Richard Hartshorne, Business Unit Director at Expleo.
At a recent roundtable discussion with leaders from major UK law firms, the conversation quickly moved beyond the potential of AI and into a more practical challenge: how to avoid vendor lock-in and ensuring today’s AI decisions do not become tomorrow’s constraints.
AI adoption within the legal sector has moved at an extraordinary pace. What started as a handful of pilots has evolved into a rapidly expanding ecosystem of legal AI assistants, document review platforms, research tools, copilots and emerging agentic systems.
As firms evaluate AI investments, the conversation is increasingly shifting from “Which tool should we buy?” to “How do we maintain flexibility while controlling cost and risk?”
Vendor lock-in has become a growing concern. Selecting the wrong platform, building processes around proprietary tools or becoming dependent on a single AI provider can limit future choices and increase costs at a time when the market continues to evolve at speed. Richard Hartshorne, Business Unit Director at Expleo, explores this in more depth.
Why vendor lock-in is becoming a legal sector issue
Discussions among legal technology leaders reveal a common theme: firms are struggling with the proliferation of AI tools and overlapping capabilities. Many are still trying to determine which solutions deliver meaningful value, while others are already reviewing their second or third generation of AI investments.
The challenge is that AI technology is changing faster than traditional software procurement cycles. A platform that appears market-leading today may be overtaken by a newer model, different deployment approach or more cost-effective alternative within months.
As a result, firms are becoming increasingly concerned about rising licensing costs, duplicated capabilities, integration challenges, difficulty demonstrating ROI and the prospect of becoming tied to a single provider just as new alternatives emerge.
For legal leaders under pressure to balance innovation, governance and profitability, maintaining optionality is becoming a strategic priority.
Focus on capabilities, not products
One of the most effective ways to reduce lock-in is to build organisational capabilities rather than product-specific workflows.
Several legal technology leaders are already encouraging teams to develop skills that work across multiple AI platforms rather than becoming dependent on a single interface or vendor. This approach helps future-proof investments and makes it easier to adapt as technologies change by minimising the risk to firms of becoming too dependent on a single product.
For example, lawyers who understand effective prompting, verification techniques, data management and workflow design can transfer those skills across different AI tools. The same cannot be said for processes built entirely around one proprietary platform.
The objective should be to create AI-enabled professionals rather than firm-wide dependence on any individual supplier.
Build around data, not tools
Another recurring lesson emerging from law firms’ AI journeys is that successful AI adoption depends far more on data quality than on the choice of model itself.
Several firms report that disappointing AI results were ultimately traced back to poor-quality or fragmented data. As a result, some have paused AI initiatives in order to strengthen data foundations and governance frameworks first.
Strong data governance provides two significant benefits. First, it improves AI outcomes regardless of which model is being used. Second, it reduces switching costs. When data is organised, governed and accessible through appropriate controls, firms can evaluate new AI tools without having to rebuild information structures each time. In other words, firms that own their data strategy retain greater freedom over their technology strategy.
Demand deployment flexibility
When evaluating AI platforms, law firms should also look beyond features and assess deployment flexibility.
Key questions include:
- Can the solution support multiple LLM providers?
- Can models be changed without disrupting workflows?
- Does the organisation retain control over data and governance?
- Are cloud and private deployment options available?
- Can AI capabilities integrate with existing systems and processes?
Solutions that separate workflows from underlying AI models provide greater resilience against market changes.
This is particularly important as different models continue to demonstrate strengths in different areas, whether legal research, summarisation, drafting, coding, compliance analysis or business operations.
Bring AI spend under control
Cost management is becoming a board-level concern. Many firms are still struggling to measure AI ROI accurately, while others are wrestling with growing numbers of overlapping licences and applications.
One emerging approach is to apply governance not only to AI usage but also to model selection.
Rather than allowing users to choose whichever model they prefer, firms are beginning to consider policies that align workload types with the most appropriate systems. Simple administrative tasks may be routed to lower-cost models, while complex legal analysis can be reserved for more advanced options.
This creates opportunities to optimise AI spending without limiting user access to innovation.
The future may be model-agnostic
Perhaps the most interesting discussion emerging from legal technology leaders is the growing belief that selecting individual AI tools may become a temporary problem rather than a permanent one. As AI ecosystems mature, there is increasing interest in platforms capable of analysing requests and directing them to the most suitable model based on performance, cost, governance requirements or task complexity.
This model-agnostic approach has the potential to reduce both vendor lock-in and AI expenditure by preventing organisations from becoming overly dependent on any single provider.
Platforms such as AI orchestration layers, LLM management frameworks and model-flexible environments are beginning to support this strategy. Solutions such as Expleo’s AI-powered development accelerator, CodePle, demonstrate how organisations can maintain flexibility through support for multiple AI models, customer-managed LLM strategies and the ability to switch models without changing developer workflows.
For law firms navigating a rapidly changing AI landscape, that flexibility may prove just as valuable as the capabilities of any single model.
Because in a market that evolves every few months, the ability to change direction quickly could become the most important investment decision of all.
Food for thought. Same time next year?
