
Democratising Artificial Intelligence
“From Craftsmanship to Prefabrication - Revolutionising the Builds”
Since December 2022, Large Language Model (LLM) QnA interfaces like ChatGPT and Co-pilot have democratised access to artificial intelligence, opening up endless possibilities for automation, decision-making, analysis, content generation, and the creation of innovative products. This shift has spurred an infinite range of applications in business, marking a significant moment in the democratisation of technology.
Challenging Traditional SaaS Models
“Missing the Mark - Cookie Cutter Smart Houses in Dover Heights”
In the first year of the LLM revolution, hundreds of thousands of LLM-driven tools were developed, but only a small number truly resonated with users. This lack of impact was primarily due to the developers and entrepreneurs behind these tools. They often relied on traditional business models and failed to fully anticipate the disruptive potential of widespread, rapid access to this new technology. Additionally, there was a rush to release products quickly to beat competitors, which led to many tools being simplistic "wrappers" around LLMs that offered only minor variations in how users could interact with QnA style chatbots.
Most creators of these tools tried to replicate the success of the SaaS model—a proven approach over the past two decades due to its scalability in addressing generic tasks. However, advancements in LLM technology and its ecosystem are encouraging a shift towards more customised solutions tailored to unique business processes and pain points.
While there will still be numerous applications for a "one size fits all" SaaS model built upon GPT’s, the trend is moving towards the development of bespoke, in-house applications. This shift is advantageous for in-house developers and consultants, and for anyone looking to integrate sophisticated AI-driven workflows to streamline their tasks. However, it may pose challenges for those aiming to become tech billionaires through broad SaaS offerings.
The early rush to capitalize on what was seen as an "innovation window" led to the premature release of many LLM products in an effort to secure market adoption before LLMs became mainstream in business processes. This urgency has only accelerated as the accessibility of the LLM ecosystem has continued to expand, particularly noticeable in 2024.
Leveraging LLMs
“Shifting from Standard Blueprints to Custom Architectural Designs”
The true potential of Large Language Models (LLMs) is realized when they access an organization's or user's unique data. While LLM-driven SaaS tools facilitate such integrations, they often lack the capability to program unique processes that dictate the LLM's responses. For instance, a real estate agent tracking lead generation may have a distinct process that differs from their competitors. Although SaaS products can enhance the retrieval of information and content generation, they typically function as support tools for completing specific task steps rather than as comprehensive agents capable of managing entire processes. To achieve this level of functionality, a SaaS product would need to incorporate workflow and automation components, or allow for the programming of custom code to meet specific user requirements. This complexity raises the question: should organizations consider developing their own bespoke products?
Bespoke AI Agents
“The Craft of Personalised Construction”
AI agents offer the potential to automate unique organizational or user process flows, utilizing LLMs to determine process flow logic. At each step, they can access domain-specific knowledge bases (such as document repositories) and databases, along with a wide range of actions and APIs to interact with features in third-party products. This capability extends the use of LLMs beyond mere content generation and information retrieval to include comprehensive process automation and task integration.
“The Cost of Couture Living”
Traditionally, the high costs of bespoke software development have created a significant barrier to entry for small-to-medium-sized businesses, leading to the popularity of one-size-fits-all SaaS products. Automating a single process can cost hundreds of thousands of dollars. Given the multitude of processes an organization may want to automate, these costs can quickly make in-house development prohibitive.
A Transformative Technology
“Robot Masons on the Horizon – Tackling the Builder Bottleneck in Booming Markets”
As LLMs open new avenues for businesses to envision transformative solutions, they are also revolutionizing the way we develop these solutions. Tools like GitHub's Code Pilot, which can automatically generate code from natural language commands, are transforming the coding landscape. This technology continues to evolve, enabling on-the-fly code development. With the advent of automated app creation, it's clear that these tools have significantly lowered the barriers to entry, cost, and time associated with building and implementing bespoke technological solutions.
Are Developers Reducing Their Rates?
Despite the future of AI increasing demand for developers, it's not immediately clear if developers are passing on cost savings to clients. What is evident, however, is that becoming a developer has never been more accessible, and this trend is likely to continue.
The Impact of ChatGPT on Learning to Code
The introduction of ChatGPT has allowed individuals with a basic grasp of logic to quickly learn how to assemble usable code and troubleshoot issues. Similarly, those familiar with any programming language can now easily learn new ones. This is akin to how automatic speech translators have enabled basic conversations with non-native speakers without prior language knowledge. Nonetheless, like these translators, coding tools have their limitations—it's never quite as effective as mastering the language itself. Despite these limitations, such tools have primarily benefited existing developers, though they have also been adopted by entrepreneurial technologists and those on the periphery of software development, such as product owners. In late 2023 and early 2024, we've seen significant advancements in the GPT ecosystem and software development sector, making software development increasingly accessible to a broader audience.
The Armchair Developer
“DIY Kits for digital construction”
In my opinion, the transformative signs of GPT-led workforce changes will only become clear when we enable individual employees to fully utilize the latest tools and technologies to automate their workflows. Imagine every employee being able to build automated workflows as easily as they use a spreadsheet—seamlessly integrating GPT calls to manage process steps, retrieve information, generate content, and activate various features from other systems and software. Major corporate players, such as Microsoft, have already introduced tools like MS Co-Pilot and their Power Platforms, which empower employees to harness this potential within the Microsoft ecosystem. Furthermore, the rapid advancement of low-code and no-code tools has significantly lowered the barriers to creating everything from personal automation to workplace apps, and from mobile applications to commercial software.
Choosing the Right Tools
“Give a builder a hammer and he’ll construct a house, give him a scalpel and he’ll struggle to perform surgery”
Selecting the right tools based on product requirements is essential before initiating builds or upskilling employees. This principle applies equally to designers and entrepreneurs, who have access to a broad array of tools varying by solution architecture, delivery modality, and specific use cases.
No/Low Code Development Tools | Process Supported | Product | Use Cases |
Bubble | UI/UX Design, Front End Dev, Back End Dev | All | Best for Web Apps |
Flutterflow | UI/UX Design, Front End Dev, Back End Dev | All | Best for Mobile Apps |
Figma | Design | All | Interface and wireframing design. |
BuildShip | Backend Development | All | Projects requiring low code back end development that’s hosted. |
Automation / Integration /Workflows | Process Supported |
Make | Automation workflows integrating web apps |
Zapier | Automation workflows integrating web apps |
Power Automate | Automate workflows between microsoft services and other applications |
Retool | Quickly build bespoke internal web apps |
Coda | Turn documents into apps |
VoiceFlow | Voice activated conversational AI workflows |
ManyChat | Create Ai Chatbots deployed via SMS, WhatsApp, websites. |
Recommendations
Integrating AI worlfows with Product Types- ManyChat
Low Code Tool for Building a Back End - Buildship
Front End Low Code Tool For Web Apps - Bubble.io
Front End Low Code Tool For Mobile Apps - Flutterflow
Corporate Desktop Automations - MS Co-pilot with power app automations
Mobile Phone Automations (prototyping mobile apps) - Siri Shortcuts
Voice Driven / Conversational Workflows - Voiceflow.ai
Beyond The Basic Builds
“The risks of handing out power tools in the office”
As we navigate the evolving landscape of AI and automation, it's important to understand that while individual employees can quickly develop minimal viable products (MVPs) and automate workflows, these efforts represent only the initial steps in harnessing the full potential of technologies like Large Language Models (LLMs). User created GPT-driven workflows may more quickly address immediate needs, but its important to ensure they’re built within the appropriate business and ICT governance context. Even still such applications utilised within silo’d knowledge domains only scratch the surface of what these new technology stacks can accomplish.
“Building a simple shed vs building a modern skyscraper”
Integrating LLMs into more complex and impactful business solutions requires careful and strategic planning, similar to the comprehensive approach necessary for large-scale project developments. MetaMorph-IT provides expertise and support across various critical areas to ensure the successful implementation of advanced LLM applications:
Business Case and ROI Analysis: Evaluating the financial implications and benefits.
Data Readiness and Strategy: Ensuring data is prepared and strategies are in place to leverage it effectively.
Technology Integration: Seamlessly incorporating LLMs with existing systems.
Model Selection and Customization: Choosing and tailoring models to fit specific business needs.
Regulatory Compliance and Ethics: Adhering to legal standards and ethical considerations.
Change Management: Managing the transition and adoption of new technologies.
User Experience and Interface Design: Designing interfaces that are user-friendly and enhance interaction.
Scalability and Maintenance: Ensuring solutions can grow with the business and are maintainable over time.
Security Considerations: Safeguarding all systems against potential threats.
Performance Monitoring and Optimization: Continuously improving system performance.
Cost Management: Effectively controlling expenses related to LLM deployment
The Crucial Role of Business Analysis
“Building Without an Architect : Foundations on Sand”
Business analysis is essential in the development of GPT-driven products or the integration of Large Language Models (LLMs) into business solutions. Often underrated, the role of a business analyst (BA) is critical, especially as domain experts, entrepreneurs, and SMEs may lack the comprehensive insight needed to transform an idea into a viable product.
Bridging Business and Technology:
BAs are pivotal in linking business stakeholders with technical teams, translating complex needs into actionable technical specifications. This bridging is crucial because business owners, while clear about their goals, often do not possess the necessary technical skills or understanding of LLM capabilities and limitations.
Aligning Solutions with Business Objectives:
BAs ensure technological solutions align with strategic business goals, which enhances operational efficiency and addresses real needs. Without this alignment, even advanced solutions can fail to add value.
Eliciting User Requirements and Providing Insights:
Business analysts excel at gathering detailed user requirements, a step vital for developing functional and user-friendly features. They also provide objective insights into process inefficiencies, offering a fresh perspective that can lead to significant improvements.
Risk Mitigation and Project Management:
BAs conduct feasibility studies and impact analyses to identify risks and manage project scope and timelines, ensuring projects stay on track and within budget.
Challenging Assumptions:
BAs challenge overestimations and misguided assumptions, crafting realistic plans that ensure technical feasibility and strategic alignment of projects.
Underestimating the role of BAs can lead to projects that do not meet business needs, wasting resources and opportunities. BAs are indispensable in ensuring projects are technically sound and strategically aligned with business objectives.
Get in Contact with Metamorph-IT today to discuss ways to better integrate LLM’s into your business or product offerings.
Comments