Spent the last two days at a local startup expo. The most common thing I heard — by a wide margin, from visitors, potential customers, and just-curious enthusiasts — was a variation of this (also mainstreamed by popular media):
“Software engineers won't be needed (or useful) anymore. I can build whatever I want with ChatGPT, Claude, Lovable, EASILY.”
I had this conversation enough times that I want to put the answer somewhere I can point to, with some background info on what I've been up to with AI tools, the good and the bad, and what's realistic to expect now.
Fair disclosure: Looking everything from infra, architecture, and cost perspective will be biased from my side, just because I have experience from that angle does not mean I have to force it over others and for every use case either.
Few conversations highlight the repititive pattern: Paying $$ or $$$ for use cases that should honestly cost $. Not because anyone overcharged them. Because they don't have the infrastructure and ecosystem knowledge to know there were cheaper, better-fit alternatives. A database used as a queue. A Pro plan for what the Free tier already covered. An always-on EC2 instance for a four-times-a-day job. The application was working, so nobody questioned it. And when nobody asks “is this priced right for the shape of work we're actually doing?”, the default option — which is almost never the cheapest — quietly bills you every month. That's a deep-human-understanding problem too. It just shows up eventually in the bills or in the bugs.
The half-truth
The reason this question keeps coming up is that the part of software engineering most visible from the outside, and generalised by the daily news: typing code into a computer has genuinely gotten much faster. AI tools really do compress what used to take days/weeks into hours or less. Anyone who hasn't watched it happen wouldn't quite believe it.
So the question is completely fair and genuine. If the typing/coding is really automated, what's left to do now?
The things software engineering is actually about
1. Deciding and specifying what to build.
- Which problem matters
- Which version of the solution
- What “done” looks like
- What you're explicitly choosing not to do
2. Verifying and being accountable for what's delivered or finished.
- Does it actually work all the time
- Does it work at scale (doesn't always mean thousands or millions of users — does it practically work for 10 or 100 users, from their perspective)
- Does it work when something core/internal changes
- Does it work in the specific environment your customers/users are running
- And — when it doesn't — owning that
3. The deep human understanding
- of the codebase
- of the business
- of the environment those two have to fit into
This is the meta-skill that makes #1 and #2 possible at all.
The typing/coding part is absolutely necessary part of the process — but not the actual work or essence of software engineering.
What AI is actually doing now — honestly
The honest part most takes skip: AI is also helping with steps #1 and #2.
It helps me think through specs. It generates evaluation scripts. It writes verification flows. It catches edge cases I'd miss on first pass. The boundary isn't AI types, and humans think.
The IKEA analogy
For the simple explanation, I used IKEA analogy. IKEA has almost every possible furniture or utility item you could imagine — built, packaged, ready to assemble. The availability is real. But for most buyers (especially in India), when you ask will you buy everything from IKEA - the answer is usually no, not for all products. Availability alone isn't enough. They also want custom-fit, better quality, durability, cost over time, and maintenance - most of which IKEA does not provide.
That's why a lot of people still hire local agency/carpenter for at least part of the home. Not because IKEA can't supply it. Because the agency understands the specific business, the specific use, what the family will and won't tolerate over five years. The agency can also say “this design will warp in two years,” which the IKEA listing can't provide you.
AI coding tools are IKEA. They can build almost anything you can describe, and fast. That's real, and it's genuinely useful — same as IKEA is real and useful. But for the use cases that matter most to your business, you still want someone who understands the room, your business, and what will (and will not) work for you over time.
- So will IKEA grow further? - Yes absolutely, its a reputed global brand already and expanding rapidly
- Will it replace local agencies/carpenters? - Not in the near future, my friend running home/office interior business has practically unlimited work
- Will local agencies run out of business? - No, the demand is huge and growing, good agencies will continue to grow and provide value to customers
Consumers across the world, uses both of the services in balance or as per the need. Either IKEA (for certain type of products) and local agencies (for custom work).
What remains
Give me all the AI assistance in the world and the value I produce will still depend on how deeply I understand both the problems and the solutions the agents (or general AI) are building for them.
It's definitely not the practical answer to “why do I still need a person if I have these tools?” — because the tools can write the code, draft the spec, and run the eval. They can't be the one who knew that the customer's accounting team uses a particular Excel macro that breaks if dates come back as strings. They can't be the one who watched the previous version fail at scale and knows exactly which assumption broke. They can't be the one accountable when the deployment runs at 2 AM on launch night.
Those are all “deep human understanding” problems. They don't compress.
So, what next!!
The conversation will keep happening because typing is what's visible from the outside; understanding isn't. From the user (consumer) side, software engineering looks like the part the AI just automated. From the inside, the AI automated the smallest and easiest part.
I'll probably have this same conversation a hundred more times. The answer will be the same. The job that paid yesterday because someone could write code will pay tomorrow because someone can understand the problem deeply enough that the code — whoever or whatever writes it — actually solves it.
That's the part that's not for sale, not yet 😅.