Nathan Hill on why you upskill the network engineer, not hire the dev
“You can't take 20 years of experience, grab a dev, and say, hey, now you know telco.”
The episode
Nathan Hill leads AWS's engagement across Australia's telcos. Before that he spent more than 20 years inside the telco industry, starting in engineering and operations, moving through pre-sales, and ending up running GTM for a challenger telco. His day job now is walking into Australia's largest enterprises and rolling out AI alongside the consultants, system integrators, and software partners.
James and Nathan get into why enterprises trip on the buy-versus-build mix, the proof-of-concept trap where nothing has a path to production, and the handover moment when the partner leaves and nobody in the building can fix the chatbot. Then a second act on go-to-market: why the car salesman never grabs a mechanic, and why engineers make the best sales hires.
What we get into
- Where enterprise AI goes wrong: buying a standard solution and expecting deep customisation, or building a complex platform to solve a problem you could buy off the shelf.
- The POC trap. Proving AI works means nothing if the build has no path to production and nobody can operationalise it. 'We've done AI' is not an outcome.
- How regulated telcos and banks adopt AI without waiting for perfect governance: automate the half of the process you can do safely now, and bank half the value.
- NASA's cool kids problem. Centralise AI capability to start, but the end state has to live in the lines of business, or you're buying consulting engagements forever.
- Why domain expertise is the moat. Ten years ago maybe 5 to 10 per cent of telco network engineers knew software. Nathan says within five years it will be 90 per cent, because you upskill the domain expert rather than teach a dev 20 years of telco.
- Go-to-market has changed. The car salesman doesn't grab a mechanic, salespeople now need to deeply understand AI, and Nathan openly hires engineers into sales.
Lines worth keeping
Tap to copy, then drop it into a post.
“If your chatbot starts spitting out Gordon Ramsay recipes instead of the answer to common questions on your website, who in your organisation can fix it? Do you pull it out of production? Can you adjust it on the fly?”
“If you looked at telco networks 10 years ago, you might say 5 to 10 per cent of those resources knew anything about software. What we'll see in the next five years is that 90 per cent of those resources will know a lot about software.”
“You don't want this cool kids aspect, where there's a centralised team of cool kids that get up in front of the town hall and talk about AI, but they're isolated from the business.”
“I used to be called a principal software engineer. I was paid $200,000 a year. I'm now managing a team of agents. My productivity is 10x. What should I be paid?”
Why it matters
Australian enterprises are buying AI faster than they can run it. This is the person selling it to them explaining how to avoid owning a system nobody in the building can fix.
Chapters
Tap any chapter to jump to that moment in the video.
- Cold open: what do you pay the manager of agents?
- Welcome: Nathan Hill, Head of Telco at AWS
- Buy versus build, and where enterprise AI goes wrong
- The POC trap: proving AI works with no path to production
- The handover moment: who runs it when the partner leaves
- NASA, centres of excellence, and the cool kids problem
- Upskill the network engineer, don't hire the dev
- Human in the loop on critical infrastructure
- The 10x engineer's pay packet
- The car salesman doesn't grab a mechanic: go-to-market has changed
- Building a go-to-market function from scratch
- The SaaS apocalypse gets an asterisk
Tech Teams Tuesday
Get the read that goes with the show
One issue every Tuesday, 7am. The moves that matter in Australian tech hiring, for the leaders making the hire and the tech professionals making the move.