I
have never considered myself a technologist in the conventional sense.
I
don’t build software or write algorithms.
But
I have always been fascinated by one question:
“If technology is evolving so rapidly, why should the user have to
struggle with the complexity?”
Looking
back at some of my old notes, blogs and correspondence, I discover an
interesting thread running through almost two decades of my thinking.
The
ideas are different.
The
technologies are different.
The
years are different.
But
the question remains remarkably similar:
Can we remove the friction between a person and a powerful new
technology?
Here
is my little journey.
My first stop on this journey goes back to 21 December 2005.
At
that time, conventional job portals were already well established.
I
had written to Ashish Kashyap, then Country Manager of Google India,
suggesting the possibility of Google integrating jobs into its search
ecosystem.
My
basic thought was simple:
Why should a job seeker have to visit a separate job portal merely
to find a job?
Search already knew how to find information.
Why
not jobs?
Years
later, Google indeed entered the jobs-discovery space.
I
subsequently wrote about the idea in:
url Google
Jobs — my earlier blog
post http://hcpreports.blogspot.in/2016/02/google-jobs.html
I
certainly don’t claim that Google’s later development was the result of my
suggestion.
But
the episode taught me something important:
Sometimes
it is simply an existing technology being applied to remove an unnecessary
intermediary.
The next significant episode came with the emergence of BHIM and
UPI.
India had created something extraordinarily powerful: a common
digital-payment infrastructure.
But I began wondering:
Could BHIM become simple enough for people who were not comfortable with
smartphones, menus, typing and conventional digital interfaces?
My answer was:
In April 2017, I wrote:
BHIM — The
Unstoppable
https://myblogepage.blogspot.com/2017/04/bhim-unstoppable.html
I also wrote:
Let BHIM Get MAD
https://myblogepage.blogspot.com/2017/05/let-bhim-get-mad.html
and:
A Beerud Named
BHIM
https://myblogepage.blogspot.com/2017/04/a-beerud-named-bhim.html
My argument was not that voice was fashionable.
It was that simplicity could be the key to inclusion.
Millions of Indians, particularly in rural areas, might find
speaking much easier than navigating a complicated digital interface.
I took the idea beyond my blog.
On 16 June 2017, I wrote to Shri A. P. Hota, then
Chairman/MD & CEO of NPCI,
suggesting:
“Could the next improvement in BHIM be a ‘Voice Activated’ version —
BHIM V 4.0?”
I referred him to my blogs and explained why I thought voice could
bring millions
of rural/less-literate Indians into digital payments.
Shri Hota replied the same day and asked NPCI’s product-development
executive,
Vishal Anand Kanvaty, to get in touch with me.
That led to the next step.
Four days later, on 20 June 2017, I met Aseem Chaturvedi
of NPCI to discuss the idea.
On the following day, 21 June 2017, I sent him a detailed
follow-up email.
The meeting was particularly useful because it brought the idea down
from the
level of imagination to the level of implementation.
Aseem’s concerns were perfectly reasonable.
For Voice-Activated BHIM to become practical, several pieces of the
ecosystem
would have to fall into place:
·
merchants would need suitable
biometric-enabled smartphones;
·
banks serving rural customers
would need to participate;
·
privacy, encryption and
security concerns would have to be addressed;
·
BHIM itself would first have to
gain widespread acceptance;
·
and, perhaps most importantly,
there was not yet a visible “clamour”
from potential users demanding such a
facility.
I agreed with the practical difficulties.
But I made one observation:
If development itself could take several months, perhaps we should
begin before the demand became obvious.
That has remained one of my favourite principles of technology
foresight:
There was another part of that 21 June correspondence that now looks
particularly interesting.
I quoted Shri Amitabh Kant, who was then speaking about
India’s digital-payment ambitions.
One of the problems he identified was the:
“multiplicity of apps”
BHIM had been conceived partly to bring clarity to that fragmented
environment — allowing customers of different banks to use the same common
UPI-based payment system.
At the time, I was also watching messaging platforms such as Hike
and WhatsApp moving toward payments.
The larger trend was becoming visible:
Why
not make one common infrastructure work underneath many different consumer
interfaces?
I
did not have the terminology for it then.
But
the underlying idea was already:
Fast-forward nine years.
Artificial Intelligence has exploded.
We now have major AI providers — OpenAI, Google, Anthropic, xAI and
others — while thousands of applications use their APIs to deliver specialised
services.
And suddenly I encountered a new version of an old problem.
Suppose I use:
·
an AI writing service;
·
an AI image generator;
·
an AI video service;
·
an AI presentation maker;
·
an AI coding assistant;
·
an AI research service;
·
an AI transcription service.
Many such services may require me to:
Register → create an account → add payment details → purchase
credits → use service → run out of credits → recharge.
Then I move to the next service and do it all over again.
I found myself asking:
Why?
So I proposed another apparently crazy idea.
Suppose I maintain:
OpenAI Wallet — ₹1,000
Google AI Wallet — ₹1,000
Anthropic Wallet — ₹1,000
xAI Wallet — ₹1,000
Now I visit an AI-powered website.
It charges ₹18.
Instead of creating another account and buying another set of
credits, I simply see:
PAY WITH
OpenAI
Google
Anthropic
xAI
I select my preferred AI Wallet.
The website receives its ₹18.
I receive the service.
That
was the seed of what I began calling:
I
am not suggesting that AI companies should merge.
Nor
am I suggesting that OpenAI, Google, Anthropic or xAI should surrender their
individual billing systems.
The
more realistic proposition is an:
The
analogy with UPI is conceptual.
UPI
allows different banks, payment apps and merchants to participate in an
interoperable payment ecosystem.
Similarly,
an AI-payment protocol could allow:
AI Provider Wallets
↕
AI Applications
↕
Consumers
to
interact through a common payment/authorization layer.
The
AI companies could continue competing fiercely on:
·
intelligence;
·
price;
·
speed;
·
reliability;
·
models;
·
multimodal capabilities;
·
developer ecosystems.
But
the payment interface could become interoperable.
And then, almost as if the universe wanted to encourage the thought
experiment,
today’s newspaper carried a fascinating development.
Mint reports that NPCI is preparing to
make UPI AutoPay interoperable,
allowing consumers to port existing
AutoPay mandates across UPI apps and
enabling merchants to move existing
mandates between payment providers
without forcing customers to register afresh.
The article describes the objective in terms of moving:
from a closed-loop system to a “many-to-many” network
and reducing consumer and merchant lock-in.
That immediately caught my attention.
Because this is precisely the architectural principle behind my
AI-UPI thought
experiment.
The problem:
Too many apps and fragmented user
experiences.
The answer:
Interoperable payment
infrastructure.
The emerging problem:
Too many AI applications and
fragmented payment/credit relationships.
The possible answer:
Interoperable AI payment
infrastructure.
The common principle is:
Looking back, I see four milestones.
Google Jobs
Why should job search require a
separate destination?
↓
Voice Activated BHIM
Why should digital payments
require a complicated interface?
↓
AI-UPI / Open AI Wallet
Why should every AI application
require a separate payment relationship?
↓
AI Agents + AI Payment
Why should an AI agent need to
ask me for payment credentials every time it
wants to buy something on my
behalf?
The AI Wallet idea may ultimately be about much more than avoiding
the nuisance of buying credits.
Imagine an AI agent saying:
“I have found the service you need. It costs ₹17. Shall I proceed?”
I say:
“Yes.”
The agent authenticates me.
The merchant is paid.
The service is delivered.
My financial credentials are not handed over to the merchant.
The transaction appears in my AI-payment history.
That could become a fundamental requirement of the agentic AI
economy.
Because tomorrow, AI may not merely answer questions.
And
if they do, AI identity, AI authorization and AI payment will have to
work together.
I don’t think the lesson is:
“I predicted the future.”
That would be too grand a claim.
Ideas are cheap.
Implementation is extraordinarily difficult.
And many ideas that sound brilliant on paper never work in the real
world.
The more modest — and perhaps more useful — lesson is this:
The
technologies may change.
The
companies may change.
The
interfaces may change.
But
the friction remains.
And
whenever I see:
“You need another app.”
or
“You need another account.”
or
“You need another wallet.”
or
“You need to register again.”
I
instinctively ask:
Looking back over 21 years, I see the same instinct recurring:
Remove the separate job-search
destination.
Remove the complexity of the
digital-payment interface.
Remove the fragmentation of AI-service
payments.
The technology changed from:
Search → Mobile → UPI →
Voice → AI
But the objective remained
remarkably constant:
Perhaps that is what I have really been trying to do all these
years.
Not predict the future.
Just ask uncomfortable questions a little before the future arrives.
In 2017, I wondered
whether a voice-activated BHIM might someday make digital
payments accessible
to people who found conventional interfaces difficult.
In 2026, I am wondering
whether an interoperable AI Wallet might someday make
the rapidly expanding AI
economy accessible without forcing every user to
maintain dozens of separate
financial relationships.
I don’t know whether AI-UPI
will ever exist.
Perhaps the AI industry
will invent something completely different.
But today’s move toward
greater interoperability in UPI AutoPay has reinforced one
belief I have
carried for a long time:
— Hemen Parekh
31 August 2026
===================================================
References :
Extract :
Sent : 16 June 2017 16:06
To: Hemen Parekh
Cc: Contact NPCI; Vishal Anand Kanvaty
Subject: Re: BHIM V 4.0
Thank you Hemen ji.
Vishal, our SVP product Development will get in touch with you.
Hota
