The idea of technology augmenting human work isn’t new, despite the current hype around AI. At home in Montana,  I don’t know any farmers who till vast fields without tractors, relying solely on manual labor. I certainly don’t know any accountants who forgo tools such as Excel to manage complex data sets by hand. For decades, we’ve embraced tools that make us more efficient, handle laborious tasks, and allow us to achieve more than we could alone. Tractors didn’t replace farmers; they allowed farmers to cultivate more land and produce more food. Excel didn’t replace accountants; it enabled them to perform more complex analyses and provide greater strategic value.

Generative AI is the next evolution in this story. While the possibilities are vast, I believe its most powerful immediate impact lies in its ability to automate the redundant, repetitive, and often tedious parts of our jobs. Think about the time spent summarizing notes, drafting standard emails, searching for information across disparate systems, or formatting reports. These are the modern equivalents of manually driving steel spikes. AI can handle these tasks efficiently, freeing up human workers to focus on critical thinking, creative problem-solving, complex negotiations, and building relationships – the areas where human insight and empathy truly shine.

However, making this transition shouldn’t fall solely on employees’ shoulders. Companies have a crucial role and responsibility in navigating this shift. We can’t simply introduce powerful new tools and expect everyone to adapt seamlessly and safely. Organizations should provide:

1. Support & training: Helping employees understand how AI tools can genuinely assist them in their specific roles, not just expecting them to figure it out.
2. Clear guidelines: Establishing policies around how AI should be used is critical. For example, you absolutely wouldn’t want employees uploading sensitive company or client data into an unprotected public AI model. Clear do’s and don’ts are essential for security and compliance.
3. Focus on augmentation: Framing AI as a partner or assistant, designed to enhance human capability, not replace it entirely.

John Henry didn’t have the option to partner with the steam drill; his story was framed as a zero-sum competition. Today, we have a choice. The future isn’t about people racing against AI; it’s about people working alongside AI. By embracing AI to handle burdensome tasks and providing the right support and guardrails, we can empower our teams to be more productive, more strategic, and, ultimately, find more value in their work.

The post What John Henry and farmers can tell us about the future of AI appeared first on Larridin.

At Larridin, we initially invested heavily in fine-tuning models using Low-Rank Adaptation (LoRA). These specialized models effectively picked up domain-specific nuances, delivering solid results early on.

However, AI moves at lightning speed. In recent weeks, frontier models like Gemini 2.5, GPT-4.5, and Claude Sonnet 3.7 have advanced dramatically. With carefully crafted prompts, these models outperform our finely-tuned LoRA solutions significantly.

This rapid evolution presents a strategic question: Should organizations continue to invest in fine-tuning specialized models, or leverage the raw power of frontier LLMs?

Fine-Tuning (LoRA + RAG): Specialized Precision

In fine-tuning, small adapter layers are trained atop large base models, optimizing them for specific tasks without retraining the entire model.

Advantages:

• Data Privacy & Security: Models run locally, keeping sensitive data securely contained.
• Cost Efficiency & Speed: Lean models ensure lower costs and rapid inference, ideal for real-time use cases.
• Deep Domain Expertise: Highly accurate for specialized tasks and organizational contexts.

Challenges:

• Resource Intensive: Requires sustained investment in MLOps infrastructure, data curation, and model management.

Frontier Models (Large LLMs + Prompt Engineering): Broad Excellence

Alternatively, organizations can directly leverage advanced, general-purpose models, utilizing detailed prompting to guide their reasoning.

Advantages:

• Advanced Reasoning: Immediate access to cutting-edge AI capabilities.
• Rapid Deployment: Crafting effective prompts typically outpaces full model fine-tuning.

Challenges:

• Privacy Concerns: Data typically processed externally, requiring rigorous anonymization and compliance measures.
• Higher Costs & Latency: API-driven inference can be expensive and slower.
• Prompt Expertise Needed: Achieving optimal results demands expert-level prompting.

Our Hybrid Approach at Larridin

At Larridin, we’ve adopted a strategic hybrid model:

We use Frontier Models when:

• Complex, generalized reasoning capabilities are crucial.
• Data privacy can be effectively managed through anonymization.
• Premium insights justify additional inference costs.

We deploy Fine-Tuned LoRA Models when:

• Strict privacy, regulatory compliance (GDPR, HIPAA), or local deployment are essential.
• Efficiency and scalability at lower costs are critical.
• Specialized tasks require deep understanding of domain-specific nuances.

Frequently, fine-tuned models handle initial data processing—such as filtering or anonymizing, classification—before leveraging frontier models for deeper analysis. We continuously evaluate our approach to adapt to evolving AI capabilities.

Privacy and Ethics: Non-negotiable

Our commitment is clear: analyzing productivity trends and organizational effectiveness—not individual monitoring. We maintain rigorous standards of privacy and ethical data handling regardless of the AI techniques deployed.

Final Thought

Selecting between fine-tuned models and frontier LLMs isn’t just a technical choice; it’s a strategic investment decision. By employing a balanced hybrid approach, we ensure our clients receive precise, actionable insights tailored exactly to their needs, also allowing us to take advantage of future developments.

P.S. At the time of writing, LLMA4 has just been released along with LLMA4 Scout—a smaller, faster model featuring a 10M token context window. This recent development further validates our hybrid approach

The post Fine-Tuning vs Frontier Models: Making the Right AI Investment appeared first on Larridin.

• Frederick Winslow Taylor revolutionized workplace efficiency in the early 20th century by introducing the principles of scientific management, which focused on optimizing tasks through time studies and standardization.
• Henry Ford dramatically improved manufacturing productivity by pioneering the moving assembly line and implementing innovative workforce management practices like the $5 workday (well above market) and the 40-hour work week.
• Jeff Bezos fostered a highly productive environment at Amazon by relentlessly focusing on the customer, implementing innovative operational mechanisms, and establishing a strong set of leadership principles that emphasized efficiency and results.

What all these luminaries have in common is that they found themselves at a point in time where it was possible to make a step function change in productivity, and they found a way to think differently to innovate. With the emergence of GenAI and automation, we are entering a new era where the nature of work itself is transforming. Integrating human ingenuity with AI capabilities demands more than just optimizing existing processes; it requires a dynamic, adaptive approach to how organizations function. With unprecedented rapid technological advancements, there is the need to constantly adapt and grow.

What is Organizational Fitness?

At its core, Organizational Fitness is a company’s capacity to continuously adapt, grow, and thrive by understanding and optimizing the interplay between its people, processes, and technology.

This concept of fitness is crucial because it acknowledges that businesses, like living organisms, need continuous adaptation and optimization to thrive. It’s not a one-time fix but an ongoing process of assessment, adjustment, and improvement in response to internal and external factors. This is in stark contrast to static organizational health metrics that provide a snapshot in time but fail to capture the dynamic nature of a company.

Traditional approaches to measuring productivity often focus narrowly on individual actions and fall short, failing to capture the collective effectiveness of teams, especially when leveraging tools like Generative AI. Employee monitoring software ‘bossware’ exemplifies this by tracking granular individual activity—like keystrokes or screen time—instead of the meaningful outcomes produced by the team as a whole– people and AI combined. 

Organizational fitness, on the other hand, aims to provide a more holistic and actionable view. By understanding what’s happening in the systems where work gets done, this approach identifies areas where the organization is thriving and areas that need improvement, enabling leaders to make strategic adjustments that enhance performance, efficiency, and employee well-being.

If you want to look at organizational fitness, there are five key differentiators.

1. Focus on outcomes and value creation This approach emphasizes what the organization and its people are actually achieving and contributing, not just the activities they perform. It accounts for the impact of tools like AI, which can enhance productivity, rather than simply measuring inputs like keystrokes or time spent online. This differs from employee monitoring (focused on activity) and organizational health (focused on enabling conditions, not direct output measurement).
2. Strategic alignment and capability building The focus is on ensuring that individual and team efforts directly contribute to the organization’s strategic goals. It prioritizes clarity of purpose, talent development aligned with business needs, and efficient resource allocation to maximize impact, unlike organizational health surveys that may only gauge surface-level employee sentiment or monitoring tools which lack strategic context.
3. Data-driven insights from work systems Analysis and information are derived from the systems where work is actually done, such as communication platforms, project management tools, and business applications. This provides unbiased, contextualized insights into real-world workflows and performance, contrasting with the often-isolated nature of employee monitoring metrics or the subjective, survey-based data common in organizational health assessments.
4. Enabling performance and streamlining complexity The aim is to empower employees and teams to work more effectively and achieve their full potential. It’s about making complex work easier and more productive by identifying and removing systemic friction, not about implementing punitive measures or creating a culture of distrust, as is often the perception with employee monitoring software. Organizational health supports well-being but may not directly address operational complexity.
5. Systemic and dynamic views vs. individualistic or static snapshots This perspective analyzes the interconnectedness of teams, processes, and workflows across the entire organization, treating it as a dynamic system. It seeks to understand how work flows and evolves continuously over time to enable ongoing adaptation and improvement. This contrasts sharply with the often individual-employee focus of monitoring software and the periodic, static snapshot nature of many organizational health assessments (like annual surveys), which may not capture the fluid reality of how the organizational system operates day-to-day.

In conclusion, moving beyond traditional productivity metrics toward Organizational Fitness isn’t just a semantic shift; it’s a strategic necessity. In an era defined by rapid technological change and the integration of AI, understanding and enhancing fitness allows companies to not only navigate complexity but also to unlock new levels of performance, innovation, and employee potential. It’s about building organizations that are resilient, adaptive, and truly fit for the future.

The post Beyond Productivity: Why Organizational Fitness Matters in the Age of AI appeared first on Larridin.

Over the last year, we talked with thousands of our former customers and partners. Again and again we heard the same thing from CEOs, CFOs, Heads of HR, as well as other business leaders – the current tools to measure and understand corporate productivity are severely outdated and not helpful in a world of AI.

Every company now has humans working alongside AI and, in some cases, machines/robots.

Our solution is designed to fill the measurement void and help prioritize investment decisions across people, processes, and deployment of AI.

After 25+ years building enterprise technology companies with my co-founder Jim Larrison, we’re ready to equip organizations for the AI-powered future along with our new partner, Ameya Kanitkar and our amazing growing team.

Let’s build the future of work, together.

Reach out if you’re interested in learning more!

The post The Larridin Story appeared first on Larridin.