Frequently Asked Questions

What is “Deep Learning”?

For context, Artificial Intelligence (AI) is a broad term. Machine Learning is a subset of AI. Neural Networks is a subset of Machine Learning. Multi-Layer Neural Networks is a subset of Neural Networks. This last subset has been called deep learning.

As the result of academic breakthroughs beginning in 2012, deep learning has enabled computers to “learn” from either images or words. Click the title for this 2015 TED talk (19:46 and over 2.4 million views): The Wonderful and Terrifying Implications of Computers That Can Learn

On the Home page, you show the User Interface. How long does it take to learn to use it?

You can learn  to use Intraspexion’s UI in less than a minute.

That’s a static image. If it’s that simple, can you walk me through how to use it, without a demo?


Referring to the Home page, see the blue line? The system takes you first to the highest-scoring email.

Step 1. Click the blue line to open the email.

Step 2. Now, for this and every other email, with subject-matter words highlighted for you, you make the decision whether the email is a True Positive or a False Positive.

Step 3. Click either the True Positive or False Positive button at the top - to save it. We’ll explain why later. 

Step 4. Click the next line to open the next email.

What about Open Email?

If you’ve found a True Positive, click Open Email to access that email in its native state.

And then?

So now, from one of yesterday’s emails, you leave Intraspexion behind. You export a True Positive to your internal matter management platform and open an internal investigation. There, you can perform email threading, access internal databases, e.g., HR performance reviews, etc.

Is the workflow confidential?

We think so. When corporate counsel enages our system, it’s doing so in anticipation of litigation. The work-product doctrine should apply. Then, when a paralegal or attorney finds a True Positive and conducts an internal investigation, the work-product doctrine should apply again. Then, if the investigation merits further action, and corporate counsel advises a control group executive, the attorney-client privilege should apply.   

Why split out the emails as True Positives and False Positives?

A deep learning engine learns from examples. Intraspexion’s engines are initially trained, for each risk, with examples from PACER litigation classifications. That’s generic. With the True (and False) Positives you’ve saved, we’ll help you customize the deep learning engine so that model is company-specific, and better reflects your company’s culture.  

How is Intraspexion’s system deployed?

On premises or using AWS (with Azure coming soon).

What technical assistance do you provide when the system is deployed?

We provide step-by-step instructions, including both text and screen shots.

Do you store any customer data?


Do you forward any customer data to anyone other than a user?


Do you have Office 365 connectivity?


Do you have any direct competitors?

Not to our knowledge.

Because of your patents?


Are they listed on your website?

Yes. See the Tab for 8 Patents.

How popular are patents for deep learning software systems?

Beginning in 2015, the # of U.S. patents has doubled or more than doubled every year. See the Tab for 8 Patents.  

Where does Intraspexion stand?

Intraspexion is in the Top 10, and is the only startup there. See the Tab for 8 Patents.

What are the key differences between what Intraspexion does and the tools used in e-discovery?

1. With e-discovery, the personnel in corporate law departments are forced to look backward for custodians of documents that are potentially relevant to a lawsuit that’s already been filed. The responsive documents are limited to the allegations and claims in the lawsuit.

2. In e-discovery, litigators are accountable to the opposing parties and to the court. But the pre-litigation focus of Intraspexion—prevention—is different, so corporate law personnel are accountable only to their enterprise.

Aren’t there other ways internal investigations can be triggered?

Yes. Someone who slams the door on the way out might say, “You’ll be hearing from my attorney.” Or the company may have a “hotline.” Of course, some companies may have hotlines but not encourage their use; or some companies may have hotlines and encourage their use, but employees may still be reluctant to use them.

Besides "employment discrimination," what other litigation threats does Intraspexion detect?

There are many business-relevant litigation categories in PACER, e.g., breach of contract, fraud, antitrust, to name only a few. We can train a deep learning model for any category of litigation pain, and we can do so in a very short time (weeks, not months). 

What's your pricing structure?

Each prospective customer has a unique litigation profile. Their prevention priorities differ. As a result, we have no set pricing structure. We customize our pricing and we keep our relationship with each customer strictly confidential.

What about the privacy concerns of company employees?

First, we care about this issue. Second, we do not monitor personal devices.

But where a “computer technology resource" (CTR) policy is in place, and an employee is aware of the CTR policy, and agrees to it, the now-settled law is that an employee has no reasonable expectation of privacy when a company monitors company-owned computers. See Holmes v. Petrovich Development, LLC, 191 Cal.App.4th 1047, 119 Cal.Rptr.3d 878 (2011) (holding that not even the attorney-client privilege bars monitoring).

For the sake of customer confidence and transparency, can you tell us anything about the software you use?

Yes. We are proud to say that our indexing and visualization capabilities are powered by dtSearch®, We also use Global Vectors for Word Representation (GloVe) and Google's TensorFlow. Both GloVe and TensorFlow are “open-source.” Both GloVe and TensorFlow are included in our system under an Apache 2.0 license, the terms of which are here and are incorporated herein by this reference. TensorFlow, the TensorFlow logo and any related marks are trademarks of Google Inc.