Connective Blog

This blog is a summary created by ChatGPT from over 150 slack messages of our Connective Developer conversations regarding AI. It is a brief picture of how we are currently viewing AI in the ever evolving world of Software Engineering. Here are key themes and takeaways from the discussion

1. AI Is Changing the Role of Developers, Not Replacing Them

The strongest theme across the conversation was that AI is a tool that augments developers rather than replacing core developer skills.

Key points:

• Developers still need to design systems, validate outputs, and guide AI.

• AI can accelerate development but cannot independently build reliable software systems.

• The real shift is from writing every line of code to orchestrating, validating, and guiding AI generated work.

One participant summarized this well:

The writing of code becomes less important while defining, experimenting, proving, and assembling solutions becomes more important.

2. The Mental Model of Development Is Changing

Several developers noted that the biggest adjustment is changing how they think about the job itself.

The emerging model looks like:

Old model Write most code manually.

New model Define problems, guide AI tools, review outputs, and assemble systems.

This shift requires developers to focus more on:

• architecture thinking

• experimentation

• validating outputs

• orchestrating AI agents

3. Fundamentals Are More Important Than Ever

Despite the rise of AI tools, participants strongly emphasized that core software engineering fundamentals remain essential.

Important fundamentals mentioned include:

• programming concepts

• architecture and system design

• understanding frameworks and patterns

• source control and engineering practices

• debugging and verification

The consensus was that AI makes weak thinking more dangerous, because mistakes can scale faster.

One comment captured this clearly:

Poor critical thinking skills will get you in trouble faster with AI than they did without it.

4. AI Is Best Used as a Learning and Acceleration Tool

Developers described using AI in several productive ways:

Examples included:

• learning new frameworks

• researching unfamiliar topics

• generating prototypes

• validating architectural ideas

• building simulation scripts

• accelerating side projects

Several developers described using AI as a tutor or collaborator rather than as an autonomous builder.

Example use cases discussed:

• building simulation models

• generating knowledge bases

• exploring complex topics

• scaffolding applications quickly

5. Side Projects Are a Major Learning Mechanism

Many developers said the best way to build AI skills is experimentation through personal projects.

Common approaches mentioned:

• building new applications

• testing AI workflows

• experimenting with agent frameworks

• letting AI assist in unfamiliar areas

These projects help developers learn how to:

• prompt effectively

• validate results

• structure AI workflows

• identify AI weaknesses

6. AI Agents and Multi-Agent Workflows Are Emerging

Some participants described experimenting with agent based development models.

Examples included:

• separate agents for frontend, backend, integration, testing, and documentation

• agents collaborating in parallel

• rules and guardrails for consistent outputs

This approach was described as:

Having multiple junior developers working simultaneously.

However, this approach still requires human oversight and architecture direction.

7. The Biggest Open Question: How Do Junior Developers Learn?

A major concern raised in the conversation was how new developers will build foundational skills in an AI driven development world.

Challenges include:

• fewer junior hiring opportunities

• AI accelerating senior productivity

• unclear training pathways

• education systems lagging behind

Participants suggested that organizations must rethink:

• how junior developers are trained

• how mentoring works

• what junior roles look like in the future

This was described as:

The $1M question.

8. Hiring Criteria May Shift Toward Human Skills

Several participants argued that soft skills and thinking skills may become more important than specific technical skills.

Important capabilities mentioned:

• critical thinking

• logical reasoning

• communication

• bias awareness

• curiosity and learning ability

• willingness to admit uncertainty

The reasoning is that AI can generate code but cannot reliably define intent or validate outcomes.

9. AI Does Not Remove the Need to Review Code

There was discussion about whether developers should still review code generated by AI.

The consensus was:

• yes, code review still matters

• trust levels vary depending on context

• developers often aim for AI to get them 80 to 95 percent of the way

Humans still provide the final validation.

10. Development Workflows Are Rapidly Evolving

Participants noted that their daily development workflows are changing significantly.

Examples include:

• using AI planning modes

• creating rule files for AI tools

• using AI to generate architectural plans

• using AI to scaffold projects

However, everyone agreed that the best workflows are still emerging and evolving quickly.

Overall Takeaway

The discussion highlighted a major transition in software development.

AI is not eliminating developers but reshaping the role from code producer to problem solver and system orchestrator.

Developers who succeed in this new environment will combine:

• strong technical fundamentals

• critical thinking

• experimentation with AI tools

• ability to guide and validate automated systems.