Automate data entry into systems that have no API

The problem

Many businesses run on a system that does its job well but can't talk to anything else. It might be a student records system, a case management tool, a desktop accounting package or an industry system the vendor stopped updating. It holds years of data, and your team knows it inside out. But it has no API, so nothing flows in or out on its own.

So people become the integration. A form arrives by email, and someone retypes it into the system. A report comes out of the system, and someone copies the figures into a spreadsheet. The same customer is entered twice, in two places, by two people. Every retyped field is a chance for a typo, and the work grows with the business.

Replacing the system is often the right long-term answer, but it is rarely quick. Data has to be moved, staff retrained and processes rebuilt. In the meantime, the retyping continues, and the people who know the business best spend their days copying it from one screen to another.

What we build

We build legacy data entry on Maestro, Worcia's AI process automation. Where an older system needs a better front end, or is ready to be replaced in stages, we build that with Custom software, Worcia's software development for internal tools and business systems.

1. We find the most reliable route in. We look for the best way to reach the system, in this order: a partial or undocumented API, scheduled file imports and exports, database access the vendor supports and, only then, screen automation. We tell you which route we chose and why.
2. Incoming data is checked first. Forms, emails and documents are read, and each field is checked against your rules before anything touches the old system: required fields present, formats correct, no duplicate record.
3. Entry happens through the chosen route. With imports, the automation builds the file in the exact format the system expects. With screen automation, a software robot signs in with its own account and fills in the screens the way a trained member of staff would. This is a form of robotic process automation.
4. Every entry is verified. After saving, the automation reads the record back and compares it with what it meant to enter. A mismatch stops the run and raises an alert.
5. A person supervises the exceptions. Anything unclear, such as a missing field, a possible duplicate or a screen that looks different from usual, goes to a named member of staff with the source document and the step where it stopped.
6. Data comes back out, too. The same routes can pull data out of the old system on a schedule, so your CRM, accounting software or reports stay current without anyone exporting and copying.
7. Every run is logged. You can see what was entered, when, from which source and what was held back.

Screen automation is the most fragile of these routes. It can break when the vendor changes a screen or a pop-up appears that wasn't there before. That is why we use it only where nothing better exists, watch it closely and keep a person responsible for the exceptions. The robot account gets only the permissions the job needs, and its credentials are stored separately from your staff's.

How it connects

Older systems usually hold data that other processes need. Once Maestro can read from and write to them, that data can move to and from the rest of your tools.

• Bookkeeping. If the older system is your accounting package, the same approach feeds bookkeeping automation.
• Onboarding. New clients, students or customers captured through client onboarding can be entered into the old system automatically, so nobody types them in twice.
• Portals. A client portal can show status that lives in the old system, read on a schedule, without giving clients access to the system itself.
• Replacement, in stages. When you're ready to move off the old system, Custom software can build its replacement while the automation keeps both systems in step during the move. Read custom software vs off-the-shelf SaaS for how to decide.

For more processes like this one, see workflow automation examples for small businesses.

What we measure

Before launch, we time a sample of manual entries with your team and record the error rate in that sample. After launch, we report the same measures against the agreed targets.

• Records entered and verified without retyping. The share of incoming records entered and read back correctly with no person typing them. Measured from the run log.
• Entry errors found in review or downstream. Records found to be wrong after entry, whether caught in review or later by a colleague or customer. Measured from corrections your team logs, compared with the baseline sample.
• Exceptions sent to a person, and the reasons. The count of held records, grouped by reason: missing data, possible duplicate, rejected by the system, changed screen. Measured from the exception queue. Recurring reasons become new rules.
• Staff hours spent on data entry into the system. Measured from before-and-after time logs for the same kinds of records.