There was a tailor in Lajpat Nagar my father refused to replace for twenty years. The man measured things no readymade shop has ever measured: the drop of your left shoulder, which side you keep your wallet, whether you sit all day or stand. He would ask you eleven questions before he touched the cloth. Two shops down, a readymade showroom sold suits in forty sizes with free alteration, and on an ordinary day, for an ordinary occasion, nobody could tell the difference.
Then comes a wedding. Wedding light is cruel light... the photographs from that day outlive you. Under that light, the difference between measured and altered is the entire point.
A patent application lives under wedding light. It will be examined years from now by a person who is paid to find the one seam that does not hold, and opposed, if it ever matters commercially, by lawyers whose whole job is to pull that seam. So when people ask me why we built eety the way we built it, I usually skip the technology and tell them about the tailor. The whole product is eleven questions before touching the cloth.
Part one
The case against template drafting
Prompting a chatbot is not patent drafting. I want to say that carefully, because I run an AI company and the sentence sounds like it is against my own interest :) but it is true and you already know it. A chatbot is the readymade shop. You hand it a disclosure, it hands you a suit. It will give you claims. It will give you an abstract. It will give you a detailed description, delivered with the serene confidence of a man who has never once been cross-examined.
The tell is simple. A chatbot answers; a good drafter asks. Sit beside a senior attorney during an inventor interview and count the questions... the mechanism, the edge case, the one detail the inventor thinks is obvious that is actually the invention. The questions are the drafting. The typing at the end is clerical. Any tool that starts typing the moment you finish uploading has skipped the part of the job that was the job.
So the moral center of eety is a gate. Eety builds a structured model of your invention from whatever you give it, scores its own understanding of that model, and refuses to draft until the understanding crosses 90 percent. Below the line it does not write; it asks. Engineer-grade questions aimed at the exact gaps, one at a time, like the tailor with his measuring tape. People sometimes call this friction. I call it the difference between a tool that asks and a tool that invents, and I know which one I would let near a client's filing.
One admission before we go further. The address of this page calls eety "the first intelligent patent drafting tool." We named the page in our louder days, and the URL stays because links live forever. I am not going to argue for "first"; I cannot verify it and neither can you. The argument I will actually make is narrower and, I think, more useful: patent work does not need a faster typist. It needs an intelligence layer across the whole lifecycle, and the lifecycle is three verbs.
Decide what to patent. Draft it. Verify everything.
The whiteboard in our Gurgaon office still has those three verbs on it, in a handwriting nobody has admitted to. The rest of this page is just those verbs, taken in order.
Part two
Decide before you draft
Actually, let me back up. I have been describing the drafting before the deciding, which is exactly the mistake this industry makes every single day.
Most patent budgets are spent on inventions nobody triaged. Disclosures pile up, a committee meets when it meets, and the filings go to whichever project argued loudest or whichever team management currently loves. The expensive question, "is this worth patenting at all," gets answered after drafting has already begun. That is appointing the wedding caterer before anyone has said yes.
I have sat in that meeting, at more than one company. Disclosures arrive all year; filings get decided in a monthly session built around one Excel sheet; the outside-counsel bill for each filing dwarfs what it would have cost to ask the prior-art question early. On one of those sheets there was a column titled "Gut Feel." Nobody in the room was embarrassed by it, because the gut was the only triage instrument anyone had handed them.
This is what invention harvesting in eety exists for. Two tokens, $20, once per matter. Eety reads the disclosure, runs a state-of-the-art search underneath it, and returns a verdict: prior-art density Low, Medium, or High, with claim charts against the closest references. The search itself is a 15-query ladder: ten queries phrased the way patents are written and five phrased the way scholarly papers are written, because engineers publish before they file, and a search that only speaks patentese misses half the shelf. Every reference comes back with a provenance label, where it came from and why it matched. Receipts, not vibes.
A $20 verdict will not tell you everything; I say more about its limits further down, honestly. What it does is reverse the order of spending. You triage for tens of dollars before you draft for thousands, on every disclosure instead of the loud ones. The Gut Feel column becomes a density column with references attached.
Part three
Draft like the firm
Every firm has a voice. Give the same disclosure to three good drafters and you will get three defensible applications that read nothing alike: different claim rhythm, different hedging, different appetite for boilerplate. A new associate learns the firm's voice the honest way, by reading the partner's old files at night. That is what eety does too. You hand it reference drafts, and it drafts to that style the way the associate would, minus the nights.
The full draft is 5 tokens, $50, and mechanically it works like a team rather than a scribe. Sections are drafted in parallel against the same understanding model, so the claims, the description, and the abstract stay consistent with one shared source of truth instead of drifting the way a single long generation drifts. Templates cover five patent offices, USPTO, EPO, PCT through WIPO, UKIPO, and IPO India, each enforcing its own formatting and conventions. Eety also generates the patent drawings, planned from the same model, numbered and referenced from the text.
And here is the part I am proudest of, which is also the least glamorous part. When eety does not know a fact, the draft says so. You get a marked needs-input placeholder at the exact spot: the operating range the disclosure never stated, the material, the tolerance. A junior who writes "TBD, confirm with client" is annoying. A junior who quietly invents the client's answer is dangerous. Every drafting tool on the market has chosen one of these juniors. We chose the annoying one, on purpose, and I would make that trade again every time.
Tools in this industry advertise 10 to 15 hours saved per application. That is their claim, not mine; I have no way to audit it, and honestly the hour count was never the interesting number. The interesting number is how many invented facts survive into a filed document. The design goal of eety's drafting is a draft where every fact is either yours or flagged, and where knowledge runs out, the placeholder is the proof that we stopped.
Part four
Verify everything
My mechanic in Saket returns the old parts. A little steel tray, worn brake pads sitting in it like evidence. He does not do it because I can judge a brake pad... I cannot. He does it because the tray changes the relationship. The tray says: check me.
Most outside-counsel drafts arrive with no tray. A clean Word file, forty pages, and by page nine you have got done with careful reading and started skimming, because the deadline reads faster than you do. You sign because the calendar signed first. Everyone in this profession knows that feeling and nobody enjoys admitting it.
Vendor review is eety's tray. Three tokens, $30. Import the vendor's DOCX and eety matches its headings to the right sections automatically, offers manual mapping when a vendor's headings are creative, and recognises an India Form 2 when it sees one. Then it runs the full legal review battery in parallel across every section, and each finding lands as an error, a warning, or an informational note, citing the actual provision it rests on, Section 10(5) of the Indian Patents Act, for instance, on claim clarity and support. Not "this looks off." This fails here, per this statute.
Then the part your vendor will actually feel: the whole review exports as a tracked-changes DOCX redline, every edit attributed by name, ready to send back. Not a PDF of complaints. A redline in the format their own tools open, with a reason attached to every change. The first redline you return like that changes the second draft you receive; checkable beats impressive. It is the tray on the counter, again.
The honest part
What eety is not
Since this whole page is about tools that admit what they do not know, let me hold myself to the same standard.
Eety is not a law firm, and nothing it produces is legal advice. It is drafting and analysis software for professionals who carry the professional responsibility.
The state-of-the-art search is a triage instrument, not a professional prior-art search. It tells you where the crowd is before you spend real money. When the stakes call for a professional search or a freedom-to-operate opinion, get one. The $20 verdict is for deciding which matters deserve that spend.
Nothing should be filed without a qualified practitioner reviewing it and taking responsibility for it. The 90 percent gate, the placeholders, the statute citations all exist to make that review faster and sharper. None of it exists to replace the reviewer, and if a tool ever tells you otherwise, ask it a hard question and watch what happens.
A note on money
Deliberately boring pricing
A token is $10 of Eety power. Each service is charged once per matter: harvesting 2 tokens, the state-of-the-art search 1, the full draft 5, vendor review 3. Signing up gives you 25 free tokens, $250 of work, no credit card. If you want packs: 25 tokens for $250, 112 for $1,000 at $9 a token, or 667 for $5,000 at $7.50. The rest of this industry sells enterprise patent software at $30,000 to $100,000-plus a year, behind a demo call and a quarter of procurement. The full comparison is on the pricing page, and when your team needs seats, roles, a shared token pool with a ledger, and an audit trail, that is the enterprise page.
Closing
The tailor, again
So, the tailor. He handed the shop to his son, who I am told does good work, and the readymade store two doors down is now three floors tall. Both are still there... which is roughly my prediction for this industry too. Templates will always be cheaper, and for some filings, maybe that is fine. But for the ones that will one day stand under the harsh light, you want the eleven questions asked first, by something that would rather ask you than invent for you.
If you want to test whether any of this is real, do not take this page's word for it. Take the 25 free tokens, run one real matter through all three verbs, decide, draft, verify, and read what comes back. Or book a demo and I will walk you through it myself.
There is a version of this essay that ends with a clean line about the future of patent practice. I keep deleting it. The tailor never made speeches either; he just measured twice.
Tabrez Alam, Founder, eety.ai