limiting, and don’t necessarily solve the problems of the financial professional.

They can maybe handle some of the simple procurement stuff, but they are not designed for complex conditions. This causes a lot of frustration across the industry, and Claira is extremely well positioned to solve these problems.

JS: If you compare the techniques of NLP to the language understanding of NLU, you can see that turning words into numbers and vectors and then comparing them can only get you so far. When you have heavily negotiated and more bespoke documents, these NLP models break down, simply because the patterns are not as easily recognisable.

However, when you use the NLU models - like the language understanding and computational approach that Claira has taken - the permeations of different words and sentences on even bespoke agreements can be run through Claira and given that insight. Claira offers a really unique approach in what we think is the next generation of document intelligence.

Q: Why do you think we are seeing increased technological innovation in the securitisation space right now?
EC: The transformation that’s happened across the market is not a new trend - nor is it unique to securitisation - but it has definitely been accelerated recently. Technological innovation took off earliest in equities and then bonds, but the securitisation market has maybe taken longer to innovate in this way because of the complexities of the deals involved. A solution like Claira can help accelerate the digital revolution, even in the securitisation market, as it works to reduce some of these complexities.

As with any financial market there might be some resistance, but digital transformations have usually been very positive for market appetite, growth, increasing the number of participants, and opening access for people looking to issue or invest. Digitisation has also been positive for increasing transparency and accountability across the financial industries, and I think it’s going to be a good thing for the securitisation market too.

JS: I’ve witnessed a lot of resistance over the past 25 years or so to the use of technology, and it isn’t normally the technology that proves to be the difficulty but rather human behaviour. People fear giving up control, but what Eric and Claira co-founder, Alex Schumacher, have been able to design takes that into account.

Claira offers clients the ability to double-check on the conclusion given by the technology, which gives people more confidence. Claira is not built to be self-driving; it is built to be driver-assisted. Financial transactions are important, and it is critical to have that transparency and that trust when you are putting capital at risk.

This model also sets a precedent for the next generation of portfolio managers, who are more familiar with digital to not just trust the machine blindly, but to actually engage with it – which is precisely what Claira is built for.

Q: What are your ambitions for Claira going forward?
EC: In the short and medium term, we will be focusing on our partnership with Citi, as well as promoting ourselves, building a brand, and having people really get to know what our solution is about. From there, I think anywhere where there are complex conditions in important agreements is what we want to tackle next.

Even in leasing, if you sign a lease, how are you meant to know it is fair? How are you meant to know what the market standard is? So, if you can analyse 500 leases, you can start to paint a picture of what is and isn’t standard, and you can start to be data-driven in some of these decision-making processes.

Insurance is another area where there are a lot of conditions, and questions about whether someone actually has coverage and so on.

JS: Insurance is definitely on our to-do list. We are working with major insurance companies, but generally in the investment side of insurance, like insurance policies. At the moment, we are seeing a lot of private flood insurance policies being written because of climate change - of which most is non-standard, and so we are working to assist in the analysis of core insurance policies.

EC: I think the opportunities for Claira are very vast, and legal negotiations and conditions are fundamental to how capital markets work. So, for us to be able to digitise those agreements, and provide data and insights about those documents, I think is going to be really helpful for the market as a whole.

Claudia Lewis