Mike Fisher draws a useful line between coding and programming.
AI is writing more code, giving developers more time for the programming side of the job: understanding complex systems, making architectural decisions and figuring out how people will use the product.
For many developers, that's the part of the job they enjoy most.
Listen to the full Leadership Lessons episode with Mike: ow.ly/131A5107XfR
9
Your team's PR count has probably gone up this year. But has that translated into business impact?
Thamsanqa Moyo, CTO at Jem, says engineering judgement increasingly comes down to taste and outcomes. Did the work move gross profit, increase revenue or save costs? Did it reduce the risk your customer needed you to manage?
In the full session, Thams explains why Jem merged product and engineering into one discipline and gave every engineer end-to-end ownership.
Watch on demand: ow.ly/k1GW5107bO7
3
201
What happens when an electronics engineer gets a 3D modelling task in a technical interview?
At BMEC Technologies, candidates have always been given tasks outside their area of expertise. Now, co-founder Raphael Smith says they also want to see how candidates use AI to work through those unfamiliar challenges and explain the decisions they make along the way.
Thamsanqa Moyo, CTO at Jem, also shares what their in-person interview tests for.
Watch the full session on demand, whether youโre preparing for your next interview or rethinking how you assess candidates: ow.ly/vCqy5105h3q
1
170
A ticket sat in the backlog for six weeks. An AI agent turned it into a PR.
Every dev knows the tickets that never make it to the top of the list. In our Next-Gen Engineer session, Nicolaas van Noordwyk, Head of Engineering at OfferZen, showed how his team used an AI agent to pick one up and turn it into a PR, ready for review.
Watch the full demo on demand ๐ ow.ly/tJfU5104nkj
2
1
155
Congratulations on your promotion. Your new direct reports work around the clock, never ask for a raise, and will confidently build the wrong thing if you give them a vague ticket.
In our Leadership Lessons AMA, Mike Fisher makes the case that every engineer is becoming a manager of agents. Breaking work into chunks an agent can execute well is a management skill, and one that's becoming part of every engineer's job.
How are you breaking down work for your agents? Tell us in the comments.
Watch Mike's full session here ๐ow.ly/xE2x5103e22
123
At Jem, coding agents run for days without anyone needing to step in.
So how do you know which tasks to hand over to AI, and how much autonomy to give it?
CTO Thamsanqa Moyo shares how his team approaches AI delegation at different levels, using a framework developed by OfferZenโs product team.
Watch the full session to understand the framework, see where your team fits, and hear what Thamsanqa looks for when hiring engineers who build with AI.
Watch the full session ๐ ow.ly/QjiN5101e5Z
3
2
182
What does responsible AI adoption look like when you manage software for a fund manager?
We spoke to Sameer Adams, Head of Software & AI Engineering at Coronation Fund Managers, about how 13 years at the company have shaped his approach.
๐ ๏ธ ๐๐ณ ๐๐ผ๐ ๐ฏ๐๐ถ๐น๐ฑ ๐ถ๐, ๐๐ผ๐ ๐๐๐ฝ๐ฝ๐ผ๐ฟ๐ ๐ถ๐. From 10 people to around 80, the principle has held. AI makes it even more important. โAI built thisโ is never an excuse when something breaks.
๐ฆ ๐๐๐บ๐ฎ๐ป ๐๐ถ๐ด๐ป-๐ผ๐ณ๐ณ ๐ถ๐ ๐ฏ๐๐ถ๐น๐ ๐ถ๐ป๐๐ผ ๐๐ต๐ฒ ๐๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐. An architect agent drafts the build plan, a human reviews and signs it off, and only then does implementation begin.
๐ ๐ง๐ต๐ฒ ๐ฏ๐ผ๐๐๐น๐ฒ๐ป๐ฒ๐ฐ๐ธ ๐๐ถ๐๐ ๐๐ฝ๐๐๐ฟ๐ฒ๐ฎ๐บ ๐ผ๐ณ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด. After testing several team models, Sameer found that delays often come from stakeholders, design choices and analysis, before an engineer has touched an agent.
๐ ๐ค๐๐ฎ๐น๐ถ๐๐ ๐ถ๐ ๐๐ต๐ฒ ๐ผ๐ป๐ฒ ๐บ๐ฒ๐๐ฟ๐ถ๐ฐ ๐๐ผ๐ฟ๐๐ต ๐๐ฟ๐ฎ๐ฐ๐ธ๐ถ๐ป๐ด. Sameer avoids reporting productivity percentages. He measures whether software breaks less often and gets fixed faster when it does.
๐ง ๐๐ถ๐ฟ๐ถ๐ป๐ด ๐ฐ๐ผ๐บ๐ฒ๐ ๐ฑ๐ผ๐๐ป ๐๐ผ ๐ต๐ผ๐ ๐๐ผ๐บ๐ฒ๐ผ๐ป๐ฒ ๐๐ต๐ถ๐ป๐ธ๐. AI fluency is the starting point. What decides who gets hired is how they work through a problem and how quickly they learn.
๐ฆ Read the full conversation with Sameer about AI, ownership and hiring: ow.ly/YHUz50ZZc2Y
Engineers have always been climbing the stack. Machine code gave way to application languages, and every new layer meant trusting more of what sits under the hood.
Raphael Smith believes AI is the next layer, and itโs arriving faster than any before it. The skill worth building now is knowing which tasks you can trust AI with, then revisiting that call every time a new model lands.
He unpacked this, along with how to keep your judgement sharp without living in the weeds, in our Next-Gen Engineer online event.
Watch it on demand: ow.ly/RJ1y50ZYjFF
1
151
โBeing able to show how I use AI in my workflow helped me justify a salary increase.โ Thatโs what one developer shared in our 2026 survey.
Pay is still the top reason developers stick around, and the most common bump this year was 4% to 6%. Close behind is meaningful work, which developers describe as challenging work that keeps them learning. AI is reshaping that too, already giving half of developers more time to focus on higher-value tasks.
Get the full breakdown in the State of SAโs Developer Nation report ๐ ow.ly/UrK450ZXueS
121
Engineers love to engineer. Hand one an AI that writes code, and theyโll start building a system to manage it before lunch.
Stephan Swart, Fractional CTO, says thatโs exactly the shift. Youโre still building. Youโre just building the thing that builds the thing: code, acceptance tests, automated tests, the lot.
Thatโs how AI capability develops in a team. You start out assisted and checking everything. You delegate more as the system proves itself. Eventually, some work runs on its own.
Watch the full Leadership Lessons with Stephan here: ow.ly/Z8vX50ZXsbL
1
150
So far, 81% of people who responded say their AI skills help them stand out in the job market. ๐
The State of AI Delivery in South Africa survey is live, and early results already offer a glimpse into how AI is shaping the way we work.
Weโre proud to support this research and help build a clearer picture of AI adoption in South Africa. Have your say: ow.ly/wixS50ZX2ee
#AIinSA #StateOfAIDeliveryZA
At DevConf 2026, we asked the community one question: whatโs your AI engineering archetype?
Are you the:
๐ Agent-orchestrating Builder?
๐ฃ System-mapping Explorer?
๐ด Output-amplifying Multiplier?
๐ก Precision-first Artisan?
๐ต Skill-forging Apprentice?
Take our quick quiz - it only takes five minutes to discover your AI engineering archetype ๐ ow.ly/WQuj50ZW9nI
If youโre heading to any of these upcoming conferences, come say hi at the OfferZen stand. You can take the quiz in person, show off your results on LinkedIn, and maybe even walk away with a figurine of your archetype. Want a sneak peek? ๐ ow.ly/F8J750ZW9nJ
* DevFest CPT: 10 October
* PyconZA CPT: 15 October
* DevFest JHB: 11 November
See you there! ๐ฆ
In every conversation we've had with engineering leaders this year, AI delivery comes up.
Most SA teams are somewhere between experimenting and embedding it. Nobody in SA has solid data on how many initiatives actually make it from pilot to production.
The State of AI Delivery ZA 26 is where that picture starts. ๐
The survey closes Thursday, 16 October. Add your voice: ow.ly/nuVl50ZVJ3A
1
157
We caught up with Andrew Baker, Group CIO at Capitec, about what adopting AI really looks like inside a bank. Here are a few takeaways.
๐๏ธ AI adoption starts with the architecture. Andrew believes the key question is not what AI can do, but what happens when it gets something wrong.
๐ง Boundaries make experimentation possible. At Capitec, having the right guardrails lets teams try new things without risking production systems.
๐ง AI amplifies strengths and gaps. One developer used AI to rewrite Capitecโs internet banking platform in roughly three months. But getting it production-ready still took a team with complementary skills, because AI amplified their gaps as much as their strengths.
๐ฅ Two responsibilities for engineering leaders. Create an environment where people can experiment and learn safely, and bring different perspectives into the work early enough to catch what one person might miss.
Check out the full conversation with Andrew about AI, architecture, and engineering leadership here: shorturl.at/jIx1Q
1
176
Thinking about how your tech team will look in 2027? If youโre hiring or just trying to keep your best people, review season is a whole new ballgame now that AI is changing how we measure, reward, and evaluate engineering work.
Our 2026 Salary and Benefits report gives you the numbers you need to make those decisions: salary benchmarks, benefits gaps, hiring trends, and career progression insights, all from over 2,200 SA tech professionals. Download the full report here: ow.ly/J59550ZTNU7
230
We're proud to be backing the State of AI Delivery in South Africa. ๐ฟ๐ฆ
The country's first survey and report with DayOne and OfferZen, Pepperplane, Velocity Labs, and Altron AI Factory, building a real, evidence-based picture of how AI is actually being deployed across South African organisations.
If you build, deploy or lead AI work in SA, add your voice: ow.ly/1khK50ZTOsx
#AIinSA #StateOfAIDeliveryZA
216
Next stop: PyConZA! ๐๐
Weโll be there on 15 October, with Lizzy, Nicky, Estine, and Jay representing the OfferZen crew.
Hereโs what weโve got planned:
๐ค Discover your AI Engineering Archetype
Take our 5-minute quiz, find out which archetype you are, and share your result on social. Tag OfferZen for a chance to win your very own 3D archetype figurine. ๐ฆ
๐ Get your OfferZen swag
Come say hi and grab a t-shirt from our latest swag drop.
๐ฆ Take a pic with Dinocorn
Yep, Dinocorn is coming too. ๐๐ธ
๐ค Catch us for a 1:1
Visit our stand during the tea break for one-on-one sessions. If youโre looking for your next job and want guidance, or if youโre navigating the hiring market right now, come talk with us.
๐ PyConZA
๐
15 October
See you there! ๐๐ฆ
Nick van Noordwyk recently interviewed engineering candidates who all claimed to be AI-fluent. After a few follow-up questions, most were just getting started with delegating to agents.
For Nick, the gap between using AI and getting good work out of agents comes down to setup. His team went from pairing with AI to running agents in the background by restructuring their repos and building the environment agents need to produce work worth shipping.
This Thursday, Nick will show what theyโve set up and walk through how they ship a feature with agents, end to end.
๐
1 October | 12:00 SAST
๐๏ธ Save your seat: ow.ly/4yji50ZSvcX
4
234













