Ninety days is not a legal period. No statute recognises it, no office counts it, and nothing bad happens on the ninety-first morning that was not already going to happen. It is simply the window in which most independent inventors make every irreversible decision about their idea — usually without noticing that they are making them, and almost always in the wrong order.
The errors repeat with dull consistency. The idea is described to eleven people before it is described on paper. Money goes into tooling a prototype before anyone checks whether the mechanism was patented and abandoned two decades ago. A provisional application gets filed in an afternoon, runs to two and a half pages, and turns out twelve months later to have protected roughly none of what its owner believed it protected. What follows is the same three months run properly, stage by stage, with the decisions flagged where they occur.
Days 1–7 · The Record
The First Week Is for Writing, Not Building
An idea, in itself, is not property. This is the misunderstanding that costs the most and is corrected the latest. "A bin that sorts recycling" belongs to nobody. What can be owned is a particular technical solution — described in enough detail that a competent person working in the field could build it from the description alone, without inventing anything themselves to fill the gaps. That standard is called enablement, and every other requirement in the system sits downstream of it.
Which makes the first week's work specification rather than invention. Not "what if there were a device that…", but: what are the components, how are they arranged, what does each one do, and by what mechanism does the arrangement produce the result? An inventor who can answer those four questions on paper has something a drafter can work from. One who cannot has an intention, and no amount of later spending converts the second into the first.
What week one should produce
- A dated written description of the problem and the mechanism that solves it, in ordinary technical language.
- Dimensioned sketches, however rough, showing how the parts relate to one another.
- A list of the variants you considered and rejected, with the reason for each rejection.
- Photographs of any mock-up, however crude, kept as separate dated files rather than overwritten.
- A note of every person who has already heard about it, and on what date.
Keep versions rather than replacing them. The trail of rejected variants is not clutter: it is the evidence that the solution was arrived at deliberately, and it is frequently the material from which dependent claims are later built. How that record should be structured, and what makes one durable rather than decorative, is worth a longer look in its own right — the mechanics of a working invention record are unglamorous and decisive.
Days 8–30 · The Silence
The Month You Say Almost Nothing
The second stage is the one people find hardest, because it consists of not doing something. A new idea generates an urge to tell people, and the appetite on the other side is genuine: invention stories are told approvingly everywhere, and the cultural fascination with the moment of innovation keeps a steady audience for them. The legal system is considerably less romantic about it.
Describing an invention publicly before filing can destroy the right to patent it. Some systems allow a twelve-month grace period in which the inventor's own disclosure does not count against them. Most do not: the rule in a large part of the world is absolute novelty, meaning that the first public description — a trade stand, a crowdfunding page, a detailed forum post, a video showing how the mechanism works — becomes prior art against the inventor on the day it appears. Rights that existed on Tuesday are gone on Wednesday, and no fee reinstates them.
"Public" is broader than it sounds. It does not require an audience, a sale, or anyone actually reading the thing. A description available to a person who was under no obligation to keep it confidential is generally enough. Conversations with a patent attorney are privileged; conversations with a manufacturer are not, unless a confidentiality agreement was signed before the conversation rather than after it.
Tell nobody, or tell them under a signed agreement. There is no third arrangement that reliably ends well.
The rule that saves the most rights
Days 21–45 · The Search
Finding Out Whether Somebody Already Did It
Novelty is a matter of record, not opinion. Before any money is committed to drafting, the field needs searching: granted patents, published applications that never granted, expired patents whose teaching is now free for anyone to use, product catalogues, trade literature and academic papers. Expired prior art defeats a claim exactly as thoroughly as a live competitor's patent, and it is the category first-time searchers forget to look at.
Searches rarely come back empty, and that is not the verdict it feels like. Almost nothing is entirely new. The question is whether this specific combination — this mechanism, applied in this way, to this problem — has already been described somewhere. Very often the search does its real work by redirecting: the headline concept turns out to be anticipated, while a detail the inventor considered incidental turns out to be the part nobody has claimed. Redirection at this stage costs a few hundred in search fees. Redirection after a rejection costs a year and a drafting bill.
This is also where most people stop working alone. Searching is a research discipline, drafting is a legal one, and neither is learned quickly on a first project. Practical help with patenting invention ideas tends to be worth more here than at any later point, because the choices made now — what to claim, what to abandon, whether to file at all — set the ceiling on everything that follows. Organisations that have spent four decades championing independent inventors tend to make the same observation: the expensive mistakes are cheap to make and slow to surface. The method itself is straightforward enough to learn, and a first search can be run before anyone is paid anything.
Days 45–75 · The Filing
Provisional, Non-Provisional, and What Each One Buys
A provisional application is a placeholder. It is never examined, never published, and never becomes a patent. What it does is fix a filing date and permit the words "patent pending" to be used honestly for the twelve months that follow. Official fees for an individual filer run in the low hundreds, which is why it is so often treated as the cheap option — and why it is so often written as though cheapness were its purpose.
It is not. A later non-provisional can claim the benefit of a provisional only to the extent the provisional actually described the invention. A thin provisional supports a thin sliver of the eventual claims, and the gap is discovered at the worst possible moment: after a year of disclosure made in reliance on it. The document should be written as though it were the full specification, because for every purpose that matters, it is. What the phrase does and does not confer in the meantime is widely misread — the practical value of pending status is real but much narrower than most people assume.
The non-provisional is the substantive filing: specification, drawings and claims. It enters a queue, publishes at eighteen months from the priority date unless publication is actively opted out of, and typically draws a first examination report somewhere between eighteen and thirty months after filing. That report is usually a rejection. This is normal and expected rather than a judgment; the exchange of arguments and narrowing amendments that follows is where the real scope of the eventual patent gets settled.
The choice between filing provisionally and going straight to a full application is a question of readiness rather than budget. If the design is still moving, twelve months of protected iteration has genuine value. If it has settled, the provisional adds a year of delay to a process already measured in years. Fields where the physical form is inseparable from the function tend to reward filing later and more completely — the case for inventor-led design in the built environment rests partly on how long such solutions take to reach their final geometry.
Days 75–90 · The Ledger
What the Ninetieth Day Should Look Like
By the end of the third month the position should be legible to somebody who has never heard of the project. Not necessarily filed — deciding not to file is a perfectly respectable outcome, provided it is a decision rather than a drift — but legible.
The day ninety checklist
- A dated, versioned record of the invention, complete enough that a drafter could work from it unaided.
- A written search result: what was found, what it covers, and why the invention is or is not distinguishable from it.
- A list of everyone who has been told, with dates, and confirmation of whether each conversation was covered by an agreement.
- Either a filed application with its receipt and its priority date, or a written reason for not filing.
- A calendar carrying the twelve-month anniversary, marked at nine months as well as twelve.
That last item is worth more than it looks. The twelve-month deadline is absolute: miss it and the provisional lapses, its date is lost, and any disclosure made in the interim now sits in the prior art against the inventor's own later filing. Nine months is when the work of preparing the non-provisional has to start if it is going to be finished calmly. Inventors who diarise only the deadline itself reliably spend the last three weeks paying for haste.
The first ninety days cost almost nothing and decide almost everything. Every later expense is spent defending the choices made in them.
Why the cheap stage is the important one
None of the five stages is technically difficult in isolation. What defeats people is sequence: disclosing before filing, searching after drafting, letting a placeholder expire while waiting for a manufacturer to return a call. Run in order, the paperwork stops being an obstacle and becomes what it was built to be — the mechanism by which one person's specific solution to a specific irritation becomes something a company has to negotiate for rather than simply take.
Ninety days, four documents and one diarised date. Everything after that is negotiation.