Before generative AI, AlphaSense built its sentiment analysis model by employing a large team for years to manually tag financial statements. This highly specialized, narrow AI still surpasses the performance of today's more generalized Large Language Models for that specific task, proving the enduring value of focused training data.
Related Insights
Recognizing there is no single "best" LLM, AlphaSense built a system to test and deploy various models for different tasks. This allows them to optimize for performance and even stylistic preferences, using different models for their buy-side finance clients versus their corporate users.
LLMs have hit a wall by scraping nearly all available public data. The next phase of AI development and competitive differentiation will come from training models on high-quality, proprietary data generated by human experts. This creates a booming "data as a service" industry for companies like Micro One that recruit and manage these experts.
Simply offering the latest model is no longer a competitive advantage. True value is created in the system built around the model—the system prompts, tools, and overall scaffolding. This 'harness' is what optimizes a model's performance for specific tasks and delivers a superior user experience.
For specialized, high-stakes tasks like insurance underwriting, enterprises will favor smaller, on-prem models fine-tuned on proprietary data. These models can be faster, more accurate, and more secure than general-purpose frontier models, creating a lasting market for custom AI solutions.
The notion of building a business as a 'thin wrapper' around a foundational model like GPT is flawed. Truly defensible AI products, like Cursor, build numerous specific, fine-tuned models to deeply understand a user's domain. This creates a data and performance moat that a generic model cannot easily replicate, much like Salesforce was more than just a 'thin wrapper' on a database.
Startups like Cognition Labs find their edge not by competing on pre-training large models, but by mastering post-training. They build specialized reinforcement learning environments that teach models specific, real-world workflows (e.g., using Datadog for debugging), creating a defensible niche that larger players overlook.
Instead of a generalist AI, LinkedIn built a suite of specialized internal agents for tasks like trust reviews, growth analysis, and user research. These agents are trained on LinkedIn's unique historical data and playbooks, providing critiques and insights impossible for external tools.
Current AI models resemble a student who grinds 10,000 hours on a narrow task. They achieve superhuman performance on benchmarks but lack the broad, adaptable intelligence of someone with less specific training but better general reasoning. This explains the gap between eval scores and real-world utility.
The company developed an AI that conducts highly technical expert network interviews, automating a high-friction manual process. This enables new, scalable content creation like monthly channel checks across dozens of industries—a task too repetitive for human analysts to perform consistently at scale.
To analyze brand alignment accurately, AI must be trained on a company's specific, proprietary brand content—its promise, intended expression, and examples. This builds a unique corpus of understanding, enabling the AI to identify subtle deviations from the desired brand voice, a task impossible with generic sentiment analysis.
