By Guo Hong, President of ZYIP

Contents

1. Deep Understanding of the Client's Business and Commercial Goals
2. Proactively Identifying Hidden Risks and Uncovering Potential Problems
3. Professional Trade-off Judgment at Critical Junctures of a Case
4. Balancing Client Demands with Legal Compliance
5. Communication, Mediation and Coordination Among Multiple Stakeholders
6. From Search Reports to Strategic Intelligence Products
7. Fallback Handling and Post-Incident Recovery in Crisis Events
8. "Translating" IP Decisions into Business Language
9. Coordinated Strategy Judgment Across Jurisdictions
10. Client Trust Built Through Long-Term Service
    Closing Remarks

Over the past two years, the impact of AI on the IP industry has been real: patent searches can be generated automatically, invention disclosures can be polished with a single click, and the speed of preparing draft application documents as well as the efficiency of office action (OA) responses have improved markedly. This has given rise to a widespread anxiety across the industry: when AI can do everything, what is left for IP agencies?

In my view: AI has already proven its ability to improve efficiency, but it has yet to pass the test on quality. The deeper layer of work — involving business judgment, interpersonal negotiation and trust-based relationships — is still beyond AI's reach. The following ten capabilities constitute the core that agencies cannot yet be replaced by in the AI era. They are not isolated individual skills, but a professional service system built around "the client's commercial success."

1. Deep Understanding of the Client's Business and Commercial Goals

Low-level service means "understanding what the client asks for"; high-level capability means "anticipating the commercial intent the client has not voiced."

Case: An AI medical imaging startup comes to you to file a patent, saying "help me protect this algorithm." A shallow understanding is to write a high-quality patent application. A deep understanding means seeing that the company is in the middle of a funding window — that investors care about the quality of the patent portfolio, the strength of the technological barrier and whether the technology is implementable, not merely the number of filings; that the product still has to clear FDA approval or domestic NMPA registration; that the core of the algorithm is better protected as a trade secret than fully disclosed in a patent application; and that you also need to anticipate the IP due diligence requirements in future M&A scenarios.

What you deliver is not a single application, but a portfolio strategy that is "funding-friendly + compliance-safe + M&A-enhancing." This kind of intent recognition spanning the legal, technical and commercial dimensions cannot be replaced by AI executing instructions.

2. Proactively Identifying Hidden Risks and Uncovering Potential Problems

Low-level service means "answering whatever the client asks"; high-level capability means "seeing the risks the client has not asked about — before they do."

Case: A client brings you a technical solution for a patent filing, and you敏锐 notice that the core process of the solution originates from a supplier's technical secret, with the client merely performing an integrated application. Filing the patent directly would very likely trigger supply chain ownership disputes and leakage of the core process.

Professional approach: Give early warning, adjust the boundary of technical disclosure and omit key parameters; prioritize pushing the client to sign a sound IP ownership and confidentiality agreement with the supplier before initiating the patent application.

Risk preemption is the most intuitive manifestation of professional value. The client may not pay for it separately, but one successful act of risk preemption is enough for the client to entrust you with more core and more sensitive business — and trust and reliance deepen as a result.

3. Professional Trade-off Judgment at Critical Junctures of a Case

Low-level service means "presenting options"; high-level capability means "giving a recommendation with a clear lean, and bearing professional responsibility for it."

Case: An office action points out a lack of inventiveness. There are usually three paths:

1. Narrow the scope of protection for a fast grant — quick and satisfying for the client, but the scope of protection is narrowed;
2. Argue vigorously on inventiveness — may be granted or may be rejected, with a long cycle;
3. File a divisional application, preserving the broad parent application — increases cost, but leaves a fallback.

The basis for the decision is not the legal provisions themselves, but the client's commercial situation. If the client is in fundraising negotiations and urgently needs a granted certificate to support valuation, recommend the "① + ③" combination: secure the granted certificate first and keep the broad-scope fallback in the divisional. If the product has not yet launched, the budget is sufficient and competition is intense, recommend the "② + ③" combination: argue the main case fully and fight for a broad-scope grant, because in a highly competitive赛道, every inch of broader scope means stronger suppression of competitors; file the divisional simultaneously as a fallback, so that even if the main case is ultimately rejected or forced to narrow substantially, the core solution still has a chance to be re-fought.

You do not place three options in front of the client and let them choose — the client does not understand patent law and cannot make an effective choice. Instead, you clearly say "I recommend X, because Y," and take responsibility for your professional judgment.

The willingness to give a directional opinion and bear the consequences is precisely the essential difference between a "patent attorney" and a "tool."

4. Balancing Client Demands with Legal Compliance

Low-level service means "giving the client whatever they ask for"; high-level capability means "daring to say no, and offering an alternative."

Case: The client asks to name an executive who clearly did not participate in the invention as an inventor. You should clearly inform them: falsely naming an inventor constitutes dishonest conduct. Although it does not necessarily invalidate the patent right, it may raise serious questions in ownership disputes and in financing and IPO due diligence, and in certain circumstances in some jurisdictions may even affect the validity of the patent.

At the same time, offer an implementable alternative: use legitimate incentives such as project bonuses and equity in place of naming, so that "attribution goes to attribution, and incentives go to incentives." The actual inventor enjoys the right of attribution and the employee invention reward as provided by law, while the executive is rewarded through project bonuses and equity incentives. This achieves the incentive purpose without crossing the legal red line.

Saying "no" is not rejecting the client — it is using professionalism to protect the client's long-term interests.

5. Communication, Mediation and Coordination Among Multiple Stakeholders

Low-level service means "acting as a messenger of information"; high-level capability means "becoming a translator between technology and law, a buffer for conflict, and a negotiator of interests."

Case 1: The technical inventor feels the attorney "changed his solution" and refuses to cooperate in reviewing and finalizing the draft. You need to mediate between the inventor's technical pride and the legal rigor of the claims — explaining, in language the inventor understands, why the claims must be drafted this way, while making him feel respected rather than overridden.

Case 2: The client's legal department worries about technical leakage, while the R&D department wants to showcase technical strength, and the two sides are deadlocked over "whether to disclose a certain technical point." You can organize a three-way meeting, cite the consequences of disclosure in similar cases, quantify the risk levels of different disclosure boundaries, and drive consensus.

IP service has never been a simple one-to-one engagement; it is the art of coordinating among multiple interest stakeholders embedded within the client's organization.

6. From Search Reports to Strategic Intelligence Products

Low-level service means "delivering a search report"; high-level capability means "delivering a strategic intelligence product."

Case: You serve a solid-state battery client. You should not stop at helping them file patents — you should also regularly provide a "Solid-State Battery IP Landscape Briefing": who is doubling down on the sulfide route, who is continuing to invest in the oxide route, which companies have been aggressively building positions over the past six months, which patents may become SEPs (Standard Essential Patents) in the future, and which targets are worth tracking for acquisition. When the client says "this report is more useful than the market research we paid for ourselves," you have already walked out of the red ocean of price wars.

Search is the means; intelligence insight is the product. AI can generate data lists, but "what these signals mean for the client" requires deep industry judgment.

7. Fallback Handling and Post-Incident Recovery in Crisis Events

Low-level service means "rescue after the incident happens"; high-level capability means "having a plan + being able to provide a fallback + post-incident recovery."

Case: The client's core trademark has been squatted overseas, and the other party has already started using it to sell similar products. The professional response runs on four parallel tracks:

① Assess the feasibility of invalidation or opposition;
② Simultaneously initiate buyback negotiations to seize the initiative in pricing;
③ Design an interim alternative brand solution to ensure business continuity;
④ Prepare non-infringement defense materials to respond to possible litigation.

Ultimately the trademark is bought back at a reasonable price, with zero disruption to the client's product line. The commercial value recovered for the client by one successful crisis handling far exceeds the magnitude of routine agency fees.

8. "Translating" IP Decisions into Business Language

Low-level service means "outputting professional jargon"; high-level capability means "making decision-makers understand and willing to pay."

Why it matters: The client's decision-makers — CEO, CFO, investors — do not know patent law provisions, but they do understand input-output, risk exposure and time windows. Whether you can explain, in language they understand, "why this is worth spending the money on" directly determines whether you get the budget.

The contrast:

Professional language: "This claim contains too many non-essential technical features and needs to be trimmed."

Business language: "If we keep these non-essential technical features, competitors can easily drop that feature and design around our scope of protection — it would be like spending RMB 100,000 to obtain a patent that is easy to circumvent. After trimming and optimization, the scope of protection can cover the three technical routes competitors are most likely to implement."

The last mile of professional capability is the ability to express it. Whether the same professional opinion turns into action by decision-makers depends on whether you can complete the translation from legal language into business language.

9. Coordinated Strategy Judgment Across Jurisdictions

Low-level service means "filing country by country"; high-level capability means "a combined-punch layout based on commercial rhythm."

Why it matters: Going global is now a necessity, but examination practices and legal environments differ significantly across countries. AI can supply the legal provisions of each country, but a combined approach such as "securing a position in China through expedited pre-examination, keeping a fallback in the US through continuation applications, and covering different embodiments in Europe through divisional applications" requires an orchestrated strategy grounded in commercial rhythm.

Take the global launch of a drone product as an example. A professional three-stage layout can help the client, before launch, secure Chinese patent rights as the foundation for enforcement and qualification; build a portfolio of continuation applications in the US that continuously suppresses competitors; and form a refined patent network in Europe that is difficult to design around. Such strategies are not automatically generated template solutions, but the product of comprehensive analysis of national systems, industry competitive dynamics and the client's commercial rhythm.

10. Client Trust Built Through Long-Term Service

Low-level service means "deliver the certificate and leave"; high-level capability means "becoming an extension of the client's IP department."

Case: The client has changed IP heads three times. At each handover, the agency proactively provides a panoramic view of historical cases, the status and to-do list of pending matters, and annotations of risk points, so that the new head can take over within the first week. When the new head says "you know these historical situations better than we do ourselves," the agency has already built a barrier that no AI tool can replicate. Trust is not formed in a single delivery; it is accumulated over a long service cycle.

Patent prosecution is a special profession: the client entrusts you with their most core secrets before the technology is disclosed. The essence of this entrustment is trust, and trust is built on long-term, non-utilitarian professional companionship — proactively flagging risks before the client has thought of them; giving honest priority advice when the budget is tight, telling them what must be done and what can wait rather than urging them to file more; and serving as that "de facto IP advisor" when no one inside the client organization understands IP.

AI has no position to take and no reputation to maintain. What the client needs is not a tool that is available on call, but a person who "can be turned to when something goes wrong, tells the truth, and has my interests at heart."

Closing Remarks

Why are these capabilities currently irreplaceable by AI? Looking back at these ten capabilities, they share one thing in common: all of them rely on "a deep understanding of the client's specific situation" and "a subject who bears responsibility for the judgment" — and these are precisely AI's two structural shortcomings today.

Technical tools are amplifiers, but an amplifier only amplifies the signal that already exists at the front end. If the front end only carries the signal of "writing documents," AI merely amplifies assembly-line document production. If the front end consists of business judgment, interpersonal negotiation and trust relationships, AI will amplify your professional barriers and service value.

But we must see this clearly as well: these capabilities are only an interim moat. At the pace AI is developing today, work that currently belongs to high-level human professional judgment may gradually be realized by AI in the future.

So what we need is not reliance on this capability list, but continuous self-reinvention. The true meaning of a moat is not "AI can never get in," but that while others are being replaced by tools, you have already moved to a floor that the tools cannot reach.

In the AI era, tools are responsible for getting things "done." The mission of a professional firm is to get things "done right" — to deliver IP solutions that serve the client's long-term commercial interests.