Predictions For The Coming Software Storm
Software development is on the verge of a massive shake-up, and it’s all because of AI
An increasingly large number of people are starting to realise that this won’t be the smooth, gradual transition many envision (or hope for). Instead, I predict a chaotic scramble as companies grapple with integrating AI into their workflows, a scramble that may leave many casualties in its wake.
Of course the topic of AI and its likely impact on jobs has been covered extensively. I shall try not to bore you, dear reader, with a repeat of old arguments, but instead offer a (hopefully!) new insight and offer some predictions that flow from it.
It’s my belief that the primary danger of this particular inflection point is the combinations of (a) deeply opposing views of what AI is for, and (b) asymmetric information about how to get to an AI future. We is curious, is that it is not as if perceptions of (a) and (b) are randomly distributed across the population; no — they are highly predictable based on job function which, in turn, sets the workforce up for a tumultuous time in the next 3–5 years.
Two Tribes And A Referee
The two tribes poised for conflict are, broadly speaking, the “engineers” and the “business” professionals.
The “business” tribe includes most non-technical roles, such as sales, marketing, product managers, analysts, and the majority of the C-suite, with the possible exception of CTOs who come from the “engineer” tribe. Their typical perspective on AI is that it serves as a short-term competitive advantage and a necessary strategy in the medium term. They believe that companies that do not adopt AI will be eliminated in the future. Therefore, they argue for the need to quickly set aside any fears about AI, embrace it, and strive for a first-mover advantage.
They also think that any development work traditionally outsourced to offshore locations should be straightforward to replace. Additionally, they believe AI can enhance lead generation through automated content creation and sales automation, such as cold emailing. Consequently, they see AI as a tool that can both reduce costs and increase revenue.
Some may even think, “I’ve seen a YouTube video where someone built an AI agent in five minutes without writing any code!” While this is a broad generalization, it reflects a belief among non-technical people that developers may no longer be necessary, and that they can accomplish tasks more quickly on their own.
On the other hand, the “engineers” are individuals with technical skills who have coded at some point in their careers. Their perspective on AI is likely to be as follows:
AI tools, such as GitHub Copilot, improve my efficiency; however, the complexity of large-scale applications still requires experienced developers. I acknowledge that the firm will need fewer AI-augmented developers than in the past, meaning job cuts are inevitable.
While AI can generate code and test cases, explaining an entire workflow in enough detail for AI to write a complete application remains quite challenging. Furthermore, a potentially substantial re-engineering project may be necessary to modernize or adjust legacy applications, allowing AI to provide even greater automation.
How will this play out?
The end state for any software company will — I predict — be the result of the interaction between these three groups, and how powerful they are.
Firm #1: Large Companies
Large, bureaucratic organizations may have many engineers, but they are also burdened by powerful, deeply entrenched stakeholders to whom management has historically deferred. In these companies, I anticipate the following outcomes:
Top-down management may initiate a frenzy of “strategic initiatives” to integrate AI into various areas of the organization, often supported by management consultancies eager to create elaborate presentations. However, these initiatives are likely to lack concrete details.
As a result, IT security and compliance teams will respond by implementing so many safeguards, frameworks, and approval processes that almost nothing can be effectively delivered.
Overburdened developers, who are already stretched thin maintaining legacy non-AI platforms, will not possess the bandwidth to experiment with new AI tools or navigate the new internal regulations. Consequently, these initiatives will falter. Interestingly, these same developers may work on AI projects in their spare time, partly as a hedge against job loss, but this does not directly benefit their current employer.
Frustrated by the lack of tangible progress, management may revert to traditional methods to spur innovation. This could involve creating segregated innovation teams that operate independently from the business and seeking acquisition targets, either directly or through sponsored incubators.
Meanwhile, smaller, more innovative competitors will rapidly erode market share, increasing pressure and frustration for the organization. This, in turn, may lead to aggressive headcount reductions as a strategy to force faster AI adoption through resource constraints.
OUTCOME: confusion, frustration, and overly restrictive policies hinder growth, ultimately leading to aggressive business decisions regarding headcount and (potentially) significant layoffs.
Firm #2: Firms With Non-Technical Founders
Small to medium-sized software companies with non-technical founders tend to focus more on the business side of the equation and view AI as a significant opportunity to expand their market share by quickly “wiring things together” using low- or no-code solutions rather than hiring developers.
The speed and effectiveness of this approach to deliver new product or replacing existing ones will differ from company to company, however. And the fact that smaller companies typically face lower cybersecurity and compliance bureaucracy, there is certainly the potential for them to move fast.
In some cases, the legacy product may not be compatible with low-code, however, which will lead to tension between technical and non-technical teams. Developers who proactively engage with non-technical “business” staff will be seen as allies in this new strategy, and are likely to improve their standing. As the saying goes, however, “if you are not on the bus, then you are under it.”
While adopting a low- or no-code approach may lead to explosive short-term growth — especially for one-off products — achieving long-term, recurring growth will be more challenging. The more a product can be assembled using no- and low-code tools, the easier it becomes for competitors or customers to replicate it directly.
The perceived decline in the value of engineering expertise may prompt more experienced developers to look for jobs elsewhere, which poses a risk in the long term if the company reaches the limits of what can be accomplished without technical staff.
OUTCOME: enthusiastic adoption of low-code solutions risks upsetting experienced developers who may view it as “dumbing down”. While this may work for short-term new product delivery, maintaining software over time will be much harder.
Firm #3: Firms With Technical Founders
A small or medium-sized software company with technical founders is likely to lean more towards the “engineering” side of operations.
Similar to companies led by non-technical founders, the expectations from “referees” will be lower compared to those of larger corporations. Since the business leaders are engineers, their requirements are likely to be both relevant and appropriate.
The understanding that engineers have for their founder’s technical background suggests that this type of firm is well-positioned to make meaningful progress towards greater AI adoption as a unified team. AI integration is likely to be more thoughtful, as this type of company seeks a more considered blend of human developers actively assisted by AI.
While there may still be job losses or a hiring freeze as AI takes on the responsibilities previously held by employees who leave, I believe this type of firm could emerge as a significant winner in the evolving landscape. They are small enough to innovate quickly, and their management possesses the empathy and pragmatism, along with sufficient technical expertise, to avoid the pitfalls that may affect others.
OUTCOME: a considered blend of human developers with AI make this group the best-placed to benefit from AI. Small and nimble enough to innovate quickly, but with a technical pedigree that understands humans are needed long-term.
Firm #4: New Start-Ups
The final type of firm to consider here are new start-ups that are founded in the age of AI. Free from the constraints and headaches of maintaining a legacy codebase, how might these green-field companies work?
Well, I suspect their approach will reflect the technical versus non-technical divide highlighted earlier for existing firms. Broadly, that suggests that:
- Non-technical founders will try to “wire together” minimum viable products as quickly as possible and try to launch products which are really a thin wrapper on top of ChatGPT (or equivalent). Hiring human developers will happen at a later point as a result, storing up a potentially painful re-platforming job just as they start to gain traction.
- Technical founders, on the other hand, face the classic trap of building technically advanced products that don’t directly solve a business need. This has always been the case of course. The difference with AI, however, is that the set of possible use-cases for AI is near unlimited (unlike other fashionable technologies such as blockchain for example), and therefore this temptation will only grow.
Starting a business has never been easy, and I predict that the challenges faced by start-up founders will only increase as we move forward, as AI makes it easier for new business owners to make bad decisions.
OUTCOME: starting-up will become faster and more accessible than ever. The applicability of AI is so broad and the barriers to entry appear so low, however, that it may encourage entrepreneurs to make decisions that they later regret. Starting up a successful, long-term business has never been easy, and AI may make it harder still.
So those are my predictions about how AI may shape the software companies in the next 3–5 years. Buckle up — it will be a bumpy ride!
What do you think? Leave me a comment below.