An in-house contract signing system with detailed analytics
The signing service showed nothing after send: whether the client opened the contract, which clause stopped them, why they went quiet.
6 manual steps → 1 buttonIn short
Signing ran on an off-the-shelf service, but it lived apart from everything else: data from the client database was retyped by hand, the income forecast was worked out in a spreadsheet and pasted in as a picture, and once the contract was sent nothing was visible — not whether it was opened, not where the reader stopped.
We built signing of our own: from a button on the client record to the finished document. The third-party service stayed inside as a tool, but the client never sees it now.
What hurt
The company manages property in the UK. A contract with an owner is not a two-page formality: it carries a year-long income forecast, commissions and termination terms. People genuinely read it, and genuinely hesitate.
Signing ran on an off-the-shelf service. The service itself was fine. The problem was that it lived apart from everything else — from the client database, from the calculations, and from the manager running the deal.
Four things that did not work
Did they open the contract? Read it through or close it halfway? Which clause gave them pause? The manager called blind.
The client left for a third-party website. No branding of our own, and nowhere to show anything before the signature.
The manager typed the address and the flat's features into a spreadsheet, got an income chart, and pasted it into the document as a picture. Every single time.
Enquiries landed in the client database while the contract was assembled elsewhere: name, address and terms were keyed in again.
What we built
We built signing of our own: from a button on the client record to the finished document. The third-party service stayed inside as a tool, but the client never sees it now.
The system does the income maths
The spreadsheet managers had filled in by hand for years moved inside. Now it is enough to enter the address and tick what the flat has — the twelve-month forecast and the seasonality chart assemble themselves.
It calculates the same way it always did: the figures in the contract match what the manager quotes on the call. There is nowhere left to slip while copying, because there is nothing to copy.
The contract is born from the client record
A «new contract» button appeared on the deal in the client database. Name, phone, address and property details fill themselves in — they were already entered when the enquiry was logged.
The manager only checks them and sets the commercial terms. Six manual steps became one.
Our own page instead of someone else's window
The owner gets a link and lands on the company's own page: a welcome, the income forecast for their flat, the full contract text, and the signature — with a finger on a phone or as a typed name.
What matters most is that there is finally room for what a third-party window had nowhere to show: first the person sees what they will earn, and only then reads the terms. They can book their onboarding right there.
You can see where someone hesitated
The manager sees what happened to the contract: opened or not, read to the end or not, which clause they lingered on, which one they came back to twice.
If someone sat on the commission for a while and then left the page for ten minutes, the call opens with the commission — not with «so, did you have a look?».
What is collected is behaviour, not content: the system sees that a paragraph was re-read, but keeps neither the text nor personal data. And it only runs with the person's consent.
AI suggests what to do next
There is a lot of behavioural data, and no manager is going to sift through it by hand. So AI sits on top: it reads how the person moved through the contract and writes a short summary — where the doubts were, how high the risk is, what to raise on the call.
The assessment rests on exact figures rather than impressions: how many times they returned to a clause, how long they spent on it, where they left the page. The AI only puts that into words and suggests what to fix in the contract itself when the same clause slows everyone down.
It runs on a button, not on every document: paying to analyse contracts nobody opened makes no sense.
After the signature everything moves on its own
- 1
The document is ready
The signed contract and the signing log go into the company archive.
- 2
The deal moves
The status in the client database changes itself — the manager sets nothing.
- 3
Emails go out
The finished document to the owner, a notification to the manager.
- 4
The property form
The owner is asked for the flat's details straight away — not in a separate email a week later.
A few numbers
The same job, as a product
The company is British, but the work is the same anywhere: a contract is put together by hand from data that already sits in the system. The same details, the same numbers, the same hour in a template.
We build this as one of six jobs an AI assistant does for a business — next to reports, moving data between systems and search across your own documents.
Result
Recognise the problem?
Give us 30 minutes and we will tell you which of this can be repeated at your company.