Editor’s Notes: In this post, The CEO talks about 10 workflows, the advantages of solving the 10 workflows, and what makes horizontal human-in-the-loop software-enabled services platform (like Invisible) superior to vertical (or horizontal) pure-software products.
What do BUYERS want? Dogma Reconsidered: Humans-In-The-Loop vs. Pure Software Solutions.
Have you ever read Euclid’s Elements? Thirteen books of Geometric proofs built on a handful of premises. If you accept the premises, you have to accept every proof arrived at, by logical extension. I’m doing the same thing here.
This morning I overturned the products > services dogma. If you accept that argument, then you should also accept me overturning the vertical > horizontal dogma, which I did over lunch. This afternoon, I will overturn the pure software > humans-in-the-loop dogma. Thankfully, there are enough VC dogmas to keep me going for weeks, so never fear. I won’t run out.
If you haven’t read the first two essays, read those first. Then come back. Like Euclid, I am invoking and extending my prior arguments…
Buyers don’t care whether your service is pure software or has humans-in-the-loop. They care about 1) whether it solves their problem, 2) quickly, and 3) the price.
Investors care, because of a dogmatic belief that when humans are involved in service delivery, operations become unmanageably complex, margins plummet, and it becomes impossible to price and scale. All of these things may have been true of traditional services businesses, but none of these things are necessarily true of software-enabled services.
When you have humans involved in service delivery, it immediately increases the complexity of scaling your business. This is true. It means that, to scale, you have to:
- Hire workers.
- Train your workers.
- Manage your workers. Weekly all-hands, daily standups, etc.
- Pay your workers. And deal with HR issues. Across states and countries.
- Coordinate your workers.
- Design processes for them to execute.
- Analyze performance and use insights to determine payment, pricing and optimization.
- Optimize processes, tools and coordination to eliminate inefficiencies.
- QA Check the work.
- Price the work.
That’s it. Just those 10 things. Just those 10 things?!?! Those 10 things are a huge pain in the ass! Pure software is way easier, right?!
Calm down. All 10 of those workflows can be managed by software, and at least partially automated. A software managed workflow can be very scalable, even without automation. For example, Mechanical Turk is a software managed workflow that presents an endless number of very easy tasks to a worker. Is this a cat, yes or no? Yes. Questions like that weren’t automated until humans on platforms like Mechanical Turk trained computer vision algorithms to guess. But Amazon proved that humans could execute that operation at internet-scale. And should we be surprised? Ford proved that humans could execute operations at global manufacturing scale in the early 20th century. Why should the internet be any different?
What we’re doing with Invisible is building a horizontal software stack that allows us to do all 10 of these workflows across any digital operation, no matter how custom it is, no matter how many apps, or which apps, our clients need us to operate.
This will result in such an explosion of potential use-cases we may not be able to simultaneously go after them all. So ironically, someday, we may just white-label the software and sell it to services businesses.
Curious… would that prove me right, or the VCs right?
As a reward for solving these 10 workflows, we get three overwhelming advantages: 1) we can sell productized-services, which buyers prefer, and investors will ultimately realize are superior, 2) we can sell a horizontal solution across a wide range of industries, role types and use cases, and 3) because we have humans-in-the-loop, we can deliver end-to-end solutions because we can solve more complicated problems than software alone can solve.
I predict that the history of digital operations automation in the 21st century will be an endless dialectic between knowledge workers and software. Work can be created faster than it can be destroyed. So knowledge workers will do something, then we’ll build either pure-software tools or human-in-the-loop software-enabled services to automate that thing. Then knowledge workers will do something new that’s more complex, then we’ll build either pure-software tools or human-in-the-loop software-enabled services to automate that thing. And so on, for the rest of the century…
What makes a horizontal human-in-the-loop software-enabled services platform (like Invisible) superior to vertical (or horizontal) pure-software products? Three things:
- We can solve problems at a faster rate. With a vertical strategy, you’re building a whole company around a single solution. With a horizontal strategy, you’re building a company around your ability to solve many problems with a single solution… or rather, you’re betting on your ability to adapt your solution to solve problem, after problem, after problem. In other words, vertical companies innovate only once. Horizontal companies innovate over and over again. Which do you think will create more economic value in the long run? Which do you think will get better at solving problems?
- We can solve harder problems. Without the 10 workflows above, managing humans is chaotic and frustrating: “manual” is a nightmare and brings out the worst in us. But with the 10 workflows above in place, software can bring out the best in humans, and empower us to do increasingly complex work. When humans are in-the-loop, you have a far larger arsenal at your disposal in attacking a problem. If pure-software can’t automate a workflow, the best that a vertical strategy can do is build Yet Another SAAS Tool. But in a horizontal strategy, if pure-software can only automate part of a workflow, your margins just went up. Then, for the rest of the workflow, you can build or buy a 3rd party tool to optimize your manual efficiency on the rest of the workflow, and your margins went up more. Also, you are tech-agnostic, so every single vertical company in the world is doing free R&D for you.
- Our market size is infinite. Vertical strategies are always shrinking market size to increase the speed and decrease the risk of finding “Product Market Fit” — their sacred cow. But the proliferation of tools and automation point-solutions explodes discovery, training, usage and other Coasian costs, which, far from shrinking our market opportunity, create an opportunity for a services company to sell solutions, become a best-practice authority, integrate and use all of these solutions. Eventually we may aggregate enough buying power to build a vendor network, drive discounts and more. Furthermore, “the labor markets” are as vast, if not more vast, than the capital markets. And as majority of labor moves online, our market expands. This market may be too large for one company to monopolize entirely. But that does not mean that a monopoly cannot be build around certain network effects, platform depth and industry use cases.
In the 20th century, very large and iconic service companies were built: Goldman Sachs, J.P. Morgan, Accenture, BCG, EY, KPMG, and McKinsey, just to name a few. Even one of the most iconic product companies, IBM, became a service company by the end of the century. Although all of these companies benefit from economies of scale, switching costs and other barriers to entry, none of them benefit from hard network effects and platform advantages like the technology giants of the early 21st century.
The future will not look like the past. That is what makes it the future. Anyone who wishes to actually learn from history, must understand the complex dialectics and interactions that drove past events, and guess at how they might play out.
It may be that by the end of the 21st century, we will see Service Giants as large as the Product Giants like Facebook, Apple, Netflix and Google that are the envy of the world.