Stop Looking for a Faster Horse Carriage in a Tesla World

Back to Blog Posts

“They are all the same” is a common refrain in ediscovery when discussing review platforms and processing technologies. Think beige vs. taupe. Thankfully, this assumption is dead wrong. 

Ediscovery tools today range from antiquated legacy technology that is about as fast as a limping horse pulling a buggy with a broken spoke to tech-optimized solutions that can search, refine, and render documents in a fraction of a second. When time is money, as is painfully apparent in the case of ediscovery, getting where you need to go fast can make or break the bank — so choosing the right tech matters now more than ever. 

Document review can be the biggest cost in the litigation budget. DISCO supercharges review not just by reducing docs that have to be reviewed, but by moving lightning fast through those that do. In the costly ediscovery race to the finish, DISCO is the fastest car that gets you across the finish line first. 

Why is DISCO’s ediscovery supercar better?

More robust technology, advanced processes, and turbocharged AI are all part of the equation for winning the ediscovery race. DISCO is a better car from the wheels up, and that leads to better results guaranteed

Cloud-native technology outpaces the legacy fog

When legacy ediscovery tech moved into the cloud, it mirrored the server architecture used to support on-premise solutions, bringing the same inherent limitations along for the ride. The result was more fog than cloud. 

DISCO, by contrast, started from scratch to create cloud-native architecture that allows our platform to do in parallel what legacy tools have to do sequentially. By taking advantage of a distributed cloud network and cutting-edge innovation from Amazon Web Services (AWS), DISCO can do more at the same time as legacy tools — and faster. 

This underlying structure is exponentially more robust than the SQL-based virtual machine architecture some legacy providers use and does not struggle with larger data sets or get bogged down by more users or tasks being performed. From 1500 to 150 million documents, DISCO still renders pages and performs searches in a fraction of a second. Adding hundreds of reviewers, conducting large scale data ingestion, running advanced AI modeling across multiple issues, and managing multiple massive cases all at the same time has zero impact on the DISCO’s speed, or power because we can scale across the AWS ecosystem to increase compute power at a moment’s notice. 

DISCO gets you to the starting line sooner

DISCO’s cloud-native AI platform is supercharged from the inception of a matter, getting your review teams into the data in a fraction of the time it takes legacy on-premise solutions and even other cloud providers. Thanks to the direct transfer power of AWS Edge-S3, DISCO’s High-Speed Uploader rapidly uploads terabytes of data directly into your review database. The platform is further supercharged with 19,000 auto-scaling Lambda groups (a kind of on-demand cloud server from AWS). These automatically load-balance in a fraction of a second to support the processing of any amount of data, while automatic parallel processing enables DISCO to conduct all 15 steps of our data processing at the same time!

Whew — that's a lot of technical speak to explain that DISCO is cloud ediscovery supercharged for better, faster insights. The end result is we get your data to DISCO quicker, no matter the data volume, and we can process and promote it to review faster than the competition. 

via memegenerator

Arrivederci, spinning wheel of death

DISCO’s cloud-optimized infrastructure puts an end to having enough lag time to read "War and Peace" as the spinning wheel of death haunts your search screen. Harnessing the computational power of the cloud and continually integrating new and innovative processes and technology (like Google’s BERT and fastText) allows the DISCO platform to achieve search speeds of 1/10 of a second regardless of the document population. 

The powerful processing engine creates a nuanced document model with over 100 discrete document attributes, which in turn generates substantially improved accuracy of search results. This supercharge level of insight is amplified across the entire data universe in keywords, metadata, work product, families, Westlaw syntax, etc. in a single search with the heterogenous search parser. The end result is dramatically more accurate search results in a fraction of a second. 

We have the need for (review) speed

An additional lever that amps up DISCO compared to legacy technology is the document-to-document speed for reviewers interacting with the review platform. Docs in DISCO don’t just load faster — they load instantly.

This feat begins with all documents being imaged on the front end to eliminate lag for page rendering. The system employs optimistic page rendering, displaying the document as soon as the first page has loaded instead of making you wait for the full document. Asynchronous parallel requests for the remainder of the document eliminate in-document page-to-page lag by retrieving all parts of the document in the background for viewing (including the image for the current page, image for the rest of the pages, metadata, related documents, highlights, work product, etc.). Thanks to a proprietary near-native viewer, even massive and complex Excel files no longer create latency. 

The net result is sub-second page to page speeds that greatly contribute to the 60% faster review speeds DISCO managed reviewers are achieving. Even a seconds-long pause leaves room for distraction. Instant document rendering means reviewers keep reviewing documents rather than getting trapped by a chatty neighbor or mentally starting a grocery list. .

AI-supercharged education

Prioritized review enables contextual review. Document review is a crawl-walk-run endeavor. In the beginning, as you familiarize yourself with the content, reviewers move slowly. Over time, as they get more comfortable with how the documents interact with the issues, they review faster. By the end of the review, case teams are able to make split-second decisions and review significantly more documents per hour. 

DISCO supercharges this approach even further because DISCO AI develops nuanced insights in less time and with fewer coding decisions than other TAR technology. This means that the system creates models that give your case team trustworthy insight into your data in a fraction of the time.

Case teams learn more about the case from reviewing these increasingly meaningful documents compared to if they were reviewing unrelated content. Our reviewers get comfortable sooner and, as a result, go directly and quickly from a crawl to run. DISCO Review teams average 88 documents reviewed per hour when the industry standard is only 55, and have reached rates nearly 3x the industry average on multiple occasions. 

What is DISCO AI’s secret sauce?

DISCO AI takes a fundamentally better approach from the ground up. DISCO’s AI models are driven by the math underlying language (abstract semantic representations) rather than keywords or phrases. DISCO AI employs Google’s fastText to parse 300 data points (vectors) to understand a document. The system examines the entirety of a document, taking into account the order, meaning of words, and sentence structure to arrive at an intelligent insight as to relevancy. As a result, DISCO AI is smarter and can understand correlations like man is to king as woman is to queen as well as differentiate between “man-eating dog” vs. “dog eating man”. This context amplified with advanced innovation like cross-matter AI yields a more refined model that can provide more accurate insights when shared with a new data set than a simple vocabulary list. 

As the model evolves the accuracy and results are continually compared to the past model to ensure that only the most accurate model is deployed on a go-forward basis. This dynamic approach keeps the model on track while ensuring the case team gets the most accurate recommendations possible. This approach effectively nullifies the risk of leveraging AI.

What does DISCO Review (supercharged by AI) look like in practice? On day one of a review, the case team found that 2% of docs were responsive. After DISCO AI took inputs from a single day of reviewers making coding decisions, 79% of what was reviewed was tagged as responsive. After a single day of interacting with DISCO AI, our reviewers went from just one out of 50 documents suggested by AI being responsive to four good learning experiences in every five documents.


Much in the way it did people very little good to be the best buggy whip maker in an era of the Model T, the speed of innovation in the legal technology space (and in the data-driven ediscovery ecosystem) can render today's most cutting-edge tech obsolete rapidly. DISCO’s founders specifically architected the infrastructure powering the DISCO platform to be modular and adaptable. 

Unlike some legacy legal technology that is still anchored to SQL databases and VMs mirroring dated on-premise architecture, DISCO was built to continually evolve. Our 200 engineers constantly test new-to-market technology for incorporation into the beast that powers DISCO. From integrating advanced natural language processing up to and including Google’s BERT to the over 200 releases and counting to improve and refine the user experience in DISCO, we never stop building new features and updating existing ones to take advantage of the latest and greatest technology.

So what does DISCO at full throttle look like?

DISCO AI made a tight production deadline possible for Kennedys Law LLP, a global law firm with expertise in dispute resolution and advisory services. Their London-based team had four weeks to produce documents from their client’s initial collection of 1.4 million documents. DISCO’s High-Speed Uploader enabled the team to upload their data and begin work within a matter of hours. Using search terms and DISCO AI, the team reduced the review population to 123,000 relevant documents. Kennedys then enlisted the help of DISCO’s Professional Services team to implement an AI-prioritized review, which helped them determine that only 21% of the relevant document population were likely to be responsive. After 11 days of review, the Kennedys team was confident they had located the responsive population with a responsiveness rate of 84%, which allowed them to cross their production finish line at a record pace. (Read more about this case.)

Leave the competition in the dust

Every step of the way, DISCO is optimized to get your case team accurate insights faster and at a fraction of the cost. The DISCO Review team serves as expert drivers with a wealth of experience effectively getting the most out of the robust DISCO platform every step of the way. Leveraging optimized workflows and this lightning-fast super accurate technology has helped clients with data sets in the tens of millions and tens of thousands alike meet (and beat) their deadlines to win the metaphorical ediscovery race without breaking the bank. We can help you too. 

What else do you need to win the discovery race?
The Shortest Route
The Secret Ingredient: It's People!

Subscribe to the blog
Quick Menu