An Insidious Threat
By the end of 2025, the torrent of uncannily crafted yet deeply mediocre AI-generated content engulfing us had swelled to a tidal wave. It is therefore unsurprising that the editors of Merriam-Webster chose “slop” (being “digital content of low quality that is produced usually in quantity by means of artificial intelligence”) as their 2025 word of the year. For the legal sector, I propose a corollary: “strategy slop.” This means law firm strategy based on output that is derived indiscriminately from Large Language Models (LLMs).
The risk of strategy slop is not that it is wrong, although it might be. It is often well-written and superficially plausible. The risk is that it is banal. It resembles a fitness strategy consisting of: “don’t smoke, cut carbs, and run five miles a day.”
Strategy slop sabotages a firm’s strategising in several insidious ways:
False precision:Polished frameworks and confident language mask a lack of traceable assumptions, data provenance, or sensitivity analysis, creating a dangerous illusion of accuracy. Worse, if they sycophantically tell you what the LLM thinks you want to hear.
Category errors: Concepts are transplanted from other industries without being calibrated to the realities of the legal sector or a law firm’s specific circumstances.
Benchmark fetishism: League tables and media coverage are treated as causal evidence of good strategy rather than the noisy signals they often are. This creates a bias toward headcount or revenue growth, or profit per equity partner (PEP). potentially triggering mergers when alternative strategies might deliver better sustained competitive advantage.
Homogenisation: Firms converge on “vanilla” strategic priorities such as “digital transformation,” “client centricity,” or “innovation”, eroding differentiation and stalling crucial deep discussion about what really drives sustainable competitive advantage.
Implementation collapse: Strategies are proposed without regard for partner buy-in, or economics, or changes that would be needed to culture, governance and other systems and structures. So, nothing material changes.
An example of homogenisation
Dan Diasio, EY’s Global Consulting AI Leader, provided a good example of strategy slop on a recent interview on Microsoft’s WorkLab podcast. To demystify AI for clients, his team ran hundreds of masterclasses using Generative AI to design products, marketing plans, and even advertising jingles. It was a lot of fun. Upon review, though, the team noticed that the results were all uncannily similar. Diasio concluded that while AI makes it easy to produce something adequate, it inevitably commoditises the output.
To achieve uniqueness and excellence, one must combine the right tools with four specific human attributes: creativity, critical thinking, systems thinking, and deep domain expertise. The shift toward agentic AI will not alter this reality because agents are subject to the same statistical regression to the mean as are the underlying LLMs. To avoid strategy slop, the workflow must remain: human > AI > human.
From incremental to disruptive innovation
Avoiding strategy slop will become even more critical as the industry shifts from using AI for incremental optimisation (enhancing current workflows) to radical reinvention.
In 2025, most firms focused on incremental improvements or dabbled in deeper pilot studies in business services and other areas where failure carried less risk. These initiatives were valuable for raising awareness and yielding efficiency gains. If 2026 brings to LLMs the speed of advance that we have seen thus far, it is possible that we might hit a disruption point in the next twelve to twenty four months. Rivalry will then likely become exponential. Lagging firms will find it increasingly difficult to catch up. Some might even be unable to, and fail.
Avoiding strategy slop
For LLM-assisted output to be valuable for strategy, it must satisfy four criteria: (a) it must answer specific strategic questions, (b) be based on clear, testable assumptions, (c) provide a secure evidence chain, and (d) define necessary operational changes.
If any of these are missing, the result is strategy slop. This is how to avoid that:
Define the “Essential Elements of Intelligence” (EEIs): Borrowing from military intelligence, identify the 3 to 5 most crucial strategic questions that must be continuously on your radar over the next 12 to 24 months.
Build a secure, carefully curated knowledge corpus for strategising: A law firm’s strategy corpus should contain high-value, low-noise material. This includes internal “ground truth” data (economics beyond the P&L, client demand trends, workflows) and external intelligence. Strict rules regarding metadata and inclusion are essential to maintain integrity, prevent hallucinations and minimise data leakage.
Interrogate options rigorously: Strategy slop thrives in ambiguity. Counter this by breaking issues into discrete parts and using “first principles” thinking. Instruct the LLM, again and again if necessary, to challenge how options enhance competitive advantage and to stress-test the resources required for delivery.
Map actions to decisions: Law firms rarely fail at making decisions. More frequently, they fail at converting decisions into behaviour. This usually occurs when new goals misalign with existing incentives and compensation structures. AI allows one to model these variables far more rigorously and present leadership with better worked-up choices, than before.
Conclusion
In an AI-saturated environment, disciplined sense-making is the new currency. Wrong decisions can cost a firm dearly and right decisions mean little if they are not translated into operational reality. Law firms that couple AI tools with the necessary human creativity, critical thinking, systems thinking and deep domain knowledge will inevitably outperform those that rely on strategy slop to navigate their futures.
