What separates a patent agent who is asked for by name
$195,750estimated top of the range · middle $95,000 / yr
AI is transforming this role
Patent Agents in the United States earn a median of $95,000 a year. Pay starts near $60,000. The top of the range is estimated at $195,750. The Bureau of Labor Statistics does not publish a separate wage series for this exact title, so this figure is derived from the closest occupation it does track and is labelled an estimate.
Source: PayCrunch estimate. Last checked 9 September 2026.
Entry level
$60,000
Top-end estimate
$195,750
Education
Bachelor's degree in Science or Engineering
Wages — PayCrunch estimate. The Bureau of Labor Statistics does not publish a separate wage series for Patent Agent; figures are derived from the closest occupation it does track and are labelled as estimates. AI-impact rating is PayCrunch's editorial assessment. Updated September 2026.
🆕 New & Trending AI Tools for Patent AgentReviewed September 2026
We track new AI-tool launches every week and refresh this list — here’s what’s gaining traction for Patent Agent work right now.
ChatGPT Gov / EnterpriseNEWEnterprise / see site
Secured version of ChatGPT approved for public-sector and enterprise use.
How a Patent Agent uses it: draft, summarize, and research inside an approved, secured environment
Microsoft Copilot for GovernmentNEWGov cloud / see site
Copilot AI inside the government (GCC) versions of Word, Excel, Outlook and Teams.
How a Patent Agent uses it: write documents, build spreadsheets, and summarize meetings in a compliant setup
Google Gemini for GovernmentNEWGov cloud / see site
Google's AI assistant in the public-sector version of Workspace.
How a Patent Agent uses it: draft and research inside a FedRAMP-authorized Google environment
NotebookLMNEWFree / $7.99 mo
Google tool that answers questions grounded only in the documents you give it — with citations.
How a Patent Agent uses it: load your own manuals, policies, or PDFs and ask questions that stay accurate to the source
MoveworksEnterprise / see site
AI assistant that handles employee IT, HR, and operations requests (FedRAMP authorized).
How a Patent Agent uses it: get IT/HR answers and routine requests handled by chat instead of tickets
ChatGPTFree / $20 mo
The most-used AI assistant — writing, analysis, research, and images from a plain-language chat.
How a Patent Agent uses it: draft emails and documents, summarize long files, and get instant answers to on-the-job questions
ClaudeFree / $20 mo
AI assistant known for careful writing, long-document analysis, and coding.
How a Patent Agent uses it: analyze big reports or spreadsheets and turn messy notes into clean, finished writing
Google GeminiFree / $20 mo
Google's AI assistant, built into Gmail, Docs, and Search.
How a Patent Agent uses it: draft and reply inside Google Workspace and research without leaving the page
Microsoft CopilotFree / $30 mo
AI built into Word, Excel, PowerPoint, Outlook, and Teams.
How a Patent Agent uses it: write documents, build spreadsheets, and summarize meetings inside Office
Papers that have to stand on their own
A patent agent drafts patent papers and files them with the patent office. The invention usually starts as a scientist's or engineer's explanation, sometimes clear, often not. The agent turns that explanation into an application a stranger can follow: what the invention is, how it fits together, and which parts the client wants protected. Drawings get coordinated with the text. Deadlines get watched. When the office writes back, the agent prepares a response the office can read without a meeting. The work is technical writing with legal consequences, done by someone who is not practicing as a lawyer.
Agents work in law firms, in companies, and sometimes in solo practices once they have a reputation. In a firm, the agent may sit with lawyers and do the technical drafting while a lawyer reviews strategy the licence reserves for lawyers. In a company, the agent may sit with engineers, harvest inventions from lab notebooks and product meetings, and send applications out under the company's name. The daily tools are a docketing system, a prior-art search the agent can explain, and a habit of asking the inventor what the product actually does when the slide deck has already moved on.
The boundary with lawyers matters on the first day. An agent may represent applicants before the patent office in patent matters. An agent does not hold a law licence, does not go to court as counsel, and does not give the broader legal advice a client might want about contracts, ownership fights, or a lawsuit. Good agents say that boundary out loud. Clients who wanted a lawyer and hired an agent by mistake create risk for everyone. Clients who wanted a technically fluent drafter, and got one, get a cleaner application.
A week has a rhythm the docketing software makes visible. New disclosures arrive from inventors who want a filing before a product announcement. Office letters arrive with a response date already on the calendar. Foreign associates send formal papers that need a technical read before anyone decides on a filing abroad. The agent blocks time for deep drafting, because an application written in stolen scraps of an afternoon tends to wander. The agent also blocks time for inventor calls while the inventor still remembers the prototype. Waiting until the end of the month to ask what a drawing got wrong is how small errors become expensive amendments. None of this is courtroom work. It is calendar discipline plus technical listening, and firms notice both.
Searching belongs to that same honesty. Before a client spends heavily on a filing, the agent looks for earlier publications that sit close to the idea and explains, in plain language, what those publications appear to show. The agent does not promise an outcome. The Office will do its own review later. What the client needs at the start is a clear picture of the landscape and a draft that does not pretend the landscape is empty. Overclaiming to please a hopeful inventor is a short-term kindness and a long-term mess. Underclaiming out of fear can give away the feature the product depends on. The agent's job is to write language the invention can support and to say, in the cover note, where the uncertainty sits.
The technical degree that comes first
Registration starts with technical education. The United States Patent and Trademark Office decides which backgrounds qualify. Engineering, computer science, chemistry, biology, physics, and other listed fields are the usual routes. A degree title that sounds technical is not a guarantee. The Office publishes the list, and the list changes at the edges. Before you spend a year preparing, match your transcript to the current list or ask the Office how a mixed degree will be read. Guessing is how strong engineers discover, late, that a missing course blocks registration.
The degree is what makes the drafting honest. An agent with a mechanical background can sit with a gearbox inventor and notice when the description skips a part. An agent with a chemistry background can tell when an example is too thin to support the language the client wants. You do not need to have invented the thing yourself. You need enough fluency to ask the inventor a precise follow-up and to refuse a sentence you do not understand. Applications full of words the agent could not define are how offices, and later opponents, find trouble.
Work experience can deepen that fluency. A few years in a lab, a design group, or a software team teaches you how inventors talk and which details they forget. Some people go straight from a degree into a firm training program. Both paths can work. The firm path teaches docket hygiene early. The industry path teaches what a product deadline feels like. Name whichever path you have, without dressing a student project up as a career.
Registration with the Office, without a scoreboard
After the technical qualification, registration with the United States Patent and Trademark Office is the credential that lets you practice before the Office in patent matters. The Office runs that registration. It confirms your scientific or technical background, your moral character through the application it requires, and your grasp of patent practice and procedure. People prepare by studying the practice materials the Office points to, often while working as a technical advisor under a registered agent or attorney. This description stops there. It does not recount a test or a timetable.
What registration proves is narrow and valuable. It proves the Office is willing to let you represent applicants in patent matters before it. It does not prove you are a lawyer. It does not prove you know every technology. It proves you met the Office's requirements to be on its register. Keep the registration current the way the Office requires, including the practical-training obligations that come with staying on the register. An expired registration is not a cosmetic problem. It is the difference between doing the job and being unable to sign the papers.
On a resume, write "registered to practice before the United States Patent and Trademark Office" and the registration number if you have one. Do not decorate it with jargon from a prep course. Hiring partners know the credential. They also know the difference between a new registrant and an agent who has carried a docket through final disposition. The credential opens the door. The docket is the career.
How firms and companies hire
Firms hire agents when their patent group needs technical drafting capacity and does not need another courtroom lawyer. A posting will name a technology: electrical, software, mechanical, chemical, biotech. Apply where your degree is real. A chemistry agent forcing a software docket, or the reverse, makes both the inventor and the partner miserable. The interview is a conversation about how you learn an invention, how you write, and how you handle a deadline you can see coming. Some firms ask for a writing sample with confidential details removed. Bring one that shows clear sentences and careful structure, not a thesaurus.
Companies hire agents into intellectual-property groups beside engineers. The interview there often includes a hiring manager from the lab as well as from the legal department. They want someone who can sit in a disclosure meeting without talking down to inventors, and someone who will tell a business leader when a filing is premature. Ask who reviews your work, whether lawyers are on staff, and how conflicts are handled if the company wants a filing the technical story does not support. That last point is where agents earn trust or lose it.
Early titles vary: technical advisor, patent engineer, scientific advisor, patent agent. Before registration, the first of those titles is common, and the work is supervised. After registration, the signature authority changes. Ask when the employer expects registration, whether they support study time, and what the pay step is when the Office adds you to the register. Get that step in writing. A verbal "we will take care of you after you register" is hard to collect a year later.
PayCrunch estimates, kept away from lawyer wages
Because the Bureau of Labor Statistics does not publish a separate wage series for this exact title, the figures here are PayCrunch estimates. They are not lawyer wages, and they are not attached to any state. Estimated entry pay is $60,000. The estimated median is $95,000. The estimated top is $195,750. The gap from entry to the median is $35,000. The gap from the median to the top is $100,750. Use those five numbers as a national sketch of the agent role, then put a real offer beside them.
Do not borrow a lawyer table
An agent and a lawyer can work on the same application and still be in different occupations. Lawyer wages describe a law licence and a different market. Quoting them as if they were agent pay will not persuade a hiring partner who knows the difference. Stay with $60,000, $95,000, and $195,750, and say they are estimates.
Entry-level estimate of $60,000 fits a newly registered agent or a technical advisor still building a first docket. The median of $95,000 fits someone who drafts independently, manages deadlines, and can talk an inventor through a disclosure without a senior person in the room. The top estimate of $195,750 is $100,750 above that median. It belongs to the far end of the title: deep technical specialty, a book of clients, or a senior in-house role with responsibility for other drafters. It is an estimate of the top, not a typical offer, and not a lawyer's pay relabeled.
A path that can stay technical
Many agents never go to law school, and that can be a complete career. They become the person a firm trusts with a technology, then the person who trains newer advisors, then a senior agent whose name clients request. In-house, the path can run from drafting into portfolio coordination: which filings are worth the budget, which countries the business actually sells in, which inventions should stay as trade secrets instead. Those judgements still sit beside lawyers when the question is a legal one. The agent's value is knowing the technology well enough to make the filing list honest.
Some agents do attend law school later, sit for admission to a state bar, and move into attorney roles. That is a new profession with a new credential, not a raise inside the agent title. If you want that route, ask the employer whether they hire their own agents into the lawyer track and what happens to your docket during school. If you do not want that route, say so. Firms sometimes assume every agent is a lawyer in waiting. Correcting the assumption early prevents a stalled promotion built on a degree you never planned to earn.
Keep a record that respects client confidentiality. Note technologies, counts of applications you drafted or responded on, and whether you worked alone or under review. You can describe a mechanical filing practice without describing the client's invention. You can say you carried matters through to allowance without narrating the arguments. The career grows by demonstrated judgement. It does not grow by collecting war stories you are not free to tell.
A pay conversation that stays inside the estimates
If you are registered and still near $60,000, the median of $95,000 is the figure to put on the table, and the $35,000 gap is the size of that step in the estimate. Tie it to work a partner can see: you draft without a full rewrite, your deadlines do not surprise the docketing team, inventors answer you because you ask useful follow-ups, and a technology group has started requesting you. Ask whether registration, a specialty, or a second year of independent drafting has a scheduled increase.
If you are already near $95,000, talk about responsibility rather than about the top figure as a mood. The estimated top of $195,750 sits $100,750 above the median. That distance is real in the estimate and large. It usually means a scarce technical specialty, supervision of other drafters, or a client relationship you originated. If those describe your job, say so. If they do not, aiming the meeting at $195,750 will sound like you imported a top figure from a different occupation. You brought an estimate for this title, and you should label it as one.
Leave lawyer wage charts out of the folder. Say, if you need the sentence, that the Bureau of Labor Statistics does not publish a separate wage series for patent agents and that these PayCrunch estimates are the figures you are using. Then stop. A calm agent who knows the boundary of the role is the same person a firm trusts with a filing. Be that person in the pay meeting too.
The top of Patent Agent pay — and how to get there with AI
$195,750top-end estimate for Patent Agent
PayCrunch estimate - derived from the closest occupation BLS tracks. This figure is PayCrunch’s estimate, not a Bureau of Labor Statistics published wage for this exact title.
$60,000entry$95,000middle$195,750top end
A patent agent reaches the top of this range by drafting claims that hold their scope through prosecution and by working in a technology area where very few registered practitioners are qualified.
Prosecution work is billed against a docket, so speed matters, but the money follows judgement. Anyone can respond to a rejection by narrowing a claim until it is allowed; the agent clients want is the one who argues the reference does not teach what the examiner says it teaches, who calls the examiner and gets the case moving, and who drafted the original specification with enough support that amendments are available years later. Prior art reading, claim charting and first-pass office action summaries are the slow parts, and a model can compress them, as long as you read the reference yourself before you rely on a summary of it.
Your playbook, by where you are now
Just startingRegistration first, then drafting
Sit the registration examination early, because until you are registered your technical degree is potential rather than qualification.
Draft constantly under someone who will mark you hard, since claim drafting is a craft and feedback is the only accelerator.
Learn the rejection types cold, meaning anticipation, obviousness and the support and definiteness requirements, and how each is really argued.
Do your own prior art searching for a year before you outsource it, because searching teaches you what claims are vulnerable to.
Learn the docket, since a missed deadline is the one mistake that ends careers in this work.
What proves it: Registration plus a portfolio of applications you drafted that issued with useful scope.
Realistic span: your first two or three years
A few years inProsecute like it is a strategy, not a form
Pick up the phone and conduct examiner interviews yourself, because cases move in conversation far more than in written argument.
Learn continuation practice properly so a client's portfolio keeps options open rather than closing on the first allowance.
Specialise in a technical area with a shortage of qualified practitioners, whether that is biologics, semiconductors or wireless standards.
Have Claude summarise long references and build a first-draft claim chart, then verify every citation against the document itself.
Start managing client relationships directly, since agents who own the relationship are treated differently from agents who are assigned work.
What proves it: Client portfolios you prosecute end to end, with interview practice you run yourself.
Realistic span: years four through seven
ExperiencedDecide about the law degree deliberately
Weigh the law degree honestly; it opens opinion work, litigation support and partnership tracks that are closed to an agent.
If you stay an agent, become the drafting authority in a technology the firm cannot cover otherwise, and price accordingly.
Consider the move in-house, where a senior agent manages outside counsel and an entire portfolio strategy.
Build a practice around portfolio counselling, meaning what to file, what to abandon and where to file abroad, rather than per-application work.
Train junior drafters and review their claims, because a firm keeps the person who makes everyone else's work billable.
What proves it: A portfolio you set the strategy for, or admission that opens the attorney track.
Realistic span: from year eight
The next 90 days
Take three applications you drafted more than two years ago and read their file histories from start to finish. Look at what happened to every independent claim. Where did scope get given away, and was it because the specification did not support a better fallback? Where did an argument fail, and would a different characterisation of the reference have worked? Write yourself two pages on what you would draft differently now. Almost no patent agent does this, because the docket always points forward. It is the fastest way to improve drafting, and it gives you concrete things to say when a client asks why they should send work to you rather than to the firm down the road.
Wage figures: PayCrunch estimate. The playbook is PayCrunch editorial guidance, not a guarantee of pay or placement.
Every figure is the national median from the U.S. Bureau of Labor Statistics (OEWS) shown on that role’s own page.
Never used AI before? Start here (2 minutes).
Start with a purpose-built patent-drafting AI — PatentPal or Solve Intelligence — inside a confidential or enterprise account. Feed it your claim set and let it generate a first-draft specification, figure descriptions, and consistent reference numerals; you then edit for accuracy, scope, and support. This is the single biggest time sink in drafting, and compressing it is what lets you take on more applications.
For prior art, use IPRally or Patlytics for AI semantic search and Lens.org or Google Patents for free searching, and Juristat to study your examiner before you respond. Reserve general tools like Claude for non-confidential learning and structuring. AI is the associate who drafts and searches; you are the registered agent who signs and owes candor to the Office.
The one rule, forever: Client confidentiality and the USPTO duty of candor are absolute. Never paste unpublished invention disclosures, draft claims, or search results into a consumer AI tool without an enterprise or zero-retention agreement — a leak can destroy patentability. Verify every AI-provided citation and technical statement: filing hallucinated prior art, a fabricated reference, or a false statement to the USPTO is a fraud on the Office that can invalidate the patent and end your registration.
The plays — exact steps, exact prompts
Do these in order. Each one is copy-paste ready. You do not need to know anything about AI going in.
1
Draft specifications and figures from claims in a fraction of the time
Why this pays: Drafting the spec and figures is the slowest, most repetitive part of a patent application. AI that generates a supported first draft from your claims roughly multiplies the applications you can file per month — and for a per-application or hourly biller, throughput is income.
PatentPalSolve IntelligenceDeepIP
1
In a confidential account, use PatentPal or Solve Intelligence to generate a first-draft specification, figure descriptions, and reference numerals from your claim set.
2
Have AI draft support for every claim element, then edit.
Copy-paste this prompt
You are a patent-drafting assistant working under a confidentiality agreement. Given this independent claim [paste claim] and these figures described here [describe], draft a detailed description that provides written-description and enablement support for every claim element, using consistent reference numerals and antecedent basis. Flag any claim term that lacks clear support.
Use only in an enterprise or zero-retention account. Verify all technical accuracy and support yourself; you sign the application.
What you'll haveMore applications drafted per month at consistent quality — the throughput that drives billings toward the top of the range.
2
Run AI prior-art searches that don't miss
Why this pays: A missed reference weakens claims and burns client trust; a thorough search lets you draft around the art and secure stronger, more valuable patents. AI semantic search covers more ground than keyword search alone — better outcomes that win premium clients.
IPRallyPatlyticsLens.org
1
Use IPRally for AI semantic prior-art search and Patlytics for analysis, cross-checking on the free Lens.org and Google Patents databases.
2
Build a rigorous search strategy first.
Copy-paste this prompt
Act as a prior-art search strategist. My invention is [describe the inventive concept generically, no confidential specifics]. Generate a search strategy: the key technical concepts and their synonyms, relevant CPC classification codes, three Boolean query variations, and the adjacent fields where analogous art may hide. Then explain what a strong 102 reference versus a 103 combination would each need to show.
Keep confidential specifics out of the prompt. Confirm every reference in the actual database — AI can miss art and can hallucinate references.
What you'll haveMore complete prior-art coverage and stronger claims — better patents and the reputation that attracts premium clients.
3
Respond to office actions faster and smarter
Why this pays: Prosecution — responding to rejections — is where applications either die or issue, and where a lot of billable time goes. AI that structures arguments and amendments lets you turn responses faster and lift your allowance rate, both of which raise your value per case.
JuristatPatlyticsClaude
1
Study the examiner's history in Juristat, then use Claude to structure the response and candidate arguments — verifying every reference yourself.
2
Outline the response before you write it.
Copy-paste this prompt
You are helping me structure a response to a USPTO office action. The examiner rejected claims [1-5] under [35 U.S.C. 103] over [Reference A in view of Reference B], arguing [summarize the rejection]. Outline a response: the strongest non-amendment arguments (teaching away, no motivation to combine, missing element), which claim amendments would create clear distinctions, and the remarks structure.
Verify every citation and what each reference actually discloses. No confidential client data beyond what is public in the file.
What you'll haveFaster responses and a higher allowance rate — more value delivered per case and per billable hour.
4
Draft stronger, broader claims with AI review
Why this pays: Claim scope is where a patent's value is made or lost. AI that critiques breadth and Section 112 issues and proposes fallbacks helps you secure broader, more defensible claims — the difference between a routine filing and one a client will pay a premium for.
Solve IntelligenceClaude
1
Use Solve Intelligence or Claude to stress-test a claim for breadth, clarity, and support, then decide scope yourself.
2
Get a critique and alternatives.
Copy-paste this prompt
Act as a claims-drafting reviewer. Here is my independent claim [paste]. Critique it for breadth, clarity, antecedent basis, and Section 112 issues; suggest a broader alternative and a narrower fallback; and propose three dependent claims that add patentable weight (structural, functional, and a design-around-blocking limitation). Explain the trade-off of each.
Use a confidential account. You make the final scope decisions based on the client's business, not the tool's.
What you'll haveBroader, better-supported claims that survive prosecution and litigation — the quality that commands premium rates.
5
Analyze examiners and art units to set strategy
Why this pays: Knowing an examiner's allowance rate and tendencies tells you when to interview, amend, or appeal — decisions that determine how fast and how cheaply a case issues. Delivering faster allowances is concrete client value that supports higher fees and retention.
JuristatPatSnap
1
Pull examiner and art-unit analytics from Juristat and portfolio context from PatSnap to shape your prosecution strategy.
2
Translate the stats into a plan.
Copy-paste this prompt
You are a prosecution strategist. My application is assigned to examiner [name or art unit — public information]. Based on typical examiner-analytics metrics (allowance rate, average number of office actions, interview outcomes), explain what strategy tends to work with a low-allowance-rate examiner: when to request an interview, whether to appeal, and how to frame amendments.
General strategy. Pull the actual statistics from Juristat or PatentAdvisor and confirm the case file before acting.
What you'll haveData-driven prosecution that issues cases faster and cheaper — measurable value that keeps clients and justifies fees.
6
Build a client book in a hot art unit
Why this pays: Top patent-agent pay follows expertise plus a book of business. Specializing in a high-demand technology and using AI capacity to market and deliver lets you attract and keep clients directly — the ownership that moves you into and past the top of the range.
PatSnapPerplexityClaude
1
Use PatSnap and Perplexity to spot active filing areas and prospective clients, and Claude to draft outreach and thought-leadership content.
2
Build a focused business-development plan.
Copy-paste this prompt
Act as a business-development coach for a patent agent. I have a background in [electrical engineering] and want to build a book in [semiconductor and AI-hardware] patents. Draft a 90-day plan: the niche expertise to deepen, where these clients are (startups, in-house counsel), a short outreach message, and one piece of thought-leadership content that would attract them.
Concrete and ethical outreach only. Never disclose one client's confidential information to win another.
What you'll haveA specialized reputation and a growing client book — the ownership that defines top-of-range patent-agent earnings.
Your 12-month sequence to the top of the range
How the plays above stack into a path from median pay toward the $145,000 tier.
Month 1
Move spec and figure drafting into a confidential patent-AI tool; measure the time saved per application.
Months 2-3
Add AI prior-art search and examiner analytics to your workflow; verify every reference against the databases.
Months 3-6
Systematize office-action responses and AI claim review to lift quality and allowance rate.
Months 6-12
Specialize in a hot art unit and use the extra capacity to build a client book — the path to the top of the range.
Gear for this job
As an Amazon Associate, PayCrunch earns from qualifying purchases. Links to books and tools are for the job on this page; we only recommend what we’d use in the work.
Independent 2026 USPTO / patent-bar paperback (ASIN B0H2YFRHL6, ISBN 979-8198619623) for leftover USPTO registration / OED this page names — Just starting is Sit the registration examination early; sources link USPTO — Becoming a Patent Practitioner (registration and OED). Independently published May 2026 study guide, not the official MPEP. Confirm B0H2YFRHL6. Not leftover Series 7 and not leftover 94 CFP. HTTP 200, add-to-cart / buy-now on /dp/B0H2YFRHL6.
What Patent Agents earn by state
This page does not show a state table, and the reason is worth stating: the Bureau of Labor Statistics does not publish a separate wage series for this job title, so there are no official state figures to show. Scaling the national median by a cost-of-living index would produce a number for every state, but it would be an estimate of living costs wearing a wage’s clothes, and PayCrunch would rather show you nothing than that.
What the national figures say: pay starts near $60,000, the median is $95,000, and the top of the range is $195,750. Those national figures are a PayCrunch estimate, not a Bureau of Labor Statistics published wage for this exact title.
This is one of the more automatable legal-adjacent jobs, and AI genuinely drafts specs and searches art. But the USPTO requires a registered human to sign, owes a duty of candor, and makes the judgment calls on claim scope and strategy that determine a patent's value. The agent who directs AI out-produces peers; the one who ignores it is the one at risk.
Is it safe to use AI on invention disclosures?
Only in an enterprise or zero-retention account under a confidentiality agreement. Pasting unpublished disclosures or claims into a consumer tool can be a public disclosure that destroys patentability and breaches client confidence. Already-published, filed material is lower risk.
Can I trust AI prior-art search and citations?
As a lead generator, yes; as gospel, no. AI can miss key references and can hallucinate ones that do not exist. Verify every reference in the actual database, and never cite or file anything you have not confirmed — a false statement to the USPTO is fraud on the Office.
Does AI change how I should draft claims?
It speeds the mechanics and catches Section 112 issues, but claim scope is strategy, not text generation. Use AI to critique and generate alternatives; you decide breadth, fallbacks, and design-around protection based on the client's business.
How does this get me to $145k?
Patent agents are paid on throughput and expertise. AI lets you draft more, cleaner applications and prosecute more office actions per month in a high-value art unit, which raises your billings and makes you the agent clients and firms fight to keep.
Methodology & sources
Salary (median, 10th, top of the range) — U.S. Bureau of Labor Statistics, OEWS.
By state — the Bureau of Labor Statistics’ own state medians, limited to states employing at least 500 people in the occupation. No cost-of-living arithmetic is applied to a wage anywhere on this page.
The plays — PayCrunch's own step-by-step guidance using publicly available AI tools. Tool names/URLs are real and current as of August 2026; prompts written to work as-is. Verify any professional output before relying on it.