A patent firm agrees to prepare an application for a fixed fee. It then uses an AI drafting tool that substantially reduces the time needed to produce the first draft. The application is reviewed by the same attorney and delivered to the same professional standard. Should the client pay less?
The client reasonably expects to share in the saving. The firm has invested in the technology, remains responsible for the work and accepted the risk that the assignment might require more effort than anticipated.
The position advanced here is that AI should change fees where it changes the economics of the work. Savings in document preparation should be measured and reflected in pricing decisions, while claim strategy and substantive review should be assessed on their own terms. An across-the-board discount is no more convincing than leaving every fee unchanged.
For a managing partner, the question is also how long an efficiency gain can remain a margin improvement before clients or competitors press for different terms. That period offers a choice: reduce prices, improve the defined service or invest in the capabilities that justify the fee.
Measuring the Savings in Patent Work
Producing a draft and completing a patent application are two different tasks. AI may shorten initial drafting while leaving inventor discussions, technical verification and substantive review largely unchanged. In other matters, it may reduce effort across several stages. Sometimes correcting its output may consume much of the apparent saving.
The useful comparison is the total work required to deliver a result of the same professional standard. Comparing the time needed to generate text with the time previously needed to prepare a filing-ready patent application will overstate the benefit.
Consider a hypothetical application that previously required twelve hours of preparation and four hours of review. With AI assistance, preparation takes six hours, but review and correction take five. Drafting time has fallen by half, while total professional time has fallen from sixteen hours to eleven, a reduction of approximately 31%.
This calculation illustrates the difference between drafting time and total professional time. It is not an estimate of the savings patent firms should expect from AI.
Nor does it necessarily mean that the cost of delivering the work has fallen by 31%. That depends on who performs the work, their internal cost and the technology expenses. If the additional review requires a more senior practitioner, the monetary saving may be smaller.
Under hourly billing, assuming the rate remains unchanged, the five hours saved would reduce the professional fee accordingly. Under a fixed-fee agreement, the saving would affect profitability, the assessment of fee reasonableness and future pricing decisions.
What Does the Research Show About AI Efficiency in Patent Drafting?
To understand how AI may affect patent drafting fees, it is necessary to examine whether it actually reduces working time and whether those savings come at the expense of professional quality.
A study by David Autor and his co-authors, Does AI Assistance Enhance or Erode Expertise?, examined the impact of AI on 133 patent lawyers at eleven intellectual property firms in the United States. The three-month study found that lawyers using AI produced higher-quality patent application drafts than those who did not use it. The improvement was particularly pronounced among junior lawyers. According to Google’s account of the research, junior lawyers completed one editing task in approximately 15% less time than the control group.
The study also identified an interesting limitation. When participants were asked to review and correct a patent application without AI assistance, senior lawyers showed improvement in their independent professional performance, but junior lawyers showed no comparable average improvement. In other words, AI helped junior lawyers produce better work without necessarily improving their ability to perform that work independently.
For patent drafting fees, the findings suggest that AI can improve efficiency at the drafting stage, but they do not eliminate the need for professional review and the patent attorney’s judgment. Moreover, the approximately 15% time saving was measured on a particular task in an experiment, not over the complete preparation of a client’s patent application. It therefore does not establish that the overall fee should fall by a similar percentage. That requires examining savings across the entire workflow, including review, revisions and professional responsibility.
How Does AI Change the Economics of Hourly Billing?
Hourly billing establishes a direct relationship between time spent and the professional fee. AI challenges the economics of that model: the faster a firm completes the work, the fewer hours it can bill, even when the professional result is unchanged or improved.
The economic implications go further than they may first appear. Suppose AI reduces the average time spent on a matter from ten hours to seven. Assuming the hourly rate and mix of work remain unchanged, the firm would need to handle approximately 43% more matters to maintain the same hourly-billing revenue. If demand does not grow accordingly, greater efficiency may reduce revenue while much of the firm’s expenditure, particularly employee salaries, remains unchanged.
A further problem is that AI does not necessarily save time uniformly. It may primarily reduce the hours spent by junior patent practitioners preparing initial drafts, while the need for review by a senior patent attorney remains unchanged or even increases. The firm then loses billable junior hours without necessarily freeing senior practitioners, whose availability may be the factor limiting how many matters the firm can handle.
The economic question is therefore not merely how many hours AI saves, but whose hours are saved and how the released time can be used. Saving time for an employee who is already underutilised is not equivalent to freeing the time of a senior partner whose capacity is fully committed.
A firm assessing the viability of hourly billing in the AI era must therefore consider not only the percentage of time saved but also its effect on the volume of work it can accept, staff utilisation and overall profitability. Where AI creates capacity but the firm cannot attract more matters, hourly billing may turn the technology’s efficiency gain into a benefit for the client alone rather than one shared by the client and the firm.
Fixed Fees: Efficiency and Reasonableness
The difficulties AI creates for hourly billing may accelerate the move towards fixed fees for patent drafting and prosecution services. The attraction is not merely economic but also psychological. Clients are accustomed to paying certain amounts for patent applications and tend to assess whether a price is reasonable by reference to past experience and the service received. When a fixed price is agreed in advance, the fact that the firm completed the work in fewer hours does not necessarily generate an expectation of a corresponding fee reduction.
A fixed fee also generally removes the need to provide the client with a detailed account of hours worked. The firm can focus on the professional result and the value of the service rather than defending the time spent producing it.
Yet the appropriate fixed fee is particularly difficult to determine at present. AI capabilities are changing rapidly, adoption within firms is still evolving, and practical experience with these tools remains limited. It is not yet easy to assess reliably how much time is saved throughout the engagement, the quality of the outputs or the additional work needed to review and correct them. Even a firm that uses AI routinely may therefore struggle to establish the true cost of preparing a patent application under the new workflow.
In these circumstances, historical prices can provide a reasonable starting point for fixed fees. They are familiar to clients, reflect accumulated commercial experience and allow the firm to operate within an established pricing framework without committing to reductions based on savings that have not yet been properly measured. They also allow the firm to retain some of the economic benefit of its technological efficiency.
That advantage may, however, be temporary. As AI becomes a routine working tool, competitors will also become more efficient and clients will become familiar with the available alternatives. Historical prices may gradually lose their force as a reference point, and competition may increasingly concern which firm can provide the best professional service at the lowest price.
The move to fixed fees should therefore not rest on an assumption that current prices can be maintained indefinitely. During the transition, existing prices can serve as a starting point while the firm measures actual costs, including review and correction. Over time, prices can then reflect the new cost of the work and competitive conditions, rather than habits formed before AI became part of patent practice.
Phase-Based Fees, Retainers and Volume Discounts
AI may accelerate a significant change in how patent services are priced: a move away from separate charges for each professional task towards broader pricing arrangements that cover several stages of patent prosecution or even the management of a client’s patent portfolio.
One advantage of comprehensive pricing is that it changes the firm’s economic incentives. When drafting the application and responding to each office action are billed separately, every additional examination round generates more work and revenue. When a fixed fee also covers prosecution, the firm instead has an economic incentive to invest effort at the drafting stage in ways that may reduce later amendments and responses. AI can help with prior-art searching, testing alternative claim formulations and identifying potential examination difficulties at an earlier stage.
Such models are already being offered. For example, Solomon Patents advertises a fixed-fee model covering application drafting and responses to office actions through grant. However, pricing focused solely on obtaining a patent may create an incentive to compromise on claim scope to secure a quicker or less costly grant. The measure of success should therefore also account for the quality of the protection obtained in light of the client’s commercial objectives, not merely the fact that a patent was granted.
Retainer arrangements may also evolve. As AI makes it easier to identify potentially patentable inventions, analyse existing portfolios and recommend how they should be managed, ongoing services may shift from handling individual legal tasks to continuous patent portfolio management. Early examples of this approach already exist. Fearn, for example, offers an annual arrangement combining a professional team, a technology platform for portfolio management and fixed fees for filing work. In such a model, part of what the client buys is the capacity to make ongoing decisions about its portfolio, rather than simply a service for each separate application.
The rationale for volume discounts may change as well. Traditionally, a client sending a large number of patent applications to a firm could expect a discount based on the volume of work. AI, however, can help clients not only reduce the cost of preparing applications but also assess more effectively which inventions warrant filing in the first place. At the October 2026 AIPLA meeting, Meta’s former head of patents described how the company had used AI to reduce both its patent expenditure and the size of its portfolio. According to that account, the firms that performed better were those offering creative approaches to outcome-based pricing.
For patent firms, this points to a potential change in the business model. A client filing fewer applications but investing more in selecting inventions and protecting them may derive greater value from its portfolio even as filing work declines. A firm that depends mainly on revenue from individual filings and examination actions may suffer from that shift. A firm providing strategic patent portfolio management can continue to create value even when the number of applications falls.
The significant change AI may bring, therefore, is not simply a lower price for each task. It may be a move from fees based on the number of actions performed to fees reflecting the scope of professional responsibility, the desired outcome and the value of managing the patent portfolio as a whole.
Confidentiality Is Part of the Cost of Service
One assumption underlying expectations of lower patent fees is that AI will allow firms to work faster and at lower cost. Patent practice, however, presents a significant constraint on realising those savings: much of the work involves confidential technical information about inventions that have not yet been published.
Not every AI tool is suitable for processing that information. Firms need to examine the terms of service, data retention policies, whether information may be used to train the system and who may have access to it. The USPTO’s guidance on the use of AI also identifies risks associated with transferring technical information to servers in other countries.
These constraints have an economic cost. A firm wishing to use AI while protecting confidentiality may need to purchase services with appropriate contractual and technical safeguards, operate systems in a controlled environment or invest in its own infrastructure. Those options may cost more than publicly available AI tools and may not always offer the same capabilities.
This creates another pricing issue. Two patent applications of similar technical complexity may have different delivery costs if one can be handled using external AI tools while confidentiality or information-security restrictions prevent that approach for the other. A client requiring its information to be processed in a dedicated, isolated environment may also impose costs that do not arise in other client engagements.
When assessing the savings expected from AI, a firm must therefore consider both the cost of protecting information and restrictions on the technology it can use. This does not mean charging separately for every security measure. Where an engagement requires special safeguards, however, their additional cost may legitimately inform the fee agreed in advance.
Competitive differences between firms may consequently depend not only on the quality of the AI tools they use but also on their ability to operate advanced tools securely and at a reasonable cost. A firm that achieves this may gain a twofold advantage: lower working costs and the ability to handle sensitive technical information without compromising confidentiality.
Professional Responsibility as a Component of the Fee
One of the fundamental differences between legal services delivered through AI and those provided by a lawyer or patent attorney is professional responsibility for the result. An AI tool may produce a legal document, but it does not bear the same professional responsibility to the client as a practitioner providing the service.
When a client retains a patent firm, it is not merely buying the work of preparing an application. It is also relying on the firm’s experience, reputation and professional judgment, and on the fact that an identifiable professional stands behind the result. If professional negligence causes harm, the service provider’s liability may be examined and compensation pursued under applicable law. Professional indemnity insurance may also provide coverage in appropriate cases, subject to the policy terms.
This is particularly important in patent practice. An error in claim drafting, the omission of an essential feature from the disclosure or a missed deadline can damage rights with substantial economic value, sometimes irreversibly. For the client, therefore, the value of professional services extends beyond the quality of the document at the time of delivery. It also includes the ability to rely on the person who prepared it and the responsibility that person bears if something goes wrong.
This distinction has a direct effect on pricing in the AI era. Even if producing a patent application draft eventually takes only minutes, it does not follow that the service provided by a patent firm should be priced according to those minutes. Part of the fee reflects professional verification, reputation, professional indemnity insurance costs where applicable, and the legal and financial risk assumed by the firm.
A distinction may therefore develop between the price of producing the document and the price of a professional service that includes responsibility for the result. As AI reduces the cost of producing drafts, technical preparation may account for a smaller share of the fee, while the relative importance of review, judgment and professional responsibility may increase.
This does not mean that professional responsibility justifies any price a firm chooses to charge. It does explain why a reduction in working time does not necessarily imply an equivalent reduction in the service’s value to the client. A client paying a patent firm is purchasing not only a patent application, but also work from a professional whose judgment can be relied upon and who may be held accountable for negligence.
What Changes in the Budget for International Patent Filings?
AI may reduce the cost of obtaining patent protection in multiple countries, principally in two areas: translation of patent applications and the handling of examination proceedings.
The change in translation could be substantial. Instead of paying for a full professional translation, firms may use AI-assisted translation and pay primarily for reviewing and correcting the result. The more countries and languages in which an application is filed, the greater the potential cumulative savings.
AI may also reduce duplication in examination work. Where similar objections arise in different jurisdictions, the analysis and response prepared in one country can sometimes serve as the starting point for another. This may reduce foreign associates’ working hours while preserving the local review and adjustments required in each jurisdiction.
For pricing purposes, it is useful to distinguish work that can be performed once for an entire patent family from work that requires local handling. A firm that centralises the shared work and reduces duplication may be able to offer the client a more competitive international budget without compromising the quality of local representation.
Official fees and other expenses unrelated to the amount of professional work, however, will not decrease merely because AI is used.
Conclusion: Moving from Hourly Billing to Flexible Fixed Fees
AI is changing the economics of patent drafting and prosecution. It may substantially reduce working time and even improve the quality of the work, but this does not mean every hour saved should lead to a reduction in fees. The client is not buying working hours alone, but a professional service that includes knowledge, experience, strategy and responsibility for the result.
The central conclusion is that patent firms should accelerate the move from hourly billing to fixed fees. As AI shortens the time required to perform work, hourly billing becomes less suitable both for the firm and for how clients assess the service. Fixed fees break the direct link between hours worked and revenue, allowing the firm to price the service according to its scope and the value it provides.
That does not require a single price covering the entire life of a patent application. A fixed fee can be set for an individual task, such as responding to an office action; for a phase comprising several tasks, such as preparing and filing an application; or for a wider process that includes drafting and the subsequent examination proceedings. Each approach offers a different balance of pricing predictability, certainty for the client and financial risk for the firm.
The challenge is to identify the appropriate fixed price at each of these levels as quickly as possible. AI capabilities are improving rapidly, practitioners’ proficiency is developing and the real cost of carrying out the work is changing accordingly. Firms therefore need to track what each task costs, which tasks can sensibly be combined into one fee and which stages still warrant separate pricing.
Precisely during a period of technological uncertainty, fixed fees can offer commercial stability. The client knows the price in advance, and the firm need not reduce its fee every time a new tool lets it work faster. At the same time, the firm can revise its prices to reflect improved efficiency and competitive conditions. It can benefit from technological progress without tying its revenue to a shrinking number of billable hours.
During the transition, historical prices may provide a starting point, but they should not substitute for updated pricing. A firm that understands the new cost of each task and phase before its competitors can offer competitive prices where AI delivers significant savings while maintaining fees that reflect complexity, expertise and professional responsibility where those remain the core of the service.
The practical recommendation is therefore to move gradually, but as quickly as feasible, to a flexible fixed-fee system that can price an individual task, a group of tasks or an entire process. There is no need to wait until AI’s impact is fully understood. On the contrary, rapid change makes it necessary to develop the ability to reprice services now, update fees as experience accumulates and choose the appropriate level of pricing for each engagement.
The firms that gain a competitive advantage will not necessarily be those using the most advanced AI tools. They will be those that translate technological efficiency into a pricing model that gives clients cost certainty while allowing the firm to remain profitable even as working methods change.
Adv. Yonatan Drori is an Israeli intellectual property attorney at Drori-Stav IP and the author of Patents Law, a comprehensive two-volume legal treatise on Israeli patent law and practice. His practice focuses on patent prosecution, intellectual property litigation and representing foreign clients in Israel.
