According to Derek Xiao¹, Software 3.0 represents a fundamental shift in where value is created within enterprise software applications. The concept is part of a three-era evolution of software:

Software 1.0 (System of Record)
Value captured through controlling vital business data (Oracle, Salesforce, Workday)

Software 2.0 (System of Engagement)
Value shifted to user interfaces and engagement layers (Slack, Zoom, Notion)

Software 3.0 (System of Work)
Value now resides in the invisible ​'work layer​' containing business logic​ (Cursor, Abacus, Replit)

In the third era illustrated above, Cursor is a 'Software Developer Copilot' whereas in my example of Vibe Coding below, I am working with Replit as a 'Citizen Developer App Builder'. I am also talking about how Software 3.0 can co-exist as an extension to Software 1.0 (and Software 2.0).

Vibe Coding, introduced.

Vibe Coding is a term first used by industry analyst Andrej Karparthy² to describe how you can generate software code by using AI to describe apps using natural language prompts, where the term 'vibe' comes from "fully giving into the vibes, and forget that the code even exists".

I see Vibe Coding as a set of new but reliable technologies that both augments and extends the design and development of SaaS apps built on Salesforce Lightning Platform³. I describe my current state experience here as the AI Augmentation and AI Extension phases.

With both AI Augmentation and AI Extensions it means that there are always humans-in-the-loop in the build activities. However, what AI can deliver today are tools that allow non-programmers - Citizen Developers - to do much more than before in creating digital innovations.

Read my Journal post Simplify to learn more about how Experimentation with Storyboards and Models enable Citizen Developer innovation. It is this bottom-up approach to digital innovation, based on two things: Citizen Developers, Empowered; and, No-Code First.

Citizen Developers Empowered
Citizens are motivated to fix the problems that directly affect their lives. In the Social Care Cloud use case this is Discharge Nurses, Social Carers, Patients and Families - not executives - who have the greatest motivation to fix bed-blocking that also causes delays to patient operations.

No-Code First
Of course, if Citizen Developers are empowered to create SaaS apps that solve compelling problems, then executive managements and IT professionals can create a safe environment for Insourcing to thrive. This is No-Code First: scalable and secure.

Salesforce, augmented.

Vibe Coding, in the context of Salesforce, is where technologies such as Abacus.ai⁴ become the AI Agents to guide Citizen Developers to accomplish tasks that were previously only the domain of tech experts.

Abacus.ai DeepAgent provides:

Conversational Code Generation
Users describe what they want in plain English, and DeepAgent generates Salesforce code (Apex, LWC, Flows) or configuration steps.

Real-Time Collaboration
Multiple users can interact with DeepAgent, review code, and iterate on solutions together.

Context Awareness
DeepAgent understands Salesforce metadata (Objects, Fields, Relationships) and tailors its responses to your Salesforce Org Data Schema.

Design Thinking⁷ and Value Engineering⁸ underpins my work at Being Guided. Design Thinking maximises stakeholder engagement by generating empathy and trust. Value Engineering builds robust business cases using ROI Models to defend value and quantify costs.

Faster
No-Code development of Salesforce Software-as-a-Service (SaaS) apps can be reduced in time and costs as days and hours versus months and weeks, if Citizen Developers are close to and active in the problem being solved.

Simpler
At Being Guided I apply Fierce Reduction: Simplify all business processes, tasks and information systems by removing redundant or non-essential elements before considering next generation app development with Citizen Developers.

Cheaper
Vibe Coding represents significant cost savings when Citizen Developers take over 90% or more of the work traditionally delivered by IT services firms operating with a billable people hours business model. This is Salesforce plus Abacus.ai.

Replit creating a Citizen CRM app.
.

Salesforce, extended.

The combination of Replit and Salesforce enables the Vibe Coding of apps that can extend Salesforce as the System of Record to a new System of Work. Replit is an extremely well-funded AI startup with a valuation of $3billion from its last funding round of $250million⁵.

As illustrated above, I have created an app on Replit, with a specific app shown here - Citizen CRM. This is an example of Vibe Coding rapidly creating an extension to Salesforce as a trusted Software 1,0 era solution coexisting with a new Software 3.0 era extension.

The Use Cases for Citizen CRM are typically where consumers or say, patients need to become Users of a SaaS app. This is where the traditional Per User Per Annum pricing model of a first generation SaaS app, such as Salesforce becomes cost-prohibitive.

As widely forecast by industry analysts⁹, AI-powered technologies, such as Vibe Coded apps built on Replit, are costed on a Usage Pricing Model, as established for Google Cloud Platform and similar cloud computing Infrastructure-as-a-Service (IaaS) options.

So, I am selling Citizen CRM as a Prototype at this stage, because I am also introducing humans-in-the-loop to rigorously test and validate predictable (v. unpredictable) usage pricing, scalabil.ity and security. This applies to both Build and Use scenarios from 10 to 1million Users.

Citizen CRM was created to run on Google Cloud Platform, where production deployments can include country-specific data residency options - eg. 'Europe West-2 (London)'.

More broadly speaking, this full stack technology is built on industry-standard options, including:

• React with TypeScript for type-safe development.
• Next.js for server-side rendering capabilities.
• GraphQL for efficient data fetching.
• Tailwind CSS for styling.
• Parcel bundler for development and build processes.
• Go (Golang) leads performance-critical services.
• Python and Node.js support for additional backend functionality
• Express.js commonly used for API servers.
• Flask for Python-based backends.
• WebAssembly enables in-browser language execution.
• PostgreSQL 16 as the primary relational database (hosted on Neon).
• Redis for caching and real-time features.
• Built-in SQL Database with automatic environment variable management.
• Drizzle ORM for database interactions with built-in security.
• Point-in-time restore capabilities.
• Google Cloud Platform for hosting.
• Kubernetes and Docker for orchestration.
• Terraform for cloud resource management.
• Google Compute Engine and Google Cloud Storage.
• NGINX as Web server and reverse proxy.
• WebRTC for real-time collaboration features.

References

1. Xiao, D. (2025, September 11). AI's UI problem is actually a new era of software. Menlo Ventures.
   https://menlovc.com/perspective/ais-ui-problem-is-actually-a-new-era-of-software/
2. Karpathy, A. [@karpathy]. (2025, February 2). There’s a new kind of coding I call “vibe coding”, where you fully give in to vibes, embrace exponentials, and forget that code even exists. X. https://x.com/karpathy/status/175347216619708042
3. Stutz, A., & Kolb, R. (2021). Salesforce Lightning Platform Enterprise Architecture (3rd ed.). Packt Publishing.
   https://github.com/PacktPublishing/Salesforce-Lightning-Platform-Enterprise-Architecture-Third-Edition](https://github.com/PacktPublishing/Salesforce-Lightning-Platform-Enterprise-Architecture-Third-Edition
4. Abacus.ai. (2025). Abacus AI platform overview.
   https://abacus.ai/platform
5. Bansal, R. (2025, March 13). How to build a CRM with Replit AI Agent: A guide. Nylas.
   https://www.nylas.com/blog/how-to-build-a-crm-with-replit-ai-agent-a-step-by-step-guide/
6. Reuters. (2025, September 10). AI software developer Replit raises $250 million at $3 billion valuation. Reuters.
   https://www.reuters.com/business/ai-software-developer-replit-raises-250-million-3-billion-valuation-2025-09-10/
7. The Hasso Plattner Institute of Design. (2004) Stanford d.school. https://dschool.stanford.edu/about
8. Miles, L.D. (1947). The Lawrence D. Miles Value Engineering Reference Center Collection.
   https://minds.wiscon.edu/handle/1793/301
9. Metronome​. (2025, March 4). The future of SaaS pricing: How AI is pushing companies beyond seat-based models.​ Metronome Webinar.
   https://metronome.com/blog/the-future-of-saas-pricing-how-ai-is-pushing-companies-beyond-seat-based-models