Comment by Accounts Recovery on March 8, 2011 at 4:25pm

What follows is a release from Ranjan Dharmaraja, CEO of Quantrax, on this Jeopardy news:

Should collection managers be worried? Yes, because Quantrax says its collection technology is smart, like IBM’s Watson that played Jeopardy!

 

In February 2011, in a much-hyped, nationally televised show, IBM’s Watson supercomputer comprehensively defeated two human champions in the intellectual game of Jeopardy. Quantrax Corporation says they were not surprised, having created technology that is smarter than the best collection experts!

Everyone is claiming that they had something to do with Watson, IBM’s creation that took and on and beat two of the very best Jeopardy champions. Novell, Apache Software Foundation, Java, Prolog, C, C++ and IBM’s clever software – they were all a part of a historical experiment in testing machine intelligence against human experts. And suddenly, everyone is thinking about the practical application of this powerful technology. 

Quantrax Corporation watched with interest, as the contest evolved. The key hardware that IBM utilized was the midrange Power 750 server. This happens to be the same family of computers that Quantrax uses for its artificial intelligence-based collection platform called RMEx (formerly Intelec). One of the best-kept secrets in the industry, this intelligent software was developed by Quantrax in 1991. Its complete collection suite is deployed in the US and Canada, in operations with 20 collectors, to enterprises with over 1000 staff.

“Other vendors have realized that you can not compete with computer systems that can make complex decisions. They are adding workflow modules, rules engines and decision-making to their products. Our response is that imitation is the sincerest form of flattery”, quips Quantrax’s CEO Ranjan Dharmaraja. How do users feel about the technology? Bill Hopkinson the CEO of CBC Inc. in Charlottesville VA says “The whole idea behind the system is to force the newest collector to work an account in exactly same way as your best and most experienced collector (or as each client has mandated). It is very powerful software. I have not found a process we could not handle.”

As with any exceptional technology, there will always be skeptics. Many will insist that their software is as good as any conventional technology. Try telling that to Ken Jennings and Brad Rutter, the human Jeopardy champions (When the contest was over, Watson had $77,147 to Jennings’ $24,000)! Steve Florczak who leads the technology group at Capital Management Services says, “We have some excellent managers who routinely review a few thousand accounts each day. But they are no match for our technology team and a system that analyzes and makes strategic decisions on several million accounts, every single day. Many years ago, we had people who would create dialer campaigns for the day. No more. We now have the accounts we need in the right campaigns, presented to the right people, at the right time, with the right phone numbers - And it is all automated.” Quantrax insists that you cannot take conventional software (data-based systems that other vendors market) and make it intelligent by changing code. Intelligent software is different and has to be created in a different way. The key difference is a knowledge base (that stores many complex rules, along with the thinking and possible decision-making algorithms for each company) and its link to a powerful software program called an “inference engine”. Combined with user interaction, the technology is smarter than the best collection experts you can find - In 20 years, that has been proven, many times over. The result? You can do more work with less people. Bill Hopkinson added “I have heard of many owners who ‘brag’ about all of the custom programming they do to their systems to make it just the way they want it.  The difference is, since we can already to it ourselves, we save time and pay very little for custom programming!”

Quantrax believes that pressure on the collection industry to find better ways to spend less and do more, will force many companies to look for better technology. Today, we have the advantage of more powerful hardware than we have ever seen, but are we doing enough? What Watson did in February, provides the answer. That answer lies in the software you use.

Quantrax Corporation was started in 1988, and has been in collection technology since 1991. They are headquartered in Bethesda, MD, and can be contacted though www.quantrax.com

 

 

 

 

 

Comment by Jeremy Mapes on March 8, 2011 at 7:44pm

Interesting conversation regarding AI, as to some point.  In my experience, working with some former military AI programs that were converted to work on business processes at some of the former companies I worked for, I found that AI is not AI like normal people think.  It is really just putting several attributes together to attempt to determine a guestimate of the outcome so that the staff can make the best decision, possibly whether even to touch the account in the first place.  More people are familiar with the outsourced version of this referred to as "scoring".  Scoring takes into account several variables that attempt to determine the outcome of an account all packaged into a neat little bow.  More advanced companies get a credit score and do some of their own decisioning based on the attirbutes of a score.  Software that is more open to user-defined windows and calculations and importing and exporting can better handle this type of work. 

 

If the Quantrax system allows the users to do their own internal "massaging" of mass data points this would be a nice addition rather than paying somone exteranlly to do that for them or, with some collection systems (you know who you are) take a piece of the action for themselves as the files get sent back and forth.  So now you're paying one group instead of three to accomplish the same results. 

 

I judge systems based on three major concepts:

1.  Flexibility - What can you do with the system yourself, as a company, without having to pay the manufacturer of the software for professional services.  In some ways that's like a software company saying "we know our software stinks so we're going to charge you extra money to fix our mistakes".  A good system is an open system that allows the users to customize the product and make their use "unique" without having to pay a fortune for it.

 

2.  Relevance or ability to keep up with current technology - Does the system you're working on have the latest technology in it?  Did the dialer you buy break out three-tone intercepts into the subcategories (disconnected no new number, disco' new number, temporarily disco') or is it treating them all as one supercategory?  Can it connect via VoIP to the telco so that you can condense your connectivity charges and how does it perform (as a centralized unit or can it be broken down to multiple sites)?  Does the software have workflow automation that can incorporate more than collections (like building your own new modules for things like PPMS, security governance, state tracking, suit process)?  Limited technology brings limited results.  Old technology can be more costly.  Compare the cost of a Unix box or an AS400 to a new Windows box.  Is it worth the extra cost?  Does the system integrate with a dialer or is the words "upload/download" part of the equation?

 

3.  Cost - Does the software vendor charge you an initial sales plus a continuing "service charge".  Do you "own" that software if you and the vendor part ways?  Does the software company live off of professional services (run, don't walk),   What does your support cost pay for (do they guarantee they'll release all custom code in their new revisions or are they going to hang that on you as an "extra" cost)"  If they host the product what are your rights and what do they guarantee and what are the penalties if they have a breach or miss their commitment?

 

I once new of a software for the collection industry that was a rapid application development tool at heart.  You could make it into anything you wanted.  There wasn't really any limitations and it hitched right into a dialer directly.  Unfortunately they lost their person that had all the direction in their head and they started building walls that lowered the flexibility and then wondered why the product failed.

Today the closest I know to an open system with a dialer would be something like Interactive Intelligence and their new BPA tool that gives you a database, a dialer (and a phone system), and four other forms of communication all bundled together and ready to be altered to the specific need of the company.  Short of a full blown collection system I'd use that all day long.  Anyone else find any great dialer/communication integrated BPA products lately or a collection system that can come close to that? 

 

I don't know if Quantrax does all of that but I'll cheer any software that does.

Comment by Ranjan Dharmaraja on March 9, 2011 at 1:50pm

Jeremy,

Let's keep this simple. This debate is not about open source, versus proprietary systems. In our opinion, there is great merit to good, proprietary systems that work. Those who cry out for open source, often do so because their systems do not do what they want them to. Collection companies are in receivables management, not software development. Their software vendors should provide the solutions and flexibility they need.

What we have is a proprietary system that does something every company needs. It allows you to clone your best collection managers and then have the system make complex decisions on millions of accounts, based on that expert's knowledge. Imagine replacing every collector in your office with your best collector? What would happen? Your collections increase a great deal. No one will disagree. Why? Because of better decision-making. An AI-based system will allow you to put your best collector's thinking behind every collector. They will have to think less. They will not need to try to make complex decisions that they are simply not able to make on thousands of accounts. The system will make account management decisions for them. And you could be out playing golf - the decisions would still be made.

That is what this is about. The industry is not what it was several years ago. Average recovery rates may be less than 10% for all types of debt. 90% is not collected and we have a difficult engineering problem, not a collection problem! Traditional designs do not work. You need software to help you, and we are proud to have designed these systems 20 years ago. Who would you have as your Jeopardy partner? Watson, or Jennings and Rutter? What type of software would you want in your collection operation? Software than can think, or software than can not think?

Comment by Jeremy Mapes on March 11, 2011 at 11:08am

For the jeopardy partner I''ll take Ben Stein...  LOL.

I wrote two kinds of programs years ago.  One was a program that "thought" and to do so we had to inject random chance on top of probability to make it so that it was not always predictable.  Thought was not what I want.  I want "probability" followed by "best guess" as a close second. 

I assume you're not saying your system literally thinks like a human that can not be dependable.  True AI is not applicable for business use in most scenarios but I think people are trying to use that term synanomously with high level probability processing which is factor based, not interjected with random chance.  I think you're saying your system takes into account multiple probabilities to derive a "best solution" based on where someone has decided what is valued to give it a base line.  I was actually complementing your system being able to bypass having to draw on external scoring to make decisions on what to do on accounts.  Most systems are using external scoring systems to accomplish because their system is not programmed to derive probabilities or what you called "thinking".

 

Regarding proprietary systems...  I have no problem with a system costing money.  I have no problem that the system is not MS SQL, which is about 90% of the systems out there now (I don't have a problem with them being MS SQL either).  I do have a problem if I as a customer can not get my data in and out without the help of the maker.  I do have a problem  if I cannot take over the system or easily find someone to take over the system if the software company goes out of business. 

There was a hospital software system (probably more than one) that a hospital client of mine was using up in Wisconsin several years ago.  They get a letter one day.  Their software company was closing and they had no future for the platform.  The hospital ended up paying a million dollars for their next platform and that was not in the budget when it came up.  They also had a nightmare finding a former programmer from the company to extract the data because it was on some proprietary type database.  So ergo my concerns.  I have no idea how proprietary your system is so I wasn't trying to poke at you.  AS400 is somewhat hard to find programmer for and brings about good old EBCIDIC that no one uses any longer, hardly but my only concern is that the comparable cost to a standard server is quite a bit more and so are the programmers.

 

I am really open to all systems that have have flexibility and provide the tools to the user to self maintain and correct and improve their system.  I think I can build a better workflow for my needs or my clients needs than a software company can anticipate but that they can program for someone like me or the user can modify to fit perfect.  I do cheer some of the newer systems I've seen coming out.  Software like Beesion and CRS Titanium and Interactive Intelligence and Collect! (I know I missed one or two others so please don't assume this is the entire list) have either demonstrated new flexible software or are planning on making their software flexible that will allow the receivables company to adjust the software to fit their business and also to be able to very flexibly push data in and out and to vendors and to dialers and other communication systems.  Most agencies are diversifying or specializing.  For instance they add a first-party services segment or a cleanup project or a messaging service or a insurance follow up campaign.  The ability to change workflow, add fields, automate processes, and move data becomes quite important.

 

We haven't even touched base on other required needs by systems.  New security demands by clients, government, and vendors (VISA), require encrypted fields, a way to lock out international access, and, for some clients, they require their vendors and the their vendor's vendors to have a real credential.  Many go beyond SAS 70 (now going away) and require a mix, like PCI and ISO 27001 certification.  I could probably lob off ninety percent of the companies by having only those with ISO 27001 certification raise their hand but I won't go that far because it's not required by everyone... yet...

 

I'd be happy to evaluate the strength of any company's software, from an unbiased standpoint, regarding their competitiveness in the industry.  If they've got a superior product they should step forward and be judged.  I can tell you I know that no software is perfect and there is always room for criticism.  But that's how companies get better.  I'd assume you've done some improvements to your software over the past 20 years as well.  Most companies change their entire platform in 20 years because you get into spaghetti code language back in those days and prior that's normally less effective to maintain.

 

Regards

Comment by Ranjan Dharmaraja on March 12, 2011 at 4:06pm

Jeremy,

Appreciate the questions and thoughts. I agree with most of your comments. As for our system, it will do most of what you refer to. The idea is to capture all you knowledge and store it in a knowledge base. You can store :

• work plans,
• how the system should think when different events occur, or when accounts are worked
• business rules
• state laws
• what actions should be taken now, or in the future... etc. etc.

The list can go on and on. To be intelligent, a system must be able to do almost anything you would want to do, without any program changes. That is what we offer. It is pretty amazing, but then, Watson was pretty amazing too!

With regard to some of your comments, we are in our 4th generation of this intelligent technology. Yes, we have made some major revisions. We believe that major revisions are called for at least every 4 years for application software such as collections. This may involve major revisions of the database. With intelligent software, you are dealing with a different architecture. You do not have to change code to offer flexibility. Users can make changes by changing the knowledge base and their work plans. (Trust me - there is much more involved than it sounds like!) We do not have to write the same type of code like other vendors, to keep up with changes that clients (or the industry) may require. I was hoping that people would be curious, because we sincerely believe that our "machine thinking" is years ahead of any of the other companies you mention!

 

With regard to your comments on the iSeries (old AS/400 platform), most of our clients do not need programmers! The system does what they want. If they want to extract / manipulate their data, they can use query tools or our BI / reporting platform. The iSeries has an integrated DB2 database. And you know that  DB2 is not a proprietary database. As for hiring programmers, you can program in a variety of languages on the iSeries. I bet you did not know that the iSeries supports C, C++, Jave, Perl SQL and PHP (among other languages)? And as for iSeries being more expensive compared to other servers - When you consider the integrated database,  reliability, single point of contact and ease of support - the iSeries can have a surprisingly low long-term "cost of ownership". 

I hope that was useful information. Check out our web site www.quantrax.com. The product section has a detailed product demonstration. 

 

Comment