Keyword Spotting

Artificial Intelligence Strikes Back …

Almost 20 years ago, I started an ambitious artificial intelligence software company — International Neural Machines (INM), specializing in neural network based pattern recognition. One of our most successful applications aimed at speech recognition and in particular — at “keyword spotting”. With the support of Canadian Police Research Center (CPRC) and National Research Council — we developed a very accurate recognition of specific words within the unconstrained audio recording.

Our app stemmed from a real need for a solution to a serious problem. CPRC was quite impressed with our technology and alerted me to the following:

1) In most cases, a police cruiser would rush to attend the accident — immediately after being dispatched

2) However, in case of a chemical spill, the cops would first open the trunk of a car — and reach for a thick operations manual, instead

3) Our objective was to build a product that would recognize the keywords such as “Chemical Spill” and retrieve the pertinent information from a manual

4) No time is wasted, as the cop is being debriefed through a speech synthesizer — while rushing to the scene of an accident


To turn the prototype into a successful business offering, I identified Verizon as a key partner and a key distribution channel. At the time, the company was known as Bell Atlantic.

They liked our innovative technology and arranged for a demonstration to one of their clients in Maryland — a maximum security prison.


All prisons at that time had a big problem:

a) The inmates were only allowed to make collect calls from within the prison — using a small number of public phones installed at each institution

b) Since many of the inmates used jails as “an office without the overhead” — all the outgoing calls were recorded to intercept the illegal activities.

c) Unfortunately, each prison ended up with hundreds of hours of audio recordings daily — but they had no manpower to analyze the calls and identify crime related activities

Verizon saw the opportunity and proudly invited us to make the presentation to the warden. Make no mistake, it took almost one hour to pass the security checks — which included the dog-sniffing exercise and the removal of all the cell phones and watches at the gate.

Our presentation went well and the warden was genuinely impressed. He understood that our software would automatically scan all the daily audio in minutes and mark the suspicious segments for an offline review by the prison staff. The productivity gain was enormous and the ability to zero-in on a criminal information — uncanny. One could review all the daily suspicions in 20 minutes!

Finally, the warden asked us for the price and wanted to know how much would it cost to install and maintain the system by a trusted telco provider — such as Verizon. Our joint response was that it would cost ~$500K for a one-time site license. That was it!

To my surprise, the reaction of the warden was not what I expected. In summary, he said the following:

a) Yes, he loves the system and the proposed efficiency gains

b) However, $500K is a lot of money, and that he could bring a lot of dogs and add many more guards to the prison, instead

c) So the answer was: thanks, but no thanks, guys — better luck next time…


Following this unanticipated roadblock, we re-grouped. It took us about 3 weeks to convince Verizon that instead of asking for an outright purchase of the system (using the capital budget) — we could offer the same system in exchange for a monthly installment of ~$8K/month. We also assumed, that the monthly installments could be easily covered from an operational budget of the prison — without the need for a budget appropriation request.

The second meeting with the warden took place shortly after. The dogs, the watches and the cell phones were all left at the gate again — and we proudly presented the benefits of our value proposition to the warden. Although intrigued by our flexibility, the warden was still haunted by the long-term effect on the number of … dogs and guards in his jail — and sent us packing, politely.

We left the jail and decided to go over the outcome of our visit at the coffee shop nearby. Verizon succinctly summarized the obvious:

a) Look, we tried kicking the can — not once, but twice

b) If our proposal doesn’t work at the maximum security prison — it won’t work at any other prison, either. End of story

However, I was not ready to give up yet, and asked Verizon the following questions:

1) Are collect calls made from the public phones inside the prison PROFITABLE to Verizon?

a. The answer was a decisive: yes! These are the most profitable charges per minute that Verizon ever gets. And guess what: there is no ALTERNATIVE to make a call from a prison!

2) Since I noticed only 3 public phones on a wall and a significant prisoners lineup behind them — would 150 phones on a wall make a big difference?

a. Once more, the answer was a resounding: yes! The collect call revenues would skyrocket!

3) My last question was: how long would it take to pay for our keyword spotting system if Verizon was allowed to install all the 150 phones at the jail?

a. A quick back-of-the-envelope calculation returned: 8–10 months

We had a winner! I asked Verizon to call the warden again and ask for a third visit — the same afternoon. The warden was visibly unhappy and greeted us with a terse message that we only have 5 minutes to make the case and that he has no desire to further discuss our value proposition — ever again.

My proposal was short:

a) I offered installing 150 phones at the prison and reduce the big lineups — a significant security risk in a first place

b) At the same time, we would also install our keyword spotting solution at no cost

c) In addition, we would even … share the 10% of Verizon’s collect call revenues with the warden


Undoubtedly, we took the warden by surprise. He repeated twice my offer — to ensure that he got it right. Then he reached out for the phone on his desk and called the …….. Governor of the state of Maryland. To my astonishment, his message to the Governor was quite simple:

· Look, Verizon is at his office and just offered to him an unbelievable deal

· He is convinced that immediately after leaving his office, Verizon is going to visit additional jails with a similar offer at hand

· All he wants — is to be the first in line and that the Governor owns him such favour!

The Bottom Line:

· It doesn’t really matter what are you trying to sell and to whom

· It also doesn’t matter how well you can articulate what is it that you offer

· The only thing that matters is: WHAT’S IN IT FOR THEM

Remember: true innovators see failure as a mere stepping stone to success. It took Thomas Edison almost 1,000 tries to develop a light bulb that actually worked …

“Success is stumbling from failure to failure with no loss of enthusiasm.” –Winston Churchill

Oleg Feldgajer is President & CEO of Canada Green ESCO Inc. Oleg is positioning the company to become a leader in financing AI enhanced green energy projects and ventures. CGE’s mission is to guide DISRUPTIVE businesses in ENERGY & TRANSPORTATION toward profitable business models. Oleg is passionate about such mission, and firmly believes that without AI based innovation, we will all prematurely choke on polluted air and dirty water. CGE delivers 100% financing (levered and unlevered) to its clients — and utilizes large equity pools, and non-recourse debt. Oleg offers creative, fresh ideas to open-minded businesses — that embrace both: logic AND opportunistic intuition. CGE stands against mediocrity & its modus operandi is quite simple: If CGE is not invited to join your BOD, or Advisory Board — we failed!



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store