Bohemian Swarm AI Podcast - Episode 4 - AI Band-Aid

A conversation with Sachin Koshy, who leads AI strategy discussions for state and local government and education at World Wide Technology, on the Bohemian Swarm AI podcast with hosts Jess McDonald and Thomas Ott.

The following is an AI-generated summary of the interview transcript.

In This Episode

This episode's guest is Sachin Koshy, who leads AI strategy discussions for World Wide Technology (WWT), with a focus on state and local government and education — supporting research universities (R1, R2, and R3 institutions) and K–12 through K–20 education. His path there wasn't linear: he started as a combat engineer with the Navy Seabees, spent 25 years in sales (from door-to-door windows and siding to pharmaceuticals to university admissions), served as a director of admissions and program chair, sat on university advisory boards, and is currently finishing his PhD. He's a first-generation immigrant from India and holds certifications from Nvidia, Cisco, Dell, and Microsoft.

That background — engineer, educator, and career salesperson — shapes almost everything in this conversation, which ranges from the mechanics of university research funding to why AI adoption inside any organization is fundamentally a people problem dressed up as a technology problem.

What R1, R2, and R3 Actually Mean

The conversation opens with a primer on the Carnegie Classification system that ranks research universities. R1 status ("very high research activity") is tied to specific benchmarks — Sachin estimated roughly 180 institutions currently hold it, though he noted that number shifts as schools move between tiers. Achieving R1 status matters because it unlocks significantly more grant funding, which in turn attracts research talent and students. Jess, who came from an R1 university herself, added that the benchmarks include research contributions and program breadth, and that R1 status is increasingly tied to a university's ability to support AI infrastructure and data centers.

Does AI Make the Traditional Degree Obsolete?

Prompted by a question from his own son about whether to study data science, Sachin and the hosts dug into a genuine tension: college costs are as high as they've ever been, some major employers have de-emphasized degree requirements, and AI tools can now do work that used to require years of specialized training. Sachin was self-taught in data science and machine learning over a decade ago, before formal courses existed, and questioned whether that same path is still available today, given how quickly AI is absorbing technical work.

The group didn't land on a simple answer. Instead, they kept circling back to one theme: the "why." Sachin argued that too many students go to college because they're told to, without ever articulating why they're doing it or what they want out of it — and that's a bigger risk than the technology itself.

"Cognitive Atrophy" Is a Real Cost

Referencing a conversation with an educator who works on math curriculum, and a nod to a mutual friend's (Pat Bodin's) writing on the subject, Sachin raised the idea of cognitive atrophy: if people stop doing the underlying reps — coding, problem-solving, showing up in ambiguous situations — they lose the ability to go deep when it actually matters.

"You don't need to go deep anymore, but you need the ability to go deep — so you can orchestrate, so you can lead."

Jess connected this to the way learning itself is shifting: the traditional academic model built for the industrial era hasn't kept pace with how people now learn skills, including through AI-assisted apprenticeship-style learning.

The Real Budget Problem: Year One vs. Year Three

One of the episode's sharpest points was about how AI and data-center projects actually get funded at universities and in government. Sachin distinguished between capital funding (which buys GPUs and hardware) and operating funding (which pays the staff who run the systems and the power bill three years later) — and noted these come from different budget lines, approved by different people, on different timelines.

"We've watched enough institutions win the capital fight beautifully. They have this big, giant thing. And eighteen months later, nobody ever wrote down who owns the operating line."

His larger point: year one is easy — it's the ribbon-cutting. The real test of an AI initiative is whether anyone secured the money and staffing to keep it running in year three.

Leave Your Ego at the Door

Sachin was direct about what makes a technology advisor actually useful: not technical depth, but restraint. He described ego — often disguised as expertise — as the single most expensive thing an advisor can bring into a client discovery conversation, and argued that consultants who lead with "what's your use case?" or "what's your budget?" are already asking the wrong question.

"It is imperative that we don't underestimate the value of two ears and one mouth."

Automation (and AI) Can't Fix a Broken Process

Drawing on his Six Sigma and process-automation background, Sachin offered a line he says he coined and has since heard echoed elsewhere:

"Automation is great if the process isn't broken. AI is the exact same discussion — you cannot Band-Aid AI over a compound fracture process. All it does is break faster."

He pointed to COVID as a real-world example: organizations that scaled remote work fast discovered just how broken their underlying network and infrastructure processes already were. The lesson he applies to AI rollouts today is the same — if a workflow takes 36 steps that should really take 15, layering AI on top just automates the dysfunction.

The Flipper Zero as a Storytelling Tool

In a lighter detour, Sachin pulled out a Flipper Zero — a roughly $200 handheld device that can scan and mimic things like RFID access badges or garage door remotes — to make a point about cybersecurity awareness. He uses it in client conversations not to alarm people, but because showing a live demo of how easily physical security can be bypassed lands harder than describing it. He traced his interest in cyber defense back to seeing college students demonstrating similar tools at a campus cybersecurity club.

Who's Actually Driving the Change — and What Do We Owe Everyone Else?

The conversation turned to adoption curves: Sachin estimated roughly 30% of people are early, enthusiastic adopters of AI, another 30% will likely never come around, and the remaining third is persuadable with the right education and context. Getting that middle third on board, he said, is the real job — not converting the holdouts.

Jess pushed further, asking about the moral obligation of the smaller group driving radical, fast-moving change on behalf of everyone else who hasn't caught up yet. Sachin's answer connected back to his time as a Navy Seabee:

"We didn't build to the plan. We built to the ground. Somebody drew up that plan somewhere else, and the ground doesn't care."

He applied the same logic to university accreditation (evaluating what actually happens in a classroom, not just the course catalog) and to AI infrastructure projects (institutions that approved big compute clusters only to discover their buildings couldn't physically supply enough power — a problem that should be question one, not a footnote discovered six months in).

Closing Thoughts

Asked to leave listeners with one takeaway each:

• Sachin: Get better at building people around you — not just prompts and agents. Listen more, put yourself on mute a little more, and be willing to say "you're absolutely right." Try to be a slightly better version of yourself every day.
• Jess: Invest in each other. That's how you build the personal connection that gets people bought in.
• Tom: AI is radically reshaping work and the workforce, but hearing Sachin talk about creating new roles and merging people together — rather than just displacement — is a reason for optimism, not doom and gloom.