Friday, 5:30 pm, in an ophthalmic surgery clinic. The secretary opens the file of the last scheduled patient, retypes their identity into the consent template, counts out the follow-up appointments on the calendar while dodging the weekend and the public holiday the following Thursday, prints the document and slips it into an envelope. The next week she will reopen that file three times: to check whether the signed consent has come back, to chase the patient who never returned it, then to scan and file the signed version. None of these tasks calls for any thought. All of them call for time.
This small piece of theatre replays for every scheduled patient, and its variants exist in every trade: a contract to get signed, a quote to get approved, a mandate to get initialled. It is the textbook process to automate, precisely because every step is mechanical, yet none of them watches over itself. Our public demo of automated signing shows what this process becomes when nobody has to carry it any longer: one entry, then eight steps that run without intervention, from working out the follow-up dates all the way to archiving the signed consent. This article tells what the demo shows, and what its apparent simplicity actually required.
Three time sinks nobody sees
The manual process is not expensive per step. It is expensive because it stacks up invisible minutes, spread across days, with an interruption every time.
The first time sink is gathering the data. The patient’s identity, the planned procedure, the surgery date, the eye concerned: all of this already exists in the file when the moment comes to prepare the documents. Yet someone has to look it up, retype it into the quote template, then into the consent, then into the calendar. Each retyping takes two or three minutes and each one can introduce a typo, on a name as easily as on a date.
The second time sink is scheduling the post-op checks. One check at three days, one at seven, one at a month: the rule is easy to state, but applying it means counting on a calendar, avoiding weekends and public holidays, finding free slots, writing down three appointments and confirming them to the patient. Ten minutes of calendar gymnastics per file, to be redone whenever the surgery date moves.
The third time sink is the worst: tracking the signature of the informed consent. Once the document has gone out, the wait begins, and that wait has no deadline, because the patient signs when they please. Someone therefore has to check regularly what has come back, spot what is missing, chase, then file the signed documents. This work appears in no job description. It gets paid for in interruptions, in mental to-do lists, and in missing consents discovered the day before surgery.
What the demo shows: one click, eight steps
The demo is public and plays out live, in five languages. The form fits in three fields: first name, last name, email, plus a consent checkbox. One click on “Generate my document”, and the pipeline runs before your eyes. The headline sets the promise plainly: “One click. Eight automated steps. Zero manual work.”
The eight automated steps chain together. The post-op dates get calculated and confirmed, then written into the calendar. The pre-op file gets generated, and so does the informed consent. The document is prepared for signing, with the signature fields placed automatically, before it goes out for signature. Then, the moment the patient has signed, the signed PDF gets archived and a confirmation email leaves with the document attached. A timestamped timeline shows each operation to the second. On our run, the first six steps took three seconds.
The demo plays it straight. The documents carry a MOCK watermark and have no medical or legal value, the calendar is simulated, the data is purged automatically. But nothing here is an animation: every step really runs, the PDFs are really generated, the signing email really lands in your inbox. It is a genuine document flow, simply unplugged from the real world.
One entry, coherent data everywhere
In the manual process, the same information lives in several copies: one version in the file, one in the quote, one in the consent, one in the calendar. Each copy was born of a retyping, and each retyping is a chance to diverge. The name misspelled on the consent, the date from the old schedule left in the quote: these mistakes are not individual lapses of attention, they are the normal product of a system that duplicates information by hand.
In the automated flow, the entry exists only once. The pre-op file and the informed consent are generated from the same source, in the same second; the calendar receives the very dates printed in the file. Coherence is no longer a virtue to keep up, it becomes a property of the system.
A word on how credible this part is: the engine that produces the demo’s PDFs is not a prop written for the occasion. It is the anonymized version of a module we built for a refractive surgery clinic in Brussels, where it runs in production. The demo documents have the structure, the layout and the density of the real ones, watermark aside.
Post-op dates calculated, not retyped
The clinic’s rule fits in one sentence: a check at day 3, one at day 7, one at day 30 after surgery, on business days, at fixed times set once and for all. The demo applies it in a fraction of a second: the three dates are calculated, shifted if they fall on a weekend, booked in the calendar with their slots, and printed into the file handed to the patient.
The manual version of this rule does not fit in one sentence. It lives in the head of a person who counts the days on a calendar while avoiding public holidays, picks slots, writes three appointments into the diary, retypes them for the patient, and starts over if the procedure gets moved. It is exactly the kind of business rule that deserves to be encoded: perfectly clear, applied hundreds of times a year, and yet written down nowhere except in habit.
The signature: sent, tracked, archived with no manual chasing
Here is the heart of it, the part that cost the most time. In the demo, the informed consent goes out for electronic signature through Documenso. Preparing the document is itself automatic: the system works out where to place the fields, sets initials on each page, an agreement box, the date and the signature, then creates the document on the platform and sends the email to the patient. The signing link is available at once, without waiting for the email.
The patient opens the document, reads through it, initials each page, ticks their agreement and signs. And this is where the process reverses. In the manual world, someone had to check what had come back. In the automated flow, the system is the one that gets told: the second the signature is completed, the platform notifies the server, which archives the signed PDF and sends the patient a confirmation email with the document attached. On our run, four seconds separated the signature from the confirmation email. The tracking page, for its part, updates on its own, without anyone refreshing it. While the wait lasts, it reads “Waiting for your signature…”.
Nobody watches anything any more, and yet nothing slips past tracking: an unsigned consent stays visible as such, without anyone having to reopen files to notice it. Chasing becomes an informed decision rather than a checking round.
One last point, which counts twice over in a medical setting: Documenso is an open source solution, and the instance that signs the demo’s documents is hosted on our own infrastructure. The documents are entrusted to no third-party service; they are born, signed and archived on the clinic’s server. This setup installs identically at a client’s premises, including entirely on site.
Under the hood: what “automatic” demanded
As always, the visible smoothness rests on invisible work. The three-field form is protected by anti-abuse guardrails: a bot trap, a minimum submission delay, limits per address and per day. The timeline scrolling live is fed by a real stream of events sent by the server, not by a timed animation. The signature notification rests on an authenticated, idempotent webhook: if the platform notifies twice, the document is archived only once, and the confirmation email leaves only once. And because the demo is public, the data is purged automatically: seven days after generation, forty-eight hours if the document is never signed.
| What the user sees | What it took |
|---|---|
| A three-field form | Anti-abuse guardrails: a bot trap, a minimum delay, limits per address and per day |
| A timeline animating live | Real event orchestration in real time between the server and the browser |
| An “instant” confirmation email | An authenticated, idempotent webhook, with error recovery |
| A public demo open to all | Automatic data purge: 7 days, 48 hours if never signed |
The method is the same as on all our projects, described in our automation method: start from the clinic’s real process, encode its rules as they are actually practised, and keep the human in charge. Automation invented no new process: it took the one that already existed and stripped out the retyping, the calendar arithmetic and the watching.
And in your sector?
This flow is medical in its setting, not in its structure. Replace the informed consent with a client contract, a management mandate, a quote to approve, an internship agreement or an onboarding file, and you find the same three time sinks: data to gather that already exists elsewhere, deadlines to calculate that follow a known rule, and a signature to track that waits on no one.
Automation changes the nature of the work rather than its apparent volume: the entry happens only once, the deadlines calculate themselves, and tracking becomes free, in the literal sense, since it no longer consumes anyone’s attention. What stays with the human is what deserves a human: deciding, checking, keeping the relationship alive.
If your business generates documents that have to be signed then tracked, transtorm.ai can build this flow with your templates, your rules and your infrastructure. Book a conversation below.
Book a call