-Mankind has always been fascinated by automation
-The latest stage of automation was no-code, the next one is AGI
-The bridge between the 2 is a multi-modal version of no-code leveraging generative AI's intelligent capabilities
-Our mission at Argil is to put AI in the hands of everyone by creating this bridge, that we call hyper-automation
At the height of the 18th century, while the American revolution was in full swing and a 17 year-old kid named Wolfgang Amadeus was writing his symphony N°25, a man was scouring a warmongering Europe with a curious machine. According to its Hungarian inventor, the wooden box was able to play chess all by its own, at the highest level.
For decades, the so-called “mechanical turk” fascinated European Aristocracies - it went as far as defeating Napoleon and Benjamin Franklin. Of course, it was all a ruse: a human was encased in the box, making the moves.
This was 50 years before Ada Lovelace was credited with inventing the fundamentals of computer science and 150 years before Asimov published the first novel of his “Robots” series.
What remains, other than an amusing cocktail story, is a testament of our ancestor’s fascination for automation.
These geniuses, preceded by illustrious mathematicians and succeeded by modern geniuses, all yearned for a world where machines would do the heavy lifting while humans dedicate their time to creativity & science.
Most scientific breakthroughs in history, and especially during the industrial revolutions, directly impacted the amount and difficulty of physical labor that humans had to endure.
The birth of modern computing seven decades ago allowed us to send people on the moon. Then came the internet, massively accelerating knowledge sharing and learning. Finally, Steve Jobs unveiled the smartphone and launched the appstore, further increasing the accessibility and portability of knowledge… an thus, pocket-sized automations.
Each of these iterations freed most of us from the need to understand a layer of technology. The first computer users had to understand how a chip worked, the second generation had to understand MS-DOS, the third had to go deep into their settings to play counter-strike with their friends, and now anyone can do pretty much anything with a phone the size of their hand.
Yet, in the 10 years that succeeded the mobile boom… nothing happened. What would be the next stage of automation?
The answer was not a technological breakthrough, but once again, an accessibility breakthrough: no-code.
See, the luxury of creating websites and building software once was exclusive to highly skilled and intelligent developers. They were the first stage of any SaaS, marketplace, on-premise software, internal process, or even “internet business”as we used to call them.
No-code was a game-changer: the emergence of several softwares (Airtable, Zapier, Make, Bubble, Webflow, etc.) empowered non-coders to build websites, create apps & tools, and automate many of their time-consuming tasks without depending on expensive developers.
However, vast amounts of skills still required basic intelligence - and sadly, any form of intelligence still had to be humanly developed. Low chaos but limited potential.
As the dust settled after no-code took the market by storm, some things still felt out of reach: horizontal skills were non-transferable. a writer still needed an illustrator - and vice versa ; writing for business or content still required human-levels understanding of the audience and context ; analyzing complex databases & excel documents was a feat that only data scientists & CFO could achieve.
Since the machine didn’t “understand” what it had to do, many alienating tasks still needed to be done by humans such as proofreading, writing, illustrating above basic levels, sorting & labelling data (ironically, most of which was done on “Amazon turk”, which name derives from our example).
Most importantly, the machine wasn’t endowed with conversational skills. Either the result is what is expected, or it isn’t - it had no ability to self-criticise or improve in real time.
And then come chatGPT - and everything went upside down. Suddenly, the machine had a human-level of understanding in many aspects. It could review its answer and improve them, teach someone, correct someone, and most importantly grant people new skills: illustrators can write a compelling story and writers can illustrate their stories.
Examples abound of tasks that took hours and that can now be done in seconds: translating, rephrasing, writing of any form, illustrating, imagining, analysing data, proofreading, teaching, coding, etc.
Not only can chatGPT and mid journey do that, but they can do it nearly for free (with prices still dropping thanks to competition). Anyone on earth -with the exception of exceptionally poor countries- can have a highly skilled tutor/ marketer/ salesperson/ proofreader/ illustrator for under $20 a month. It is truly unbelievable.
With gen AI, we’ve unlocked a new stage in our timeline:
(Where we are now)
AGI, or “artificial general intelligence, designates a digital form of intelligence that is capable to learn and improve on its own. AGI would thus be able to resolve nearly all human intellectual problems, and if given access to robotics, any form of problem at all.
AGI seems inevitable - the only question that is fiercely debated by researchers & entrepreneurs is “when”. Regardless, there’s a whole new set of tools to bridge the “No-code” generation and AGI.
We call that stage: hyper-automation.
In order to grant an improvement representing at least one order of magnitude, we believe several criteria must be met:
With this, we have a unique opportunity of putting AI in the hands of everyone. But for that, we must conquer the adoption curve.
Every new technology follows the adoption curve: going from the innovators (your friend that’s always showing you new stuff) to laggards (you family member that still doesn’t have a smartphone).
So far, generative AI has been mostly constrained to innovators and a few early adopters.
For a couple of months, only the very early technical adopters & passionate creatives were using AI. The equivalent of people using Midjourney one year ago are people experimenting on Runway today for video. Many people use chatGPT for light usage, but it is not yet a true part of their routine, as it lacks flexibility.
How do we unlock the next stage? The answer is simple:
By building tools that are flexible, simple to understand, with a sleek UI, moderately priced (but insignificant compared to the added value), that leverage YOUR data. A way to build an almost fully integrated mini-software in minutes.
This is what we want Argil to be, and more. The first stage of our platform, a sandbox that allows training image and text to one’s data and creating multi-modal workflows, will benefit from a growing community of builders that will in turn help us build more powerful features. A couple clicks allows to:
And these are just the visible part of the iceberg.
Our overarching vision is simple: allowing anyone to build their own digital intern in a few clicks, and spend their energy on what matters: creativity & added value.
We’re just getting started - let’s conquer that curve.