Disrupting Sri Lanka’s Legal Practice: A conversation with Paralegal founder Elijah Hoole

Going to courts is a tedious option for many Sri Lankans. With a backlog of over one million cases, delays and challenges are rampant at every step in the process. Courts are also intimidating bodies, with obscure processes and going-ons. With this poor access to legal information, a question of democratic right has arisen. If you want to find a court decision or even an up-to-date law, you’ll have to do some time-consuming digging, mostly in-person at a government office. But an AI-search engine — paralegal.lk — is trying to make the courts more accessible to the masses.

In a world where technology-driven solutions to social problems are big money, where big tech dominates conversations, where states push products upon people with little consultation, platforms like paralegal.lk stand out. Recognising the issues that arise with keeping courts inaccessible to the public and understanding the local context behind it, the small team behind paralegal.lk is attempting to drive change in quiet ways.

CommonEdge.Asia sat down with the founder of Paralegal, Elijah Hoole, for a conversation on how ‘law tech’ is rapidly impacting legal practice and increasing the public’s accessibility to the law.

Below are excerpts from the interview:

1) What’s the story behind paralegal.lk?

The intellectual curiosity for paralegal.lk came while reading for my MSc in Artificial Intelligence at the University of Edinburgh in 2022. For a text technologies module, we were tasked with a large scale information retrieval project. That’s when I first thought of building a search engine for Sri Lankan law but none of my teammates were thrilled at the idea.

I returned to Sri Lanka in December 2023. I started consulting for a US medtech company upon return and had quite a bit of time on my hands. My case law project idea was still at the back of my mind, so I started looking at websites to scrape Sri Lankan court decisions. That’s when I discovered Commonlii, our Supreme Court and Court of Appeal websites, and Lawnet. I had to catch them on their good days because they were mostly offline.

Once I had about 10 years of Supreme Court and Court of Appeal decisions in computable form, I found that there were people with niche information requirements that would have been impossible to fulfill without computational methods. For instance, one person wanted a list of all Supreme Court rulings on freedom of speech. Another wanted to investigate the jurisprudential evolution of academic freedom. So, besides running my web crawlers to scrape appellate court decisions, I spent the first year (2023) fulfilling such bespoke requests.

During the April Avurudu holidays of 2024, I could not leave Colombo (I forget why). As usual Colombo had emptied out and I was mostly just bored. So I had a crack at building version zero (v0) of my ‘a search engine for Sri Lankan law’ idea from Edinburgh. It was super fun. Four days of non-stop, sleepless hammering away at the keyboard in my apartment with some beer and beef on the side.

I have a few friends who practice law who tested the search engine after their Avurudu slumber. Their verdict was that paralegal.lk v0’s results relevance was better than what was already available on the net.

2) Do you have a legal background?

Well you can say that I ought to have or could have had a legal education. As a mid-teen, I had perhaps a misplaced fascination for legal practice, entirely down to binging on John Grisham novels. After my Advanced Level exams, I was quite set on pursuing a law degree. It was my late father’s wishes, along with the bickering of close friends who were themselves pursuing science-related degrees, that eventually persuaded me to choose engineering over law.

But, clearly, John Grisham left a lasting mark and here I am.

3) Can you give us a rundown of your operations and how they align with your goals for paralegal.lk?

Since 2024 Avurudu, I guess I have spent about four to six hours per day on paralegal.lk. Sometimes more, sometimes less. And I try to do all of it by 10.30am, because I work full time elsewhere to pay my bills.

To date, paralegal.lk has made a gross sum of Rs. 3,000; that too only this week. So that’s one and a half years of hacking away without making barely a cent. In retrospect, there is something to say about that fact though. I guess I have been at it primarily out of a deep association or fascination with the work. The inability to monetise this work has not bothered me yet.

At tech consultancies like the one I am employed at, the work can get a bit repetitive as inbound projects assume a similar flavour based on the consultancy’s core competencies. On this count, I have found paralegal.lk to be a nice counterbalance as there’s constant discovery, iteration, and tinkering. All the more so because I am not really sure what I am doing half the time.

That said it is critical that paralegal.lk starts generating some income, at least enough to sustain itself. Earlier, I had planned to monetise the search engine. But I think differently now: paralegal.lk will remain free to access, forever…anyone can just sign up and start searching.

As a matter of principle, I think some of these resources should be free access. For instance, today, in Sri Lanka, you can’t look up the consolidated version of an Act (i.e., an Act updated with all its amendments) without paying a private company. It is quite frankly an absurd situation that a citizen has to pay to access the country’s laws in their most current form and we are currently working on a consolidated legislation viewer.

And I intend to keep our foundational tools such as the case law search engine and (in-progress) legislation viewer free to access and build an intelligence layer, i.e., AI enabled services, on top of these free tools that I hope I can sell. The Rs. 3,000 we made is in fact from a trilingual translation tool we launched recently. We also have an AI legal research assistant – paralegal-dot-ai – that can query the case law search engine and legislation repository to give fairly robust answers to difficult legal questions being trialled by some lawyers.

4) What are your key challenges?

Only a sliver of what happens in our courts is digitised. In the 21st century, what is not digital is effectively invisible. Put differently: in an age of pervasive computing, non-digital is non-existent.

For example, in Sri Lanka only the Supreme Court and the Court of Appeal publish their decisions online, and even that began only around 2009/2010. Setting aside our colonial judicial history, we lack a representative picture of what transpired in our courts even after independence. While law reports such as the New Law Reports and the Sri Lanka Law Reports exist, they selectively capture a very tiny fraction of our jurisprudence. We remain mostly clueless as to the workings of our lower courts because decisions are not published and can only be accessed on an ad hoc basis.

If a decision isn’t online, one has to actually visit the court registry and pay for a certified photocopy. Court decisions are public documents, so it shouldn’t be so hard to get a hold of them. Today, data storage is so cheap and the marginal cost of serving a digital file is vanishingly small. When court records are freely and publicly accessible, it enables analyses and insights that are otherwise impossible to perform or gain.

For instance, a major public complaint about our courts concerns the time they take to resolve a case. Yet we don’t even know something as basic as how long, on average, it takes to resolve a lease dispute in the district court. This is a simple counting problem. The only reason why we don’t have an answer is because our records aren’t digital.

According to the Constitution, a fundamental rights petition must be concluded in two months. In reality, you’d be lucky to have your “fundamental rights” restored in three years. And without access to case records, it is impossible to know what actually causes such delays.

There are plenty more questions, much more interesting ones, that become immediately tractable once you have digital records. Buried inside cobwebbed cupboards of our court registries are very interesting answers to some of these very interesting questions, many of which may be very unpleasant to one too many.

In any case, our courts have consistently recognised that court records — pleadings, petitions, objections, proceedings, minutes, decisions, etc. — are “public documents”. But one could argue that their practical approach with respect to public access to the same is at odds with their own recognition, especially in light of rights granted under the Right to Information Act.

With the new Chief Justice highlighting ‘digitisation’ as a central goal of his tenure and recent, wholesale changes within key institutions like the Judicial Services Commission, there appears to be positive movement within the judicial system on this front. Those of us on the outside must take advantage of these small openings and push for progressive change.

5) You’ve singlehandedly come up with a solution to make court records more accessible to the Sri Lankan public. Why do you think the state struggles, where you succeeded?

Singlehandedly is a stretch. I have been fortunate to work with super talented interns.

But to answer the question, the Sri Lankan state suffers from a problem of competence. The core problem is there are scores of technically stronger software engineers than me in this country but none of them are employed by the state. Why? The state is broke and cannot pay market rates.

At paralegal.lk, on the tech side, there is a certain level of technical complexity and some difficult challenges but these are all solved problems in computer science. You don’t have to invent anything new to build a functional search engine. In fact, there is a very mature open source stack. So the challenges are integration and maintenance. It is hard, therefore, to understand why the state, broke as it may be, can’t keep Lawnet up.

On the other hand, over the past year, the Supreme Court and Court of Appeal websites have become noticeably more available and reliable. There may yet be some hope left.

6) I think we are at a turning point in technology and how it’s influencing the way society functions. Do you think our value systems will change as a result of these technological inventions, for instance, like AI?

There’s a lot about AI I don’t fully understand but what I know is that some human endeavors were never meant to be about raw productivity. Personally, I’m glad ChatGPT didn’t exist when I was a student. I’d have had it to write every assignment and ended up illiterate.

When you Google something, you stumble on tangents, wander into unrelated sites, and pick up stray knowledge you never planned to. You may not use it right away, but it sticks, and later your brain connects the dots. ChatGPT skips all of that. It hands you the end product but robs you of the process.

For the skilled and proficient, AI is a time saver but its long term effects of over reliance are yet to be fully understood. But for students, in its present forms, AI is mostly just a trap stunting their growth. Generally speaking, when we rely on AI, we end up narrowing our search space. We trade off serendipity for speed which I imagine is not always a good thing.

This much is certain though: AI is going to upend knowledge work. And its rise carries a humbling lesson for those of us who make a living at a desk; for those of us who are, quote, unquote, in knowledge work. It turns out it’s far easier to get AI to write code or draft a petition than to get it to iron a shirt or clean a manhole. A lot of people who have thought of themselves as “very sharp” are about to learn they’re not so sharp after all.