The AI Reskilling Mindset: Journeys in low code tools, workflows and products

At the Oxford AI summit, we are announcing the Erdos Research Community - which is based on my teaching and various methodologies which I have used in reskilling for AI 

Over the years I have been teaching, both at Oxford and beyond,  I have spoken to many of my students and seen their AI reskilling journeys first hand. Today, in the age of low code / vibe code AI - where coding is being democratised - reskilling for AI is actually not about technology - but it’s rather about the reskilling mindset. 

From the many journeys I have personally seen so far, this democratising of AI skills beyond traditional coding -is a cause of optimism. 

This ability and mindset is like finding a hidden treasure / superpower  - which can transform us for the good. 

Ironically, I do not see our work as a start-up. Having named our venture after an eccentric Hungarian mathematician, I see our work as a community or a fellowship. 

Essentially, I see Erdos Research as a Reskilling AI mindset Fellowship

Based on my teaching of AI primarily at Oxford, I have seen the journeys of many people as they see their lives change through the transformative power of AI

Specifically

Claudia Saleh

Anne- Marie VERDIN-MULOT

Aditya Jaokar and David James

Ayşe Mutlu

Magnus Smarason

Anjali Jain

Dr. Amit Chougule DPM,MD,MRCPsych

Mayank Sharma

Aysha Nasim

David Catalano

and of course Jesus Serrano

I also think that mindset in other domains could apply here ex 

The book Boys in the boat suggested to me by Anne- Marie VERDIN-MULOT

Lewis Hamilton's biography

Tiny experiments by Anne-Laure Le Cunff, PhD

Hence, the mindset of AI reskilling we have learned from our teaching is the foundation of Erdos. Its shared in the form of interviews, books, best practise and our newsletter 

We (personally in our team i.e. me, Anjali Ayse and Mayank) use  Lovable , Cursor, windsurf, OpenAI , Langflow, Lyzr AI and Langgraph 

But, from the extended team we work with, there are many more excellent tools - and more will come. However, its clear that AI tools and platforms are making it easier to build and deploy AI itself. This includes all aspects from coding to design to deployment of software systems.  

Consider a platform like Lovable

If we abstract the characteristics of the mindset of successful AI reskillers using Lovable -  we see some interesting commonalities. 

First, let’s look at the individual level characteristics. 

1. The platform mindset - Successful reskillers view AI not as isolated models, but as interconnected workflows where data, reasoning queries, and dashboards form a cohesive system. They see Lovable not merely as a user interface, but as a framework for structuring logic, configuring prompts, and interfacing with databases and APIs in a way that mirrors system-level thinking.

2. This makes the low-code first. They are comfortable with low-code tools like YAML and SQL, they may not be traditional programmers, but they can effectively manipulate structured prompts and define schemas. They intuitively model knowledge, causal, and relational structures within Lovable, making them efficient creators of AI-enabled workflows without needing deep coding expertise.

3. Being pragmatic and outcome focussed, these individuals prioritize real-world impact over theoretical perfection, approaching AI with a problem-solving mindset. They rapidly prototype solutions using available components, learning by doing and refining iteratively. Their core question is, “What can I build with this now?”—favoring immediate utility over abstract mastery.

4. Prompt based approaches focus on intent - and this makes them comfortable with abstraction. They don't require access to raw model internals to create value; instead, they trust structured data and declarative tools to do the heavy lifting. This mindset allows them to focus on intent—expressed through reasoning prompts or schemas—while relying on Lovable to manage execution complexity.

5. The reusable mindset - With a focus on scalability, they design modular templates and components that can be adapted and reused. Whether crafting a YAML prompt or a dashboard view, they treat each element as part of a growing toolkit, ready to be repurposed across projects or shared with others in their community.

Now, let’s look at the collective characteristics

1. Pattern seeking behaviour: They identify reusable logic patterns across domains, turning one solution into a blueprint for many. Whether building dashboards for sales or compliance, they recognize the underlying query structures and maintain a mental playbook of use cases, reasoning patterns, and outcomes that guide future projects.

2. Inherently collaborative: These reskillers excel in cross-functional settings, asking insightful questions about data relationships and reasoning needs. They bring AI tooling into collaborative conversations with business owners, analysts, and engineers—bridging the gap between technical and non-technical teams through curiosity and clear communication. This could lead to more forward deployed roles

3. Reusable mindset: With a focus on scalability, they design modular templates and components that can be adapted and reused. Whether crafting a YAML prompt or a dashboard view, they treat each element as part of a growing toolkit, ready to be repurposed across projects or shared with others in their community.

4. From reskillers to co-creators: Their journey doesn’t stop at learning; they grow into creators of original workflows, prompts, and schema designs. Over time, they contribute back to the community, mentoring peers and setting standards—transforming from learners into leaders within their AI ecosystem.

We hope to show these ideas in real examples through our platform at the Erdos Research Labs.

I feel we are addressing  a large and complex problem - which will impact us all. Although, we have learnt a lot from our teaching, I think we are just at the outset of the impact of AI on our lives

 If you want to attend the Oxford AI summit, please see the link.

Many thanks to David Knott for his insights and feedback

If you want to know about Erdos Research please message me or comment here