A 10-Step Guide to Future-Proof Your Business
Intelligence is dead, long live niche intelligence!
The rise of large language models (LLMs) like ChatGPT has fundamentally altered the value of intelligence in business. General intelligence and widely accessible knowledge are no longer competitive advantages. If your business relies on common knowledge — the kind ChatGPT can replicate — your business model will soon become obsolete.
The best survival tactic is to create a business where you will always be steps ahead of ChatGPT (I create custom LLMs all day so I know I am biased in my prognosis, but I do really believe this!).
My 10 Step Guide to Beat ChatGPT
1. Use the best LLM
This is not the same adage as, ‘If you can’t beat them, join them’, but more like, ‘Use the best of them, to beat them’. At the moment, for me that’s ChatGPT.
Why? Building ontop of the smartest LLM ensures you will always be at the cutting edge.
How? Evaluate leading LLMs based on their capabilities and alignment with your goals. Build your custom AI solution on this foundation.
2. Have specialist data
Data is the fuel for your LLM. The more unique and specialized your data, the greater your competitive edge.
Why? Proprietary data ensures your LLM answers questions in ways no off-the-shelf AI can replicate.
How? Collect data that reflects your customers’ unique needs, workflows, and preferences. This could include industry-specific insights, customer interactions, or performance metrics.
3. Utilise your industry expertise
Your team’s domain knowledge is crucial. Guide the LLM’s focus by applying expert insights.
Why? Expertise ensures your AI aligns with real-world problems and delivers high-value answers.
How? Engage domain experts to define key questions and validate outputs. Their intuition and “taste” will fine-tune your AI’s decision-making capabilities.
4. Augment your data
Specialized data is powerful, but it needs to be structured and augmented to maximize its utility.
Why? Clean, comprehensive data helps your LLM understand nuances and fill knowledge gaps.
How? Invest in data preprocessing, enrichment, and transformation. Use techniques like data labeling, normalization, and augmentation to ensure your LLM has a robust knowledge base.
5. Fine tune your LLM
Out-of-the-box LLMs provide generic answers. Fine-tuning aligns the model’s responses with your business’s unique requirements.
Why? Tailored models outperform generic ones by delivering more accurate, relevant answers.
How? Train the LLM on your proprietary data and industry-specific use cases. Regularly refine it based on real-world performance.
6. Use Dynamic agentic frameworks
Instead of relying on a single model, create systems where multiple LLMs collaborate dynamically. Leaps in super (general) intelligence will come from developments in this area.
Why? A network of specialized AI agents can tackle complex, multi-step problems better than a single model.
How? Develop frameworks that allow LLMs to share tasks, query one another, and work together to solve problems. This approach produces deeply reasoned and articulated answers at its best.
7. Create feedback loops
Your LLM should constantly improve through feedback and iteration.
Why? Continuous learning ensures your system stays ahead of competitors and adapts to changing demands.
How? Either use key KPIs and/or design workflows where users provide structured feedback on outputs. Use this feedback to retrain and refine the model over time.
8. Use human experts in the loop as guardrails
AI is powerful, but human intuition remains irreplaceable.
Why? Experts can identify gaps, stress-test outputs, and provide nuanced judgment that AI lacks.
How? Establish roles for human experts to review, monitor, and guide your LLM’s performance. This ensures high-quality, reliable results.
9. Supercharge content generation
Leverage your custom AI to generate content faster and more cost-effectively.
Why? Efficient content generation frees up resources for innovation and optimization.
How? Use your LLM to produce text, code, visuals, videos, dynamic BI, presentations etc… This enables your team to focus on higher-level tasks.
10. Create use case specific workflow UX
Finally, embed your custom LLM in UI that evokes Marie Kondo levels of joy in the daily working lives of your users & colleagues.
Why? Intuitive, delightful user experiences drive adoption and maximize your system’s impact.
How? Invest in UI/UX design that simplifies workflows and enhances productivity. Tailor interfaces to your users’ specific needs.
Conclusion: Thrive in the Age of AI
The commoditization of general intelligence is a challenge, but it’s also an opportunity. By following these 10 steps, your business can carve out a competitive niche, deliver unique value, and stay ahead of even the smartest generic AI models.
In the rapidly evolving world of LLMs, adaptability is your greatest strength. Start building your moat today, and ensure your business thrives in the age of niche intelligence.
