Community

Anything is possible with Don in charge but, as of now, I think Trump Tariffs are applicable on countries, not counties.

Can't believe there was so much fracas about One Big Beautiful Bill Act for merely $3.3 million extra debt, I think it's $3.3 trillion. 

Does it really matter?

History is replete with several examples of successful startups in regulated industries that have leveraged regulatory gaps, become too big to have to comply, and gotten regulations changed to fit their innovative business models ex post facto. More in my blog posts Fintechs Need Guts More Than Lawyers and Fintechs Need Marketers And Lobbyists, Not Lawyers. In other words, as the old saying goes, "it's better to seek forgiveness than to ask for permission"!

As Reid Hoffman once reportedly said, "Had we known about all the credit card fraud and chargeback rules, we wouldn't have founded PayPal". 

PS: (1) Not legal advice (2) Description not prescription.

Great post!

Explains the pain area with incumbent IAM systems and proposed Identity Fabric solution very succinctly. 

IMO, Invention is a new thing. Innovation is a new way of doing an existing thing that amplifies impact through broader reach. By that token, Google Glass was an Invention. But, since it didn't enjoy broad reach, it was not an Innovation.

Invention (per my definition) / Innovation (per your definition) is going outside the box. Regulation / Compliance (per general understanding) is staying inside the box. Regulation-compliant invention / innovation is an oxymoron (at the point of occurence of invention / innovation).

Some of the greatest business value has been created when startups did invention / innovation that leveraged regulatory gap - if not outright broke prevailing laws / rules - at their inception, gained scale, became Too Big To Be Shut Down, and got regulators to modify regulation ex post facto to fit the new capabilities enabled by their invention / innovation e.g. AirBnB, Uber, UPI. Which is not saying anything negative about regulators - Regulators are not Entrepreneurs. 

Yet another Spy v. Spy use case in genAI! (H/T MAD magazine)

IME, as I highlighted in my blog post Hiding Your Secret Sauce, when somebody volunteers a secret sauce, the real secret sauce lies somewhere else! 

I've been asked many times about the difference between stored procedure, CRON job, script, and RPA on the one side and AI Agent on the other. No matter how much spiel I give about genAI, some people simply refuse to believe that Agentic AI is the next level in automation technologies. Going forward, I'll describe the above incident: No stored procedure, CRON job, script or RPA has even been tasked to run a vending machine business - or bungled it as badly as Claude AI Agent!

Maybe n00bs in the payments industry won't know this but, for 20+ years:

(1) Traditional payment fraud detection and prevention systems do not always rely on after-the-fact analysis. They include algorithms like "Velocity Check", which act on individual payment transactions before the fact. To give an example, a coworker swiped his credit card at Denver airport, then boarded a flight, landed at Dulles airport, when he tried swiping his credit card, it was rejected because "how can anyone travel 2000 miles in 3 hours?" Obviously, this was a false-positive but it does illustrate the availability of realtime features in trad FD&P systems.

(2) Recommending the best payment method for the individual has been the staple of many traditional payment apps. For example, given a merchant, which is the best credit card to use depending on the highest rewards for that merchant spend category. 

(3) PSPs have provided "Intelligent Payment Routing" based on availability, least cost, etc. 

If these are all the features that AI has up its sleeve, I'm not very gung-ho about its value add for payments. 

Great post. Totally resonates with what I wrote in section titled 5. Makes GIGO Obsolete of my blog post Ten Revolutionary Things About AI - Part 1. That said, Agentic AI takes autonomous decisions based on data, so the stakes of bad quality data are arguably higher in the case of AI projects compared to previous technologies like ERP, CRM, etc. 

Nice post. 

I specify the criteria to pick stocks to ChatGPT but, after I do that, ChatGPT does its own thing including access whatever sources it accesses. Given that workflow, keen to know I can "Be sure your sources are compliant and transparent".