Junior Talent in the Age of AI
The Hay Harvest by Pieter Bruegel the Elder
Many small agencies struggling with low profit margins enter 2026 in a familiar conundrum: they can't improve their project profits, because they can't charge more for their existing services; and they can't offer higher-priced services, because they can't afford to bring on the senior-level talent needed to deliver it.
It's an unenviable position to be in, and with AI increasingly making low-margin, production work more competitive, it's untenable. Agencies that found themselves increasingly pitching work against freelancers at the end of last year are in for more of the same this year, as advancements in technology lower the barrier to entry and unemployed creatives try their hand at freelancing. It's going to lead to increased pricing competition for certain types of work, which will further erode existing agency margins and unfortunately, put some of them out of business.
These agencies have at least two options:
- Deliver the same services to a less price-sensitive client (i.e., clients able to pay higher fees). Agencies that have under-invested in business development and marketing may have an opportunity here. It could be that they're simply under-charging for their existing services and could command higher fees by getting in front of new customer segments.
- Deliver different services that command higher fees and stronger profit margins. It may be time for the agency to evolve to meet new demand.
This article will consider the latter in light of the recent trend of avoiding entry-level or junior hires. I'll explore it from the perspective of software development, but parallels can be drawn across other disciplines like design and strategy.
"There has never been a worse time to be an entry-level software developer"
This is the worst job market for entry-level software developers in decades.
Among college graduates ages 22 to 27, computer science and computer engineering majors are facing some of the highest unemployment rates, 6.1 percent and 7.5 percent respectively, according to a report from the Federal Reserve Bank of New York. That is more than double the unemployment rate among recent biology and art history graduates, which is just 3 percent. ("Goodbye, $165,000 Tech Jobs. Student Coders Seek Work at Chipotle, The New York Times")
Part of this is due to the jump in graduates in these fields: in 2026, the number of undergraduates majoring in computer science and computer engineering was 170,000, which is 2x the number in 2014.
In addition to more graduates, there are fewer jobs:
Compared with five years ago, the number of active job postings for software developers has dropped 56 percent, according to data compiled by CompTIA. For inexperienced developers, the plunge is an even worse 67 percent. ("Should You Still Learn to Code in an A.I. World? The New York Times")
And anecdotally, AI is contributing to a slowdown in hiring for junior level developers. Kevin Roos writes for the New York Times:
In interview after interview, I'm hearing that firms are making rapid progress toward automating entry-level work, and that A.I. companies are racing to build "virtual workers" that can replace junior employees at a fraction of the cost. [...] One tech executive recently told me his company had stopped hiring anything below an L5 software engineer — a midlevel title typically given to programmers with three to seven years of experience — because lower-level tasks could now be done by A.I. coding tools. ("For Some Recent Graduates, the A.I. Job Apocalypse May Already Be Here," The New York Times)
And so it would seem that the tech industry is avoiding junior hires in favor of mid-level and senior-level talent. Digital agencies may be thinking along the same lines.
There has never been a better time to be a junior developer I learned to program in 2012 with the help of Chris Pine's aptly titled: Learn to Program. I had quit my job in digital marketing in the spring and just gotten home from a road trip around the U.S. in predictable, male, twenty-something form.
Reading books, reviewing documentation, going down rabbit holes on StackOverflow, and attending Boston Meetups allowed me to build a Ruby on Rails app and land a junior software engineer role in the spring of 2013. Getting that job was one of the few times I've been genuinely proud of myself, and it changed my life.
These days, I occasionally return to coding side projects. In my latest foray, I started using Claude, and without belaboring the point, AI tools make learning software development infinitely better. I learned more about bash and cron in 20 minutes than I could have learned in 4 hours reading docs. Here's how AI changed my experience of learning:
Claude had the ability to answer questions within the context of the problem I was trying to solve.
Claude could "meet the user where they are" and explain things in elementary or advanced terms.
Finally, Claude allowed me to probe deeper and ask really specific follow-up questions – activities that previously took time traversing links of discussions in StackOverflow and only tangentially addressed what I wanted to know.
Are the answers going to be 100% accurate all of the time? No. But they don't need to be, because I can probe Claude until I understand what it's doing and then correct it.
All that is to say, that there has never been a better time to be a learner. Software developers should be able to learn faster and understand more deeply than they ever have before.
Junior developers are more valuable than they were a year ago
If AI tools can help junior developers find information easier, go deeper on answers, and debug problems to unblock themselves faster, then junior developers are more efficient than they were a year ago.
Since we haven't seen a corresponding increase in junior developer salaries, the increased productivity makes junior developers more valuable than they were last year. And as the models improve and familiarity with the tools grows, they'll only continue to become more valuable.
So why are companies slowing hiring of junior developers or looking to eliminate the number of jobs available?
⁂
The thinking is that senior developers can run multiple AI agents and prompt them to do junior-level development work. For this to be efficient, a few things need to be true:
A large portion of developer time needs to be comprised of doing work that AI does well (e.g., writing code)
Senior developers need to find the work satisfying (or they'll quit)
Senior level developers need to be able to sustain an increased workload (or they'll burn out)
Let's review each of these assumptions from the context of a digital agency.
Developers need to be coding
For one senior developer to do the work of one or more junior developers, they need to be able to rely on AI to do most of the work. But in a 40-hour workweek, how much time is actually spent writing code versus the following?:
Attending company meetings to keep everyone aligned and coordinated
Scoping and prioritizing project team work
Joining client meetings to keep stakeholders confident and happy
Conducting architecture planning
Reviewing code
Updating tickets and other project management tasks
Mentoring other team members
Being a non-productive human who jokes around in Slack and peruses Hacker News
Senior developers do a lot more than write code, and many of the things that they do can't be done by AI.
So companies need to be realistic about the efficiencies that AI introduces. Developers aren't coding 90% of the time; in fact, they're probably not writing code 50% of the time. So eliminating junior roles that have real work that is not coding transfers that work to senior developers without creating an equal amount of additional capacity.
Developers need to find the work satisfying
In the new world where senior developers and their AI agents are doing all the work, the senior developers need to be satisfied with the work. Otherwise, companies risk losing their most talented team members.
Many of the best developers I worked with loved mentoring others. And in my experience working with senior and mid-level developers, they want to develop those skills and grow in that way in their own careers. Companies that eliminate junior roles risk losing developers who want to be future Engineering Managers or surrounded by a team with diverse skill levels.
If the work isn't fulfilling, expect senior developers (who stick around) to place a greater emphasis on compensation. So senior salaries will increase at a time when most companies don't actually know what the true cost of AI agents will be over the course of a year. And how likely will those seniors be to stay when a competitor posts a job description for their role with a higher salary?
Developers need to sustain an increased workload
How many AI agents can a senior developer run concurrently and how many code reviews can they conduct in a day? Before companies can retire junior roles for AI agents, they need to understand what a sustainable workload for senior developers looks like.
The other thing to consider here is: how many AI agents could a junior level team member support? The answer isn't necessarily 0 or only 1, so companies should observe actual differences in productivity between juniors and seniors in the age of AI.
Risks in skewing senior
Agencies that eliminate junior developer roles from their business model, especially if they've relied on them before, are introducing some risks.
Risk of overgeneralizing
Big Tech and large ad agencies are making headlines with layoffs and elimination of junior-level jobs.
Imitation is common in tech, given the need to keep pace in a fast-moving industry. But these enterprise tech companies are very different from small and mid-sized agencies. Agency leaders sometimes overgeneralize when seeing similarities between their business and others in creative tech. They borrow from Big Tech or a 1,000-person agency, thinking that the best and brightest minds are leading those companies. But those organizations are fundamentally different from those of an agency with fewer than 100 employees. Sundar Pichai would not make the same decisions leading your agency as he makes leading Google.
A 20-person agency shouldn't emulate organizations 10x their size.
Risk of senior developer concentration
By eliminating junior roles, agencies are increasing their dependence on senior developers and exacerbating all the hiring and retention problems they've had at the senior tier before.
They're also eliminating their in-house pipeline for future mid-level and senior developers. So hiring teams need to be especially sharp in evaluating and bringing on mid-level and senior talent from outside the company. Beyond how difficult that is, a revolving door of incoming employees introduces risk around maintaining the company's culture.
Those who wear Patagonia vests and believe that "culture eats strategy for lunch" should tread carefully (I'm an REI man, myself).
Risk to profit margins
If the industry generally moves towards more experienced hires, a few things will happen:
Salaries for senior level talent will increase due to the increased demand
Salaries for junior-level (and early mid-level) talent will decrease due to lack of demand (and increasing supply of junior-level talent entering the workforce)
So an agency that tries to eliminate junior roles is going to be paying more for senior talent or will need to hire senior talent of a lower caliber (since the strongest senior talent will join the best agencies or agencies offering the strongest compensation packages). The uncertainty around just how much more expensive senior talent will become is a risk. The average agency risks building a team of expensive, "brilliant jerks" who just missed the cut elsewhere.
Equally important, this agency may be missing out on junior or mid-level talent that is increasingly affordable and which will lower or maintain the existing Cost of Services. And as I already noted, this group is becoming more productive in the age of AI.
Investing in the junior and early mid-level tier is the clearest route to improved profit margins for agencies.
An alternative approach
In a discussion on Lenny's Podcast with Lenny Rachitsky in October, Dylan Field of Figma indicated that his company of over 1,600 employees views AI as an opportunity to grow the team rather than cut jobs.
This puts Figma in a position to increase productivity per employee and deliver more value to customers. While requiring more upfront investment, companies that take this approach put themselves at an advantage over those who are simply reducing costs by replacing humans with AI without delivering more value. Companies focused on realizing more value for customers feel they won't see diminishing returns from larger teams, and if they're right, they'll potentially create operational efficiencies that make it very challenging for competitors reducing productive capacity to catch up with.
To put this in terms of a general recommendation for the average mid-sized agency that is well-utilized and profitable:
Prepare to pay your senior talent more in the coming years to retain them.
Focus on retaining senior talent that:
Excel at the most difficult technical or creative challenges (i.e., Technical Architects, Principal Designers).
Are adopting AI in their work. A healthy skepticism is fine, but they need to be willing to try things out and adopt what works.
Enjoy and excel at mentorship.
Keep this tier of high-paid service workers (that's a dev joke) minimal, with just enough coverage to keep more junior team members unblocked while minimizing key person risk.
In terms of junior and early mid-level team members, hire and retain folks that:
Demonstrate a high propensity for learning. These are the team members who will not just use AI for output, but also for learning and thus quickly growing in their capabilities.
Exhibit strong interpersonal and collaboration skills. Not only will these traits be important as AI assumes more of the coding work, but these skills will allow senior-level team members to delegate more and stay focused on high-value development work.
I've talked a lot about entry-level team members and the most advanced, senior-level individual contributors. The roles that haven't been covered much are the ones that I think are most threatened in the coming year.
What are they?
Typing pool at the Navy Department, Washington DC ca 1918-19 (Wikimedia)
Early senior-level team members are expensive, and in the small agency environment, if these team members are not making the most critical design or technical decisions on projects, they might not be the right fit. They are likely a better fit at larger agencies where they can lead less challenging creative or technical projects (e.g., a website redesign vs. a web application project). Unfortunately, the cost per hour of these team members is too high if they're picking up work that a mid-level team member could be productive on with guidance from a design or technical lead + assistance from AI.
Experienced mid-level team members who aren't likely to become creative or technical leads may also be a poor fit at agencies in the age of AI. These folks can cost 30% - 50% more than a junior developer, but that gap just isn't justified anymore given what a junior team member can accomplish with Claude Code at their side + a Technical Lead on standby. The need for "hands on keyboards" to complete production design or development work – the calling card for mid-level team members – is quickly dissipating.
Those who know me know that I hold worker-forward beliefs. So I call out early senior-level and experienced mid-level roles as being at risk not to promote the elimination of these roles, but as a rationalization of how digital agencies might adapt in the coming year. Hopefully my perspective lights a fire for folks who have deprioritized learning for a bit or procrastinated catching up on AI tooling.
Talent strategy moving forward
Each agency will need to create their own roster strategy based on a number of dimensions:
Their mission.
The nature of the services they provide (and the skill levels required for each type of service).
The size of their workforce
For example, a small agency delivering challenging services likely can't support any junior-level talent to begin with. In this case, agencies doing small, complex projects are going to face increased competition from senior freelancers who may see the greatest benefit from AI, if it allows them to take on new categories of projects.
On the other hand, a mid-sized agency delivering less demanding services may be in a position to transition to a more junior-level team supported by senior talent that oversee many projects. The teams could target projects that hold more scope than a small team of freelancers could take on, for example. And with a greater concentration of junior-level talent, they'll be able to deliver the work with lower costs and stronger profit margins than just a year ago.
