As a freelancer, pricing a project effectively while protecting yourself from potential risks is important. Based on my experience, I believe milestone-based payment allows you to manage the project effectively while protecting your finances and time.

Benefit of Choosing Milestone-based Payment

In Nigeria, there is a pattern where clients try to actually take advantage of freelancers because they perceive them as less experienced and younger. By breaking down projects into distinct milestones with their respective payments, you create natural checkpoints that protect both the client and the freelancer. If the client fails to pay after the first milestone delivery, you can pause the work without wasting your time. This minimizes potential losses (time and money). It also encourages professional conduct from both parties.

How it Works

1. Establish your desired monthly income. This becomes your baseline for calculations. Break this monthly income into a daily rate. Use the standard 20-day work month.
2. Assess your delivery capacity. How many endpoints or equivalent unit of works can you complete in a day? This helps you estimate realistic time frames for each milestone.
3. The formula = (Your daily rate) X (Number of Days to complete the largest milestone)

This calculation ensures that each significant project phase adequately compensates you for your time and expertise. Larger milestones naturally command higher payments, reflecting the increased work and complexity involved.

How to Handle Research Intensive Features

1. Be transparent with the clients about the research requirements.
2. Incorporate additional time buffers into your estimates. For example, 1.5x or 2x your normal estimation time frame. Factor this into your pricing calculations.
3. Build a proof of concept before you commit into full implementation. The reasons are because you validate technical feasibility, identify potential challenges, and give clients early visibility into progress.
4. Develop a systematic learning plan. Break down the required learning into specific areas. Identify quality learning resources. Allocate dedicated research time. Leverage AI.

Conclusion

In the freelancing space, it is important you protect your interests while delivering value to your clients.

Remember: Your expertise has value, and it’s essential to price it accordingly while implementing systems that protect that value throughout the project lifecycle.

As the world grappled with the isolating effects of COVID-19 lockdowns in the physical realm, another phenomenon was quietly incubating in the digital realm. OpenAI released GPT-3 in May 2020. Two years later, in November 2022, ChatGPT exploded onto the scene, went viral on Twitter with a speed rivaling that of the coronavirus. But why did it go viral? The chatbot’s ability to generate human-like text on a vast array of topics captured the public’s imagination, offering a glimpse into a future where AI could perform many tasks that humans already do, and possibly more. This was both exciting and worrying. After the novelty wore off, the debates began. What is the future of work? What about students’ homework? If AGI can do everything, how will people make money? While these concerns are valid, we must be honest: AI will take many jobs, if many have not already been disrupted.

According to an article published by the BBC, the increasing use of AI in the copywriting industry has led to job losses and devaluation of writers’ skills. The BBC explained how a company fired its team of copywriters, except for one, whose task was editing ChatGPT text to make it sound more human. Throughout January 2024, the Society of Authors conducted a survey of its 12,500 members, receiving nearly 800 responses on their experiences with generative artificial intelligence (AI) systems and their views and concerns about the future impact on creative careers. The survey revealed that a third of translators and a quarter of illustrators lost their job to AI. Additionally, a WIRED investigation found that major video game companies like Activision Blizzard, which recently laid off staff, are using generative AI for game development. What about freelancing marketplaces? An analysis of Upwork freelancing jobs from November 1, 2022, to February 14, 2024, revealed which jobs are being negatively impacted by AI. The study asked: Which job categories had the biggest decrease in volume? Which job categories were the least affected? Which job categories had the biggest decrease in hourly pay? And which AI skills had the biggest increase in job postings? The answers showed that video editing/production, graphic design, and web design jobs saw increases in volume and hourly pay. On the other hand, writing, translation, and customer service jobs had the largest declines, with 33%, 19%, and 16% decreases in volume, respectively. Even the head of Tata Consultancy Services said that AI could kill off most call centres. Like a storm sweeping away everything in its path, the AI purge has begun.

Before the rise of computer-generated imagery, animation was made using hand-drawn techniques like cel animation and stop-motion animation. The process was labour-intensive and required large teams of animators, inkers, painters, and other specialists. In 1995, Pixar released the first entirely CGI feature film, Toy Story, which changed everything. The industry blossomed. Before CGI, being an animator required knowing how to draw. After CGI, being an animator required knowing how to use software. Consistency and precision were handled by the computer. Now, with AI, the only skill needed is English language proficiency. Oh my, AI will enable the largest commoditization of skills and disruption of jobs. This is different! It’s no longer about new skills; it’s about a new mindset.

In the age of AI, there are two major categories of jobs: those competing against AI and those enhanced by AI. Choose a career where your craftsmanship, multiple experiences, and level of diligence give you an advantage in using ChatGPT. If you can’t influence the input, it means you can’t be unique. If the quality of the output doesn’t change significantly based on the quality of the input, it means you can’t be unique. Uniqueness is what makes you valuable. If you’re not unique, since most tasks will be done by ChatGPT, your boss will hire someone willing to be paid at a lower rate than you. If zero-shot prompting can give any employee, regardless of experience, a good enough response, what makes you unique? The output from your skills is competing against “good enough” from ChatGPT.

The End of Bullshit Jobs

In 2018, David Graeber, an anthropologist, popularized the term “bullshit jobs” in his book titled “Bullshit Jobs: A Theory.” He argued that in modern society, many people feel that their jobs are meaningless, unfulfilling, and do not contribute to the betterment of society. Graeber identified five types of bullshit jobs: Flunkies-jobs that exist only to make someone else look important; Goons-jobs that exist to bully or intimidate others; Duct Tapers-jobs that exist to solve problems that could be avoided with better design; Box Tickers-jobs that exist to check boxes but accomplish little else; and Taskmasters-jobs that exist to manage and control others, often in unnecessary ways.

Some examples of bullshit jobs, based on David Graeber’s categories, are: Flunkies-executive assistants who do little more than schedule meetings and make travel arrangements; personal shoppers for wealthy individuals; social media managers who only post content created by others. Goons-telemarketing script readers who harass people with unwanted sales calls; debt collection agents who intimidate people into paying debts they may not be able to afford. Duct Tapers-consultants who charge high fees for obvious advice. Box Tickers-middle managers who only exist to check boxes on a to-do list; compliance officers who only ensure that companies meet minimum regulatory requirements. Taskmasters-project managers who micromanage every aspect of a project; quality control specialists who only check for errors rather than improving processes.

Artificial Intelligence (AI) has the potential to significantly impact the workforce concerning bullshit jobs. AI can automate routine, repetitive, and administrative tasks, making many bullshit jobs redundant. For example, Flunkies-AI-powered virtual assistants can schedule meetings, make travel arrangements, and perform other tasks, reducing the need for human flunkies. Automated email and chatbots can handle customer inquiries, making social media managers who only post content obsolete. Goons-AI-driven telemarketing systems can automatically dial numbers and play pre-recorded messages, eliminating the need for human telemarketers. AI-powered debt collection systems can send automated reminders and negotiate payments, reducing the need for intimidating debt collectors. For Box Tickers and Taskmasters-AI-powered project management tools can automate task assignments, tracking, and reporting, reducing the need for middle managers who only check boxes and project managers who micromanage.

The Case for Skill Stacking in the Age of AI

There is a concept called the Dreyfus Model of Skill Acquisition. It describes the five stages of learning, which are: novice, advanced beginner, competent, proficient, and expert. A novice follows strict rules and guidelines, doesn’t understand the big picture, needs clear instructions, and applies them in a non-real-world context. The advanced beginner can recognize patterns and situational aspects, needs clear rules and instructions, and applies them in the real world. The competent can plan and execute tasks independently, understand the importance of goal-setting, and troubleshoot issues. The proficient can see the big picture, use intuition to assess situations, and draw on experience to fix issues. The expert has a deep understanding of the skill and uses intuition and experience to operate effortlessly.

Unless you want to invent a new concept in a narrow domain, expertise is probably overrated. Expertise is only required in rare cases. Most of the time, just being proficient, having a core problem-solving mindset, and Google-fu are more than enough.

Furthermore, how many industries allow the option to invent new things? If you are a social media marketer, you might not invent new tools. If you are a designer, you might not invent new tools. If you are a data analyst, you might not invent new tools. These tools are built by software engineers. Apart from being an engineer (software and hardware) or a PhD candidate and holder, most domains don’t afford you the opportunity to invent new concepts in a narrow domain.

In addition, it is hard to stand out in a mature domain. If you work for a small company, they might not value your expertise because they don’t have problems that require your expertise. If you work for a big company, you need permission. The easy solution would be to share your expertise on the Internet, but being noticed in the crowd is the real challenge. On the Internet, you don’t need permission, but noise can drown you out.

Apart from the mentioned issues, there is now AI. There is a popular line on Twitter and LinkedIn that ChatGPT is not going to take your job, but someone who uses ChatGPT will. There is some truth to it. But what about the second-order effect? If everyone believes this and uses ChatGPT, how will they stand out? In a particular domain, if everyone uses ChatGPT for almost similar tasks, the responses collectively become average. A recent study by researchers Anil Doshi and Oliver Hauser at University College London and the University of Exeter, respectively, published in Science Advances, found that while AI can enhance individual creativity, particularly for those who struggle with writing, its widespread adoption may lead to a loss of collective novelty and a homogenization of ideas. The study’s findings suggest that the increased use of generative AI in creative tasks may result in less unique and more similar outputs, perpetuating a cycle of blandness that undermines collective creativity. I am not saying you should stop using ChatGPT; I am only saying you should assess how you use ChatGPT.

You might ask, what should you do? This is where skill stacking comes in.

Skill Stacking is the ability to develop knowledge and know-how in multiple subjects instead of specializing in one area. Your audience follows you not because you know how to use the same tool they already use. They want to know how you use it with other tools.

Don’t use ChatGPT to answer questions that can be found on Google. Don’t outsource understanding and reasoning to ChatGPT. Use ChatGPT to build proficiency. You are not using ChatGPT to answer questions; you are using ChatGPT to build expertise. The proficiency from multiple subjects is where you find patterns and insights. These are patterns that even ChatGPT might not find. ChatGPT’s responses are built on data shared on the Internet. Insights are forged from real-world use cases. Do you think ChatGPT could have generated an idea like Uber or Airbnb in 2008? Obviously, no. The reason is because, before the sharing economy changed people’s behavior, parents all over the world advised their children not to talk to strangers. In a place like Nigeria, a sprinkle of superstitions and cooked-up stories would make it more dramatic. ChatGPT in 2008 would never have generated an idea to live with a stranger. With ChatGPT, you are either bringing proficiency or building proficiency. It is no longer about knowledge (ChatGPT has access to the Internet) but wisdom, which can only be forged in the real world.

ChatGPT is a tool. The quality of the output is based on the quality of the input. Garbage in, garbage out. Furthermore, the quality of the input is dependent on the craftsmanship of the user. ChatGPT is like a wand, and how you wield it is more important. Your wizardry with the wand makes you dangerous.

Hierarchy of Jobs and Their Vulnerability

1. Tier 1 — The requirement is to be human. Just being human is no longer an advantage. Some examples of jobs that fall into this category are models and many “bullshit” jobs. Startups can generate AI images much more beautiful than you. Fashion companies have started using AI to generate digital models. But will they be beautiful? Beauty is in the eye of the beholder. The evaluation of the output is decided by the client or manager. The client or the manager will probably be influenced more by the reduction of cost and time than by how beautiful you are. Ten times reduction of cost and time is a no-brainer. By the way, people underestimate how companies prioritize cost savings. The core of the Boeing issues is cost savings and profit prioritization. Interestingly, this situation reflects an inversion of the Moravec Paradox, which states that tasks easy for humans (like image recognition) are often difficult for AI, while tasks challenging for humans (like complex calculations) can be simple for AI. In the industry of modeling and image creation, you can see that AI can now excel in areas once thought to be uniquely human. This shift highlights how rapidly AI is evolving, challenging our assumptions about which tasks are inherently “human” and which can be automated.
2. Tier 2 — These jobs don’t require domain knowledge and experience. They require only reading, writing, and basic communication skills. Any educated person can do them. The basic skills learned on the job tend not to be transferable and offer no clear path to career advancement. Some examples are call center jobs, voice-over artists, and many “bullshit” jobs. AI startups can generate digital humans that can speak better than you. Can you speak 50 languages fluently? I thought so too. Synthetia AI can generate AI avatars that can speak over 130 languages fluently. Many of these jobs can be done by AI agents. Do you need to sleep at night? Your competitor doesn’t sleep at all.
3. Tier 3 — These jobs require reading, writing, and domain knowledge or skill, and less significant human judgement and less complex processes. When there is decision-making, the trade-offs are minimal.
4. Tier 4 — These jobs require reading, writing, domain knowledge or skill, and significant human judgement and significant complex processes. These decisions involve deep trade-offs that are context-dependent.

Job AI Resilience Framework

I created a framework to help people make career decisions in the face of AI disruption. The Job AI Resilience Framework is a structured methodology designed to evaluate the resilience of a job in a rapidly changing job market influenced by the rise of AI. It aims to equip individuals with the insights needed to adapt and thrive in an AI-driven landscape. There are three components that make up the framework: Domain Knowledge, Task Requirements, and Task Outputs. Each component has a set of questions.

1. Domain Knowledge: It refers to the depth and specificity of expertise required for a particular task. This component includes specialized knowledge, skills, experience, and the need for continuous learning.
2. Task Requirements: It involves the complexity and variety of task requirements. This component focuses on how tasks require human judgment, creativity, nuanced understanding, abstract thinking, contextual judgment, adaptability, and human interaction.
3. Task Outputs: It defines how the outputs generated from an LLM are used.

Job AI Resilience Index

The Job AI Resilience Index is a composite measure designed to assess and quantify the overall resilience of a job to AI disruption.

Job AI Resilience Index Score

The Job AI Resilience Index Score is a numerical value that represents how resilient a particular job or career is to AI disruption on a scale of 0 to 100.

How to Calculate the Job AI Resilience Score

1. Domain Knowledge
2. Task Requirements
3. Task Outputs
4. Calculation = ∑(Score for each question)/80 x 100
5. Interpretation

Domain Knowledge

1. On a scale of 1–5, how frequent does this career require continuous learning and skill updates? (Scale: 1–5; where 1 is “Rarely or never”, 2 is “Occasionally”, 3 is “Sometimes”, 4 is “Frequently”, and 5 is “Very frequently”)
2. To what extent does the career involve making decisions based on incomplete or ambiguous information? (Scale: 1–5, where 1 is “No decision based on incomplete or ambiguous information”, 2 is “Rarely”, 3 is “Sometimes”, 4 is “Often”, and 5 is “Frequently make decisions based on incomplete or ambiguous information”)
3. To what extent is the need for deep understanding of complex systems or processes? (Scale: 1–5, where 1 is “No deep understanding”, 2 is “Minimal”, 3 is “Moderate”, 4 is “Significant”, and 5 is “Deep understanding of complex systems or processes”)
4. On a scale of 1–5, how significant is the need for creativity or innovative thinking in this role? (Scale: 1–5, where 1 is “No creativity”, 2 is “Minimal”, 3 is “Moderate”, 4 is “Significant”, and 5 is “Highly creativity or innovative thinking”)
5. On a scale of 1–5, how would you rate the depth of domain-specific knowledge required for this career? (Scale:1–5, where 1 is “General knowledge”, 2 is “Basic specialized knowledge”, 3 is “Moderate”, 4 is “Advanced” and 5 is “Highly specialized knowledge” )

Task Requirements

1. As the state of progression in the AI industry, how many tasks can an LLM not perform conveniently in this career? (Scale: 1–5, where 1 is “one task”, “two task”, 3 is “three tasks”, 4 is “four tasks”, and 5 is “more than five tasks”)
2. How context-dependent are the tasks in this career? (Scale: 1–5, where 1 is “Not at all context-dependent”, 2 is “Minimal”, 3 is “Moderate”, 4 is “Significant”, and 5 is “Highly context-dependent”)
3. To what extent do the tasks require human interaction or emotional intelligence, compared to the current state of AI development? (Scale: 1–5, where 1 is “No human interaction required”, 2 is “Minimal”, 3 is “Moderate”, 4 is “Significant”, and 5 is “Extensive human interaction required”)
4. How much creative problem-solving is involved in the typical tasks of this career? (Scale: 1–5, where 1 is “No creative problem-solving”, 2 is “Minimal”, 3 is “Moderate”, 4 is “Significant”, and 5 is “Constant creative problem-solving”)
5. How frequently do tasks in this career require adapting to new or unexpected situations? (Scale: 1–5, where 1 is “Rarely or never”, 2 is “Occasionally”, 3 is “Sometimes”, 4 is “Frequently”, and 5 is “Very frequently”)
6. To what degree do the tasks involve making ethical decisions or judgments? (Scale: 1–5, where 1 is “No ethical decisions required”, 2 is “Rarely”, 3 is “Sometimes”, 4 is “Often”, and 5 is “Frequent ethical decisions required”)

Task Outputs

1. How critical is the accuracy and reliability of the task output? (Scale: 1–5, where 1 is “Errors are inconsequential”, 2 is “Errors have minor consequences”, 3 is “Errors have moderate consequences”, 4 is “Errors have significant consequences” and 5 is “Errors could be catastrophic”)
2. How do you use the output from an LLM? (Scale: 1–5, 1 is “No changes”, 2 is “Minor changes”, 3 is “Moderate revisions”, 4 is “Significant changes”, 5 is LLM cant solve the problem)
3. How frequently does the task output need to be updated or maintained? (Scale: 1–5, where 1 is “Rarely or never”, 2 is “Occasionally”, 3 is “Sometimes”, 4 is “Frequently”, and 5 is “Constant updates required”)
4. How complex is the process of maintaining or updating the task output? (Scale: 1–5, where 1 is “Simple, straightforward updates”, 2 is “Moderately simple updates”, 3 is “Moderately complex updates”, 4 is “Significantly complex updates” and 5 is “Complex, nuanced updates”)
5. To what degree does maintaining the output require deep understanding of its context and history? (Scale: 1–5, where 1 is “No context needed”, 2 is “Minimal”, 3 is “Moderate”, 4 is “Significant”, and 5 is “Extensive contextual understanding required”)

Calculation

1. Calculation = ∑(Score for each question)/80 x 100

Interpretation

1. High Resilience (75 -100): Jobs in this range are likely to be enhanced by AI rather than replaced. These roles typically require deep domain expertise, complex problem-solving, and produce outputs that need significant human oversight. AI tools are more likely to augment human capabilities in these roles, potentially increasing productivity and allowing professionals to focus on higher-value tasks.
2. Moderate Resilience (40–74): Jobs in this range may see significant changes due to AI integration. While not at high risk of complete automation, these roles may require adaptation and upskilling. Professionals in these fields should focus on developing skills that complement AI capabilities and be prepared for shifts in their job responsibilities.
3. Low Resilience (0–39): Jobs in this range are at higher risk of being significantly impacted or potentially replaced by AI technologies in the long term. Professionals in these fields should consider diversifying their skill sets, exploring ways to work alongside AI, or potentially transitioning to adjacent fields that have higher resilience.

Conclusion

In the rapidly advancing technological era, the Job AI Resilience Index offers crucial insights into career adaptability. This tool, along with its accompanying score and the Job AI Resilience Framework, helps people make informed decisions and develop necessary skills. Recognizing the importance of continuous learning and adaptability, you can turn AI-driven challenges into opportunities for growth, securing your place in tomorrow’s workforce.

Thank you for reading! Subscribe now to receive free updates on new posts and the exclusive launch of my upcoming AI product (a new way to consume knowledge)

• AI Is Eating Software: 4 Strategies to Build AI-Resistant Products in the Age of OpenAI and Big…
• How to Build a Software Product from Nigeria (Africa) and Sell to the World

Originally published at https://theriseofai.substack.com.

What is a General-Purpose Technology? General-purpose technologies are technologies that can be used across many industries and sectors, have the potential to increase efficiency and productivity in existing industries, and probably create new industries. Some examples of general-purpose technologies are the internet, electricity, computers, and the steam engine. The computer is a general-purpose technology. In 2011, Marc Andreessen claimed in an article that “Software Is Eating the World.” This is true. The biggest direct marketing platform is a software company — Google. The biggest bookseller is a software company-Amazon. The biggest recruiting platform is a software company-LinkedIn. You probably know the story of how Netflix, a software company, disrupted Blockbuster. We now have banks with no physical branches. So yes, software has eaten the world. AI is also a general purpose technology. It can be integrated into a wide range of industries and sectors. AI has or can be applied to industries like healthcare, finance, transportation, manufacturing, retail, education, etc. What Marc Andreessen probably didn’t envision was that, in 2023 and beyond, AI is going to eat software. AI will radically change the way we build enduring software products.

ChatGPT is a killer app. In one week after launch, it racked up about 1 million users. Two months later, it reached over 100 million monthly active users. This is impressive! This information is both a blessing and a curse. It is a blessing for OpenAI (millions of users and billions in revenue) and a curse for the tens or hundreds of startups ChatGPT made obsolete. There are three categories of startups or products that ChatGPT and Big Tech have made or will eventually make obsolete.

Three Categories of Products Vulnerable to Obsolescence

1. Single Use Case Feature — The obvious example is Grammarly. It existed long before the launch of ChatGPT. The company was founded long before OpenAI was founded. They found product-market fit. I use it. They have millions of users. Grammarly is everywhere. They have a web application, a mobile application(a keyboard), a Chrome extension, a Google Docs extension, and a desktop application. This is great, but there comes a time when disruption and obsolescence comes knocking. This happened to Blockbuster. This happened to Kodak. I think this might eventually happen to Grammarly. Let’s go back in time to the 1980s. Did you know that there were companies that sold standalone spell checker applications? One company was Random House which catered to the pre-Windows PC market. These applications ran alongside word processors. Eventually, the demand for standalone spell checkers declined when word processing software like Microsoft Word integrated these features. Before the rise of LLMs, the most recent techniques a company like Grammarly would have used include advanced natural language processing and machine learning techniques like contextual word embeddings (BERT), sequence-to-sequence models with attention mechanisms, and smaller-scale transformer models. They probably combined rule-based systems with machine learning, pre-trained models on large corpora, and employed advanced statistical parsing. Then human and user feedback improved the model performance. Now with the rise of LLMs, startups like Notion have integrated these features into their core product through an API from companies like OpenAI. Recently, Apple announced that they were going to launch brand-new systemwide Writing Tools built into iOS 18, iPadOS 18, and macOS Sequoia, users can rewrite, proofread, and summarize text nearly everywhere they write, including Mail, Notes, Pages, and third-party apps. Basically, a free Grammarly integrated into the operating system. Last year, Microsoft announced the integration of LLMs into all Microsoft Office applications like Word, Excel, PowerPoint etc. In March 2023, Google announced the integration of Generative AI into all their applications like Gmail, Google Docs, Google Maps etc. Since startups have integrated Grammarly into their core product and Big Tech has done the same thing, where is the need for a standalone Grammarly? I love Grammarly, but selling a feature is no longer an enduring strategy in the age of AI.
2. OpenAI Thin Wrappers — These are startups that built a product by leveraging OpenAI APIs. Startups like Jasper.ai and Copy.ai built copywriting and marketing tools, raised millions of dollars, hired many people, and spent a lot on marketing to acquire users. My sister used Jasper.ai, and she knew about Copy.ai. In November 2022, OpenAI launched ChatGPT. She found out about ChatGPT and stopped using Jasper. ChatGPT launched as a free product. How were these startups going to compete against free? I think the problem was that they didn’t know OpenAI was going to launch a consumer product. To be honest, there was no differentiation: text input and text output. The last I heard, Jasper laid offsome of their staff and Copy.ai pivoted (pivoting is not bad).
3. OpenAI Wannabes — They are basically OpenAI clones. They raised hundreds of millions of dollars. They have an impressive team of research scientists and engineers. They built their own model (no wrapper), probably not as good as GPT-4o, Gemini, and Claude. Their business model is to sell access to their model through an API and a chatbot. I don’t think the strategy is sustainable. A model is not a moat and there is no differentiation. It is expensive to build. In less than a year after building, it becomes obsolete. They need to always invest in new models, which become obsolete a year later. This is not the type of business that can be funded by venture capitalists because after raising a version one that nobody uses (I focus on the latest open-source and frontier models), they need to raise money for a version two that nobody will use. Another strategy would have been to be acqui-hired by Big Tech, but how many will be acquired? Inflection raised over a billion dollars and built a chatbot called Pi. The traffic was ridiculously low. How many chatbots does one need? Eventually, Microsoft hired the CEO, the team, and paid back money to the investors. It was not a normal acquisition. This happened to Adept too. They raised over 400 million dollars. They probably couldn’t raise enough to build new versions of their models and Amazon hired the team. The same acquisition trick. The reason is that the antitrust regulators in the US and Europe were going to block the acquisition because bIG tEcH is bLoCkInG iNnOvAtIoN. Let’s talk about clustering and winner-take-all market. Since you are reading this article, you know Uber-the ride-sharing app. I researched, gathered, and analyzed the data about the Uber clones. This is the summary: 47 clones, 27 existing, 13 collapsed, and 7 acquired. There is a saying that history doesn’t repeat itself, but it rhymes. I am here to break it to you that in the industry of building AI models history will not rhyme. Let me explain why. The reason there are 27 existing Uber clones is because of the dynamics of the ride-sharing business. A startup will raise money and focus on their local or regional market. The riders in Lagos only care about the drivers in Lagos. The riders in London only care about the drivers in London. The clustering is local. A startup in Lagos can give Uber a run for its money since they only care about Lagos drivers and riders. Compare these with chatbots and LLMs. Users from all over the world have access to ChatGPT. This is a winner-take-all market. OpenAI partnered with Bain and Company. Cohere partnered with McKinsey. Microsoft partnered with McKinsey. This is a great distribution strategy. The consulting firms deploy AI for their clients on behalf of the AI providers. If your startup just raised money to train your first or second version of your model, you are probably too late.

How to Build an Enduring Product

1. User Experience — First of all, the core of your product should not be a chatbot. ChatGPT has a performant free tier. Claude has a performant free tier. Gemini has a performant free tier. Microsoft Copilot has a performant free tier. You are competing against free. Whatever product idea comes to your mind, make sure the core of your product is not a chatbot. Did you not see the use cases of Artifacts, the feature launched by Anthropic? When I started seeing the different and impressive use cases, I thought to myself, “God, I have to think harder.” I don’t have a choice but to think harder, because if I don’t think, these big startups will make a mess of me. So far, it seems like OpenAI, Anthropic (Claude), and Google (Gemini) are investing in chatbot interfaces that can do everything. Based on that insight, think and rethink how your product can’t be replicated inside a chatbot interface.
2. Outcome-Driven Design — The output of an LLM is ephemeral. You give an LLM an input, and it returns an output. Yes, you can store the response (ChatGPT does so), but you most likely don’t need the information again when you are done with the task you needed it for. Build a product that focuses on the outcome, not just the output from an LLM. An example is OpenAI GPT-4o language demo or Google Translate versus Duolingo. I have a friend who can speak French (he is Nigerian, and our official language is English) and he uses ChatGPT as a conversation buddy. OpenAI is for people who know enough about a language and need a conversation buddy or a tool that helps with navigating a new country. Duolingo is for beginners who want to learn the language. Duolingo doesn’t just provide the curriculum (something ChatGPT can provide); it also tracks your progress.
3. Process Knowledge — If your product can be generated through zero-shot or few-shot prompting by ChatGPT users, I don’t think you have an enduring AI product. This was a problem with products like Jasper.ai and Copy.ai. The core of their products was the LLM. Build a product that requires multiple prompts (or fine-tuning), integrated by software engineering, and unique process know-how. An input field and output page do not constitute software engineering. This means AI is not the only component that makes up your product. Consider podscan.fm as an example. This service monitors podcasts for mentions of your product. The application consists of several components: one transcribes audio to text and uses natural language processing (including LLMs and named entity recognition) for content analysis. This particular component uses an LLM. Another component gathers podcast episodes from various sources, while a third sends customized alerts and notifications to users. The latter two components require robust software engineering.
4. Unique Data — The innovation with LLMs is that the marginal cost of content creation is zero, making personalization cheaper. The limitation of LLMs is that they depend on data. Your moat is unique data that frontier models don’t have access to. If you can curate complex data, fine-tune an off-the-shelf open-source model, and then build a product around the curated data, you will have built a unique product.

Conclusion

This article provides a framework for thinking about building an AI product in the face of competition from OpenAI and Big Tech. I am working on an AI product ( you can subscribe to this newsletter and be informed when I launch), and this framework guides me on which features to build or if I need to think harder.

In the face of disruption and obsolescence brought by ChatGPT, AI startups must confront the realization that their initial strategies might not suffice. The rise of ChatGPT has redefined the landscape. These startups must rethink their approach. Instead of clinging to outdated paradigms, they must embrace what can be. By leveraging open-source models (in LLaMA we trust) and frontier models, these startups must dismantle their obsolete product sense and rebuild. They should create innovative solutions teeming with potential and high user engagement. In this new era, AI startups can thrive, not by resisting change, but by adapting and using the tools at their disposal to build a future that meets the evolving needs of users.

Originally published at https://theriseofai.substack.com.

Introduction

Building a successful product is hard. The first step involves developing the product. Developers enjoy building, but there is more to a successful product than building. The quote “if you build it, they will come” no longer holds. To make a product successful, you need to market it, engage with customers, get feedback, and iterate on what to do next. The quote no longer holds for some reasons. The first reason is that hundreds and thousands of products compete for attention and space on your phone or browser tabs. So, if you build and expect people will come, you may be waiting for a long time. Customers are less likely to try new products unless they make their lives ten times better than the previous product or hack they currently use. Another reason is the market. Yes, you have built a great product, but it is not enough. You need to ensure that there is a market willing to pay for the product. This is known as product-market fit.

Nigeria

Let’s zoom in on Nigeria. According to the World Bank, Nigeria has about 200 million people. Some entrepreneurs tend to use this population figure to convince investors that they will make money because the population size is large. Most of the time, this is not true. Over and over, Nigerian (and African) startups raise funds at a high valuation, build a product, enter the Market, and get punched by reality.

Nigeria is the poverty capital of the world. Nigeria has more poor people than India, even though India has more than five times Nigeria’s population. The inflation rate for May 2024 is 33.2%. It is high and it does not appear to be going down anytime soon. The Nigeria-dollar exchange rate is high. A year ago, it was about 460 naira per dollar. In May 2024, it is over 1400 naira per dollar.

According to an article from BusinessNG, Netflix has about 169,600 paid subscribers out of 106 million bankable adults. South Africa has over 1 million paid subscribers. Furthermore, according to Techcabal, there were only 1.6 million admissions in 2023. Lagos alone boasts 20 million people, yet there were only 1.6 million filmgoers in Nigeria for nine months. In another Techcabal post, IROKOtv started 2022 with 192,174 active users, but by the end of the year, those numbers had dropped to 46,000, a 76% decrease. Since then, IROKOtv’s subscription numbers have steadily decreased.

All these go to show that the population is not equal to the target addressable market, Nigeria is poor country, and Nigerians are poorer than they were a year ago.

Why are Nigerian Startups not Building for the World

It all starts with an idea. Y Combinator’s motto is to make something people want. To build what people want, you need to understand what people want. It is not enough to have a good idea; it must resonate with the people you hope to sell to. You might ask yourself: How do you get ideas that might resonate with the world? The answers are as follows: new technology, new customer behaviour, and new regulations.

Let’s start with the new regulations and laws. New Regulation inspires new ideas and new startups. For example, in the United States. A new regulation enabled startups to sell cannabis over the counter. Another example was when Amazon’s One-Click patent expired. Some startups, such as Bolt, built this technology for other eCommerce stores. This is good, but for a Nigerian in Nigeria, it is difficult to keep up with new US regulations. Also, because it is a new regulation, I think the founders of the companies must be US citizens.

Secondly, new customer behaviour enables the formation of new startups. When Airbnb founded the company in 2008, the founders had a difficult time convincing both the hosts and the customers to use the service. The idea didn’t make sense. Why would anyone choose to offer their home to strangers, and vice versa? The 2008 housing crisis in the United States, which had a global impact, prompted people to seek alternative sources of income. Airbnb was there to provide the service. As a Nigerian in 2008, it was practically impossible because culturally we are a superstitious society. Even in 2024, it is relatively popular in only places like Lagos and Abuja. This idea will never come to a Nigerian mind let alone attempting to pitch it to VCs, early adopters, and the world. The Nigerian culture is conservative and less receptive to innovative technological solutions.

What about new technology?

New Technology is the Answer.

1. Build digital products. I love software. It is much easier to build and sell digital products. For distribution, you can pay Google Ads, and Facebook Ads and target customers in whatever part of the world you want to prioritize (US please :winking_face:).
2. Build products on top of global companies and platforms. There is a shared context and global target addressable users. For example, you can build a product on top of Shopify. Shopify has over 2 million merchants and 4.6 million websites. Shopify is a Canadian company, yet over 50% of its merchants are in the US. The United Kingdom and Australia are among the top three. If a Canadian company focuses on the US market, I think Nigerian startups should do the same. Another example is a Nigerian developer who built two Twitter bots. The bots were successful. They had over 500k followers each. The reason they were successful was because distribution was zero, there was shared context among all Twitter Users, and he monetized through Google Adsense. It was making about 1k USD monthly before Elon Musk took over. Of course, the revenue is small for a startup, but you get the gist.
3. Solve computer science problems. There is a shared context among tech companies and developers. The target addressable market includes tech companies and companies that use computer science technology. Hashicorp, for example, started with a small team building developer tools for infrastructure automation. Their first product, Vagrant, solved a specific development environment problem, and they later expanded to other infrastructure tools like Terraform. The company now generates hundreds of millions in revenue. Another example is Vercel, which started as a small team building deployment tools for frontend developers. They created Next.js, which became one of the most popular React frameworks, and now their platform is used by developers worldwide. A Nigerian developer built Chakra UI, a component library for React. He has started selling a pro version too. The product is popular among developers globally. These examples show that small teams can build significant businesses by solving specific computer science problems, regardless of their location.
4. Build tools for digital workers. There’s a growing global community of people who make their living through digital means — content creators, newsletter writers, social media managers, digital marketers, and online entrepreneurs. These digital workers constantly need better tools to create, manage, and grow their online presence. The advantage here is that these workers share similar needs regardless of geography. For example, YouTube creators worldwide use tools like TubeBuddy for keyword research and video optimization, newsletter writers use email marketing tools, and social media managers need scheduling and analytics tools. These tools can be built from Nigeria and sold globally. The target market is clear: companies hiring remote workers and independent professionals working digitally. This market is particularly attractive because digital workers are typically early adopters, willing to pay for tools that improve their workflow or increase their income, and they actively share useful tools within their communities. For example, Calendly, though not Nigerian, was built by a Nigerian founder and became successful by solving scheduling problems for professionals worldwide. The distribution can happen through platforms where these workers already gather — Product Hunt, Twitter, YouTube, or relevant online communities. Plus, you can start by solving one specific problem for a particular type of digital worker and expand from there based on market feedback.

What I would not do

Don’t try to build highly regulated industries. It is expensive to go global if you are a Nigerian in Nigeria. Many of them have died before selling to the world. One just died today.

Conclusion

When I was growing up, I read a story about hunters and birds. The birds realised that hunters had gotten good at shooting without missing. They advised that they should learn to fly without perching. This story also applies to Nigerian startups and Nigerians generally. The macroeconomic reality are like the hunters, and they have gotten good at shooting at Nigerian startups without missing. The solution is for Nigerian startups to learn to fly without perching. The only way is to build and sell to the world (US and Maybe Europe).

The Story

I was diagnosed with Amnesia. I forgot. I forgot my name, my vocabulary, programming, people’s names etc. I forgot how to brush my teeth. I forgot how to crack a joke. I forgot almost everything I knew. How did it happen? Let me say I don’t know.

On July 1st, at midnight, I sent a resignation email to my now ex-boss. Later that day, at 7 am, woke up to workout (I have a routine). At 7:15 am, I had a headache, so I decided to sleep. The next thing; I found myself in the hospital.

My landlady found me on the floor, with the door to my sitting room open. She called my estate agent, who then called my friends that “they don’t know what happened to me.”

The doctor thought I smoke and take hard drugs. My friends were sure I don’t. The doctor felt I did it on a low. My mum insisted that I don’t. One of my friends went to show the doctor my blood donation card. The last time I donated was April. I have donated blood 20 times — That’s a win. The doctor believed. The doctor asked for a clinical test to know what got into my bloodstream. Let’s say omo!

After about two weeks at the hospital, the doctor discharged me. After about four months, I remember most of the things I used to know (words and things). About coding? Omo!

I am grateful for my friends and family. “One who has unreliable friends soon comes to ruin, but there is a friend who sticks to closer than a brother.”

Wins

Before I forgot

• I got a new machine (read laptop).
• I got my first rented apartment and set up my kitchen. I don’t like cooking and I don’t cook, but having a functional kitchen makes me happy. It makes me sound like I am doing something with my life. Sometimes, late at night, I enter the kitchen to check it out.
• I bought a washing machine. About three years ago, I stopped washing my clothes. I used a washer.
• I bought a new phone.
• I bought a fridge.

After I forgot

• After about four months of resting and remembering, I can code again. I built an AI Twitter bot to summarize any type of information. I will launch in the new year.
• Improve my CSS skills. Learnt Tailwind CSS.
• Learnt Gridsome.
• I got into Y Combinator Future Founders Track.
• I learnt and improving my product design skills.
• I got a contract role, so yeah, I can code.
• Started a newsletter. Read a book in 15 minutes.

Lessons

• “When you are faced with the possibility of early death, it makes you realize that there are lots of things you want to do before your life is over.”
• People don’t die from not having a job. But my savings omo!
• Read books you love. Read summaries because there is so much to learn.
• Count your wins and name them one by one.

2021

Build.

About five years ago, if anyone told me that Nigeria will have four Google Developer Experts, I would say it is not possible, but oh well! that just happened and there are even more Nigerians that are qualified to be called “Google Experts”. These Google experts didn’t become experts because they were more “Nerdy” than every other developer, they became experts because of their contribution to the tech community in Nigeria. These experts have counted their “beans” and most importantly, how to communicate the value of their “beans”. This article is not one of those “become an expert in 7 days” category of articles. It is more about “fake it till you make it” type of article. This article is targeted at developers that want to join the train of Google experts. So if you want to be recognized as an expert, a celebrity developer; seat back, tighten your seat belt, shit is about to get real.

What is Public Speaking?

Let’s go back in time to when you were in secondary school and you were called out by your teacher to come read to the class, what you did then is called Public speaking. Project defence is another illustration of what public speaking is all about. For those that don’t know what happens in that room filled with bald-headed, pot-bellied lecturers; final year students are required to communicate the contents of their project work with these lecturers and the confidence exhumed by students goes a long way to determine their results. Lecturers don’t care whether you have faith in God or your name is Faith, your fate is partly determined by your public speaking ability. Even Jesus Christ and Mohammed were public speakers, so let’s not bring religion to this matter.

Debug your Public Speaking Skills

Presently, I am a student of the University of Ibadan, the President of my faculty Literary and debating society. I know a thing or two about public speaking, so if I see a public speaker, I immediately feel the connection. The first developer conference I attended was the first Python Nigeria conference and it held at Lagos. One of the speakers at the event was the fire brand developer and evangelist, Prosper Otemuyiwa. When he got on stage and started his talk, I felt the public speaking connection. The connection was stronger than the very popular 4G wireless network in Nigeria and it made me Smile (pun intended). I knew the son of Otemuyiwa prospered not only because he is a Laravel guru, but also because of his top-notch public speaking skill. If I have a chance to meet him again, I will ask my “Uncle”about his public speaking career, especially while at Uni. I have a strong feeling he was a public speaker.

I believe one of the reasons the audience enjoyed Uncle Prosper’s talk is because of the informal way he delivers his speeches. He makes sure his talks aren’t boring by adding comments that people find interesting and funny. Are you the type that comes on stage and it is business as usual? Boss, your audience will find it hard enjoying your talk, despite the fact you are speaking to technical people. We all know the popular saying that all work and no play makes Jack a dull boy. This time, “all talks with no laughs makes the audience bored AF”. Some people even go as far as making the audience bored to death. That shouldn’t be the way it should go. Even while coding, it is best practice to add comments, so my point is; what’s the point?

Public Speaking Syntax and Best Practices

Delivery: One of the very important skill required by a public speaker is delivery. This is how you say what you want to say. Delivery is a broad aspect of public speaking and many factors contribute to a good delivery.

• Confidence: The Merriam Webster dictionary defines confidence as the feeling or belief that you can do something well or succeed at something. This definition can be further explained with the following statement: Knowing that you know something. There are developers out there that are very good at what they do, but they don’t know it. This affects their confidence level. Nobody loves a speaker that isn’t confident. It makes people doubt your ability. Even while coding and you are not confident in your code, you may never click the “run” button.
• Eye contact: As developers we can stare at screens for hours, but we then start to feeble when we are on stage. Some developers stare so much at their screen that even bugs begin to debug themselves. Maybe it is karma, because I don’t understand why developers don’t look into the eyes of their audience. Eye contact is very important and I am not talking about screens, I’m talking about real people in real life.
• Practice: It is important to rehearse your presentation before you go deliver it. This will enable you become familiar with the talk. I agree that you wrote the talk, but don’t forget that practice makes perfect. Practice! Practice! Practice!

2. Comments: These statements determine whether your audience will enjoy your speech or not. It also goes a long way to indicate if your audience will even remember you spoke to them. What’s the point of a talk when the audience can’t remember? It is as good as never delivered. I was able to share my experience at Python conference, Nigeria, because I remembered Uncle Prosper’s talk, which indicates I wasn’t bored AF. Comments are very important because you never know the attention span of your audience, it is not like it is written on their forehead. Hence, comment your talks, not just your code.

• Sarcasm: This is not even important at all. What’s the point of making your audience enjoy your talk? I even heard that some people in the audience prefer to be bored to death. Some of them came to the conference because they are tired of being confronted daily by bugs and are just looking for an easy way to kick the bucket. So boss, kudos to you! You are doing a great job. That’s all I have to say before my audience start complaining that my medium post is getting too interesting and they are no longer interested. They “kuku” like boredom. Let me just stop here, I don’t want the post to become too good
• Exaggeration: Another way to make your presentation interesting is by exaggerating some obvious claims. Even Uncle Prosper exaggerates. When he was giving his presentation at the Python conference, he exaggerated a lot. He even went as far as saying that he attended a Jollof association in Lagos. I agree that I am an Ibadan boy but haha! I am not sure, but I think he even said something about “Association of Jollof Rice Ambassadors of Nigeria”. He further went as far as saying something about representing Nigeria at the Jollof Rice World conference. We know he is a Jollof rice ambassador, but which one is now association of Nigeria. Although it is not jollof rice, the takeaway is to exaggerate popular claims, it makes the audience laugh.
• Pun: This type of comment can be tricky and hard to use while giving your talks. A popular use of puns by a developer is Ace_Kyd’s “Open source is the new sauce”. That’s a very good use of puns. So as developers, let’s check PunHub (pun not intended). I mean sites that share puns so that we can become punny.

In conclusion, the power point of this post is that to excel as a developer, you need to have an edge over your counterparts and that can come with the way you use word(s). Remember don’t become the explorer that makes the audience crash.

In the spirit of GitHub’s Hacktoberfest, this post is licensed under the Creative Commons Attribution-Share Alike

This means:

• You are free to Share i.e. to copy, distribute and transmit this post

• You are free to Remix i.e. to adapt this post

• You are free to use it for commercial purposes