In today's rapidly evolving technological landscape, businesses are increasingly harnessing the power of Large Language Models (LLMs) to transform industries. A recent panel discussion, Eheads 15: AI Fireside Chat, brought together CEOs and CTOs from three AI-driven companies to share insights on integrating LLMs into their solutions. The panel featured Beato Bongco from Anycase, Ragde Falcis from ChatGenie, and Kevin Philip Gayao from IOL INC.

Building AI Businesses: From Ideas to Market

Each of these companies started as side projects before scaling into full-fledged businesses. Anycase, initially targeting B2B law firms, pivoted to a B2C model after realizing it was easier to onboard individual lawyers rather than convincing entire firms. Today, they serve over 5,000 users, with 1,000 paying customers.

ChatGenie began as an AI experiment at a Techstars demo event, eventually developing a chatbot system that functions like a human sales representative. IOL INC took a more pragmatic approach, initially learning AI technologies out of necessity to secure funding, eventually building over 20 AI-powered business tools.

Each business has embraced LLMs differently, but all agree that “adapting to user needs” and “iterating quickly” are key to staying competitive.

The Tech Stack & Cost Considerations

The core technologies powering these businesses include:

• Model Provider APIs: OpenAI, Anthropic, AWS Bedrock, and Together.ai.

• Cloud Services: AWS, GCP, and Azure provided infrastructure support, with cloud credits being crucial in the early stages.

• Vector Databases: Pinecone and other retrieval-augmented generation (RAG) tools ensured AI responses were grounded in accurate data.

• AI Development Tools: GitHub Copilot, Amazon Q, and other AI-powered IDEs enhanced coding efficiency.

To manage costs, Anycase optimized every part of their inference pipeline, while ChatGenie monitored input/output tokens and flagged spam interactions. IOL INC strategically shifted API costs to B2B clients and placed soft usage limits on B2C users.

Challenges: Hallucination, Data Quality, and Model Accuracy

One of the most persistent challenges in AI development is hallucination—where AI generates incorrect or misleading outputs. The panelists shared strategies to mitigate this:

• RAG Triad Approach: Measuring the correctness of AI-generated responses, the relevance of contextual data, and the model’s reliance on external knowledge.

• Chain-of-Thought Prompting: Using structured AI reasoning to improve accuracy.

• Multi-Agent Frameworks: ChatGenie implemented an orchestrator AI agent that dynamically determines which AI subsystems to engage, reducing hallucination rates.

• Model Selection & Fine-Tuning: IOL INC explored different models to optimize responses for specific use cases.

Beyond hallucinations, maintaining data quality was a significant focus. Anycase employed lawyers to validate AI-generated legal documents, benchmarking performance against the Philippine Bar Exam. Meanwhile, ChatGenie flagged inconsistencies in customer FAQs, ensuring businesses kept their information up-to-date.

User Feedback & Business Impact

User feedback played a crucial role in refining these AI products:

• ChatGenie saw success when users couldn't distinguish between AI and human sales representatives.

• IOL INC helped an Australian law firm save significant time by automating document workflows.

• Anycase had a breakthrough moment when a lawyer witnessed AI-generated contracts in real-time.

Each company emphasized the need for continuous improvement based on user interactions to ensure their AI systems remain effective and trustworthy.

The Future of AI in Business

TThe panelists expressed optimism about AI’s potential while acknowledging the risks of job displacement. However, they framed AI as a tool to augment human intelligence rather than replace it.

ChatGenie highlighted the role of AI in automating mundane tasks, allowing humans to focus on creative and high-value work. Anycase advocated for using AI to make specialized knowledge more accessible, particularly in underserved areas. IOL INC, with its accounting background, noted the inevitability of AI in automating financial and legal processes while stressing the importance of regulatory frameworks.

Conclusion

The discussion underscored that while AI adoption presents challenges, the opportunities far outweigh the risks. Businesses that leverage LLMs effectively can drive efficiency, reduce costs, and enhance customer experiences. However, success depends on responsible AI development, robust data validation, and continuous user-driven improvements.

As AI continues to evolve, these companies demonstrate how strategic implementation can lead to real-world impact—reshaping industries, improving access to services, and setting new standards for AI-powered business solutions.