Blend's Position on AI

The needs of American War Machine in the 1860’s put a tremendous pressure on the manufacturing base to produce war materiel more consistently, reliably, efficiently, and cheaply. Every gun needed to work interchangeably with every bullet and bayonet. Over the decades, this led to factory processes where workers were no longer doing piecework, they were instead completing one step in a long process along an assembly line. The expertise was no longer held within the individual worker, it was instead built into the system. No one person knew how to make a Model T, and yet two million of them rolled off of Detroit’s assembly lines in 1924, at a cost of $290 each (about $5,400 in today’s money).

The Industrial Revolution primarily affected manufacturing, but by doing so it changed almost every aspect of western culture. A skilled wagon builder in 1860 must have found his job to be unrecognizable only fifty years later. The advances to technology and process created a sea change that touched every aspect of how the average American lived and worked, and every company at every level, and not always for the better.

Around the start of my career, we experienced a similar revolution — this time, to the flow of information. The rapid evolution of the Internet from a university-driven testbed to a public global communications network aimed at connecting every major economy on Earth. And, while the Industrial Revolution changed the face of manufacturing over a few decades, the already growing presence of computers and telecommunications infrastructure helped change the face of communications over only a few years. In 1991, there were 10 websites on the Internet; a decade later, the estimate was at 40 million. The jobs of knowledge workers, journalists, and many others are today unrecognizable to their counterparts from 20 years ago. As with the Industrial Revolution, not all of this change has been positive.

How infrastructure affects the speed of change.

We’re on the cusp of another sea change now. While it may not be true intelligence, large language models (LLM), intelligent agents, and general purpose transformers are able to distill vast sums of knowledge down to a conversational interface that can answer with human-friendly responses and even take actions on our behalf.

What the Industrial Revolution did for manufacturing, and the rise of the Internet did for access to information, AI will now do for knowledge, experience, and expertise. And while the Industrial Revolution took decades, and the Internet took years, AI is likely to take only months.

Part of this speed is due to existing infrastructure. In the 1800s, part of the more gradual change was because infrastructure was being built alongside the change itself. They needed to buy and build new equipment, use new power sources, and erect new buildings. The growth of the Internet required some new infrastructure — networks, modems, and etc. — but it also relied on existing computers, helping ease change along with a bit more speed. For this new wave of artificial intelligence all of the infrastructure is already in place; equipment changes won’t be a hindrance to adoption. The next major announcement from OpenAI, or Anthropic, or Google, could make major waves for entire industries within a quarter or two.

Blend’s position on artificial intelligence.

With all of the change upon us, and the uncertainty that comes with it, it would be easy to feel like many of those coopers and saddle makers must have felt in the early 1900’s: “let’s put our heads down, do what we’ve always done, and hope this is all a flash in the pan.” This tendency was certainly evident in the early Information Age, and I remember a lot of discussion on how the Internet would never really become mainstream.

At Blend, as a company, we’re taking a more pragmatic approach: right or wrong, love it or hate it, you can’t ignore the fact that change is happening. So our choice, as developers and as an industry, is to either master this change or be left in its wake. I have little doubt that by 2028, the tools we use, the jobs we do, and the things we build will look unrecognizable to today’s digital agencies.

With this in mind, Blend chooses to see progress within the AI space as a force multiplier, in which designers, marketers, and developers combine their knowledge and expertise alongside a team of virtual agents. These agents take care of the repetitive and rote portions of their jobs, allowing professionals time to focus on the creative and innovative components.

We are not alone in this position — hundreds of agencies are working to figure out how to integrate AI into their existing workflows. Rather than seeing AI as a job replacement for talented experts and creative professionals, we are focused on how these tools help level up our process. However, there are certainly companies that do not share this view — that look at the opportunity of AI tools to increase throughput and use this as a measure to cut staff and automate key production positions.

We feel this is short-sighted, for a couple of reasons:

• On paper, increasing the speed of a team by a certain percentage may look like an invitation to trim capacity. This ignores the fact that everyone else is learning these tools at the same time! If you determine that AI can help you remove half of your marketing staff, but your competitor uses the same insights to perform twice as much marketing work with the same staff, you’re in for a rough surprise
• AI relies on digesting a tremendous amount of training data, and then regurgitating that data to fit your requests. So while it’s really good at doing what’s already been done, what it lacks is innovation. The machine can’t have an idea that’s not already in the training data. Many organizations have often struggled to innovate and move things forward under the load of keeping up with day-to-day requests. If we can automate more of that day-to-day, it leaves team members more room to experiment, try new things, and learn in ways that break patterns, where AI is instead following those patterns.

While we’ve always constantly improved our process as new technologies and techniques have come available, the pace of change within our industry has led us to double down on this practice in a number of ways:

• We’ve always kept an eye towards new solutions, but we’ve set up new procedures to evaluate and experiment with new AI tools as they come available (and re-evaluate regularly; tools are moving quickly enough that capabilities can change month to month!)
• We’ve added systems to our business processes that encourage our team members to gather and share new use cases and results as we try new methods and tools.
• Blend is working with our clients to identify use cases within their existing solutions that might be good candidates for re-evaluation in the light of the new capabilities that AI-based tools provide. There are many instances where processes that were manual, required human intervention, or required too much human time to implement, may now be possible.

Our job as consultants is to research, experiment, trailblaze, and then bring that leadership and expertise to our clients. While we strive to stay ahead as times and technologies change, we’ll always keep the integrity of our solutions and advocacy for our clients at the center of what we do.

Footnotes

• https://www.pingdom.com/blog/how-we-got-from-1-to-162-million-websites-on-the-internet/
• https://en.wikipedia.org/wiki/Second_Industrial_Revolution
• https://thehustle.co/clifford-stoll-why-the-internet-will-fail