Straight talk from a conversation between Sajeel Qureshi (CEO of Computan) and Joe "Triple J" Jerome (CRO, Computan) on the Unwatchable podcast.
Sajeel Qureshi traces the origin of "developers for marketers" back to an early Computan client, a marketing agency that had inherited a custom WordPress plugin project for a cemetery client. The deadline slipped, and instead of scrambling to cover it up, the agency owner turned to Sajeel directly.
Sajeel recalled the moment the relationship shifted from vendor to partner, saying
"a client was actually a partner of ours."That single project reframed how Computan approached agency work going forward. Instead of chasing one-off jobs, the company positioned itself as the technical bench an agency could quietly lean on.
Sajeel described the appeal of staying out of the spotlight this way:
"we kind of liked being the secret weapon."Since then, Computan has supported what Sajeel estimates at 300 to 400 agencies, ranging from solo operators to multinational firms.
Joe Jerome, who ran Brand Builder Solutions before joining Computan, pointed to a specific and common failure pattern: agencies that only have one type of developer on staff.
"Your $10,000 a month strategic retainer is now in jeopardy."
Joe's take is that this gap directly costs agencies revenue:
"people are missing out on the revenue all over the place."
Sajeel explained that most agencies do not come to Computan through strategic planning. They come during a crisis.
Sajeel described his own default posture in these conversations:
"I'm more like the firefighter, not like the strategic person." He connected this directly to how agencies operate day to day, saying they are effectively "living in an inbox" with new jobs constantly arriving and staff spread thin.
One of the more pointed parts of the conversation covered why deals stall even when an agency owner is enthusiastic. Sajeel and Joe agreed that owners and staff are simply optimizing for different things.
Sajeel used a hospital analogy to make the point about wasted capacity:
"That's money lost. That's opportunity left on the table."
The idea: every hour a senior person spends on work a junior person could handle is an hour not spent on higher value work. Sajeel said this mindset tends to show up only at the most senior levels of an agency.
Joe connected this to compensation structure directly:
Sajeel summed up the founder mindset versus the staff mindset this way:
"Founder to founder, right?" The relationship changes entirely depending on who is actually in the room.
This was the central question of the episode, and both hosts landed in the same place: no, but the job is changing shape.
Sajeel drew a comparison to word processing software:
"Did Microsoft Word or Apple Pages kill authors?"
His point was that tools change how work gets produced without eliminating the need for the person who understands how to produce it well. He added a line that captures the whole argument:
"Technical people will always be in need."
What changes, according to Sajeel, is the nature of the problems technical people solve, not whether they are needed at all.
Joe brought the conversation back to practical reality with a direct example. Clients still come to Computan to rebuild Shopify stores and set up integrations rather than doing it themselves with AI tools. His reasoning:
"somewhat like working with a person."
Joe also pointed out that shortcuts have always existed in development, from Stack Overflow to code generators to autocomplete, and none of them eliminated developers. AI is simply the newest version of that same pattern.
Midway through the episode, the conversation shifted from whether developers are needed to what they should actually be building. Sajeel introduced the idea of "knowledge architecture," the underlying context, data structure, and governance that AI tools need before they can be useful.
Sajeel explained the logic simply:
"You need to educate the brain before the brain can think critically."
Joe connected this to real work Computan has been doing internally over the past several months, including:
Joe argued that setting up this kind of knowledge architecture is a legitimate, billable, high-value service category in its own right, not a side effect of using AI tools.
Despite all the hype, Sajeel was blunt about how little AI adoption has actually happened at the operational level inside agencies.
"a big red flag."
Joe backed this up with a concrete data point about HubSpot's new AI agent marketplace. Despite HubSpot having roughly a million users and around 7,000 agency partners, most listed agents had fewer than 10 installs. Joe's read on it:
"the adaptation is not there yet."
Both hosts compared this moment to the early 2010s, when inbound marketing and SaaS were new and traditional agencies struggled to adapt. Sajeel put it simply:
"people just don't know what they're going to do next."
When Joe asked directly what advice Sajeel would give agency owners today, the answer was short and practical:
Sajeel's closing line on the topic captured the overall stance of the episode:
"instead of doubling down with bad data, I'd rather change my mind."
No. Both Sajeel Qureshi and Joe Jerome concluded that the underlying need for technical talent has not gone away. What has changed is the type of work developers do and the tools they use to do it, similar to how word processing software changed writing without eliminating the need for writers.
A single developer usually specializes in one platform, such as HubSpot CMS. When a client needs something outside that specialty, like WordPress or Shopify, the agency either has to turn the work away or bring in a second vendor, both of which put the client relationship and the retainer at risk.
Knowledge architecture refers to the context, data sources, rules, and governance an organization sets up so AI tools can work reliably. Without it, AI lacks the information it needs to produce accurate, useful output, which is why Sajeel Qureshi describes it as educating the brain before it can think critically.
According to Sajeel Qureshi, most agencies are still using AI only for surface level, client facing content creation. Very few have deployed it for internal operations like QC, task management, or time tracking, largely because the tools and workflows are still new and unfamiliar.
Agency work tends to be reactive by nature. Most outside technical engagements start with a specific, urgent problem, such as a missed deadline or a staffing gap, rather than as part of proactive strategic planning.
Yes. Joe Jerome noted that AI still makes mistakes and hallucinates, which means it needs to be supervised similarly to a person. Structured code and integrations, by contrast, behave predictably and do not carry that same risk.