The AI Savings Question: Why the Answer Depends on Who Is Counting

The Survey Nobody Wants to Fail

A question is circulating across the legal industry this week: has law firm AI use actually delivered cost savings for clients? It sounds straightforward. It is anything but. The difficulty is not technological. It is definitional. Cost savings compared to what baseline? Measured over which time horizon? Reported by whom, the firm or the client? Until those questions have agreed answers, every survey on the topic will produce noise rather than signal.

For in-house teams, this ambiguity is more than academic. General counsel are under board-level pressure to demonstrate that their external spend is being managed intelligently. If firms are deploying AI and not passing efficiency gains downstream, that is a governance question as much as a commercial one.

What Firms Are Actually Optimising For

Law firms have historically billed by the hour. AI tools that reduce hours should, in theory, reduce bills. In practice, firms are navigating a tension: they are investing heavily in AI infrastructure and training, and they need a return on that investment. The rational short-term response is to use AI to do more work at the same price, rather than the same work at a lower price.

This is not cynicism; it is economics. A firm that cuts its revenue to reflect AI efficiency gains before it has recovered its implementation costs will struggle to fund the next generation of tooling. Clients who understand this dynamic are better placed to negotiate intelligently, linking fee arrangements to output quality and turnaround speed rather than hours logged.

The firms that will win long-term client relationships are those that are transparent about where AI is deployed and what that means for the value delivered, not just the time recorded.

The In-House Perspective Is Different

When an in-house legal team deploys AI directly, the calculus changes entirely. There is no billing relationship to preserve. Efficiency gains translate immediately into capacity: the same team can handle more contracts, more negotiations, more review cycles without adding headcount. That is a saving that shows up in the legal department budget and, eventually, in the broader business.

This is precisely the territory where contract lifecycle management matters most. A company processing hundreds of commercial agreements each month is not going to see meaningful savings from a tool that drafts faster if that tool still requires a lawyer to reconstruct context from scratch every time. The compounding benefit comes when the AI understands the company's own positions, its standard fallbacks, its risk tolerances, and the legal requirements of the jurisdiction where each contract operates. Speed without that context is just faster guessing.

Adira is built around exactly this principle. Drafting in a company's own voice, reading contracts from the client's side of the table, and applying jurisdiction-specific legal knowledge means that the time savings are real rather than cosmetic. A first draft that already reflects the company's negotiating posture is not a starting point; it is a material head start.

Measuring the Right Things

The industry would benefit from a cleaner framework for measuring AI value in legal work. Three metrics are worth tracking seriously.

First, cycle time: how long does it take from contract request to signed agreement? AI that genuinely accelerates this reduces commercial risk and opportunity cost, neither of which appears on a legal invoice but both of which matter to the business.

Second, exception rate: what proportion of contracts require escalation to senior lawyers or external counsel? Good AI should handle routine matters routinely, freeing human judgment for genuinely complex questions.

Third, consistency: are the company's standard positions being applied uniformly across contracts, geographies, and counterparties? Inconsistency is a hidden cost that rarely appears in any survey but creates significant liability exposure over time.

Firms and clients who agree on these metrics before deploying AI will have a much more productive conversation about value than those who rely on hourly rate comparisons alone.

The Honest Conclusion

The question of whether AI has delivered cost savings for clients does not yet have a clean industry-wide answer, because the industry has not yet agreed on what it is measuring. That is a solvable problem, but it requires both sides of the client-firm relationship to move beyond the comfort of familiar billing structures.

For in-house teams willing to bring AI capability inside the department rather than waiting for firms to pass savings along, the opportunity is immediate and measurable. The tools that deliver genuine value are those built for the specific realities of contract work: legal context, commercial voice, and jurisdictional accuracy. Everything else is productivity theatre.