Operations · Data entry
Manual data entry, down almost 90% at one client.
Records typed by hand from one system into another, moved by a workflow with checks. What doesn't meet your rules goes to a person.
Book a free strategy callHow do you automate manual data entry?
Start with a list of what gets typed by hand, from where and into where. The workflow reads the source (a portal, an email, a spreadsheet, a partner export), checks each record against your rules, writes it into your platform through its API and sends anything that fails a check to a person. At one client, in production, manual data entry fell by almost 90%, freeing about one full-time person's time for other work.
- Who approves
- The person who owns the data on your side
- You own
- The workflow, agent logic, integrations, evals and logs
When people are paid to copy data from one screen to another
Data entry by hand costs twice: the hours, and the errors someone finds later. Often it shows up as a contractor line on the monthly IT bill.
“We pay contractors to copy data from one system into another, and we still find typos.”
“Every new record has to be typed twice: once in the portal, once in our platform.”
“When the person who does the data entry is on holiday, the data stops.”
From the source to your platform, with a check at every step
- 1
Map what's typed by hand
Which records, from which source, into which system, how often and by whom. Steps nobody needs are removed before anything is automated.
- 2
Read the source
Through an API where there is one, or from structured exports and emails where there isn't.
- 3
Check against your rules
Required fields, formats, duplicates and the values your business rules allow.
- 4
Write through the API
Valid records are written into your platform by a user that can only create and update the records the job needs.
- 5
Exceptions to a person
Records that fail a check go to a queue with the reason, for a person to fix or reject.
What the AI can touch, who approves, what gets logged
Each solution runs with the narrowest access that does the job. Drafts stay drafts until a person sends them.
It can
- Read the sources you name
- Create and update the records the job needs in your platform
- Flag records that fail a check
- Report volumes and exceptions
It can't
- Delete records
- Change records outside its scope
- Approve its own exceptions
- Send anything to customers or partners
- Who approves
- The person who owns the data on your side. They work the exception queue and approve changes to the rules.
- What's logged
- Each record read, the checks it passed or failed, what was written and when, and who resolved each exception.
In production today, and what it changed
It runs in production, moving records between real systems with a check on every record and the exceptions sent to a person. Manual data entry fell by almost 90% at one client, freeing about one full-time person's time for other work.
- manual data entry, about one full-time person's capacity freed for higher-value work
- −88%
Measured in production: manual data entry before and after automation.
What we measure, before and after: Hours of manual entry per month, error rate, and records waiting in the exception queue.
What you provide, what we set up, how long it takes
You provide
- A list of the data entered by hand today, with a sample
- API access, or exports, for each source and target system
- The rules a record has to meet
- One person who owns the data and the exception queue
We set up
- A map of the data flow, with the steps to remove first
- Connectors to the sources and targets
- Validation rules and the exception queue
- The log and the before/after measure
Embed time
The first run is in parallel with the people doing the work today, so the output can be compared record by record.
What we adapt to your company
The read, check and write loop stays the same. Sources, targets and rules are yours.
- Your sources: portals, emails, spreadsheets, partner exports
- Your platform or ERP, through its API
- Validation rules and duplicate checks
- Where exceptions land and who works them
Case study · Operational Discovery · Automation · DevOps · Cost Reduction
Mid-size company: the IT bill nearly halved
Monthly IT spend down 47%, the hosting plan halved and manual data entry down almost 90%, with one accountable partner in place of several outsourced suppliers.
- monthly IT spend, compared month by month
- −47%
- hosting plan
- −50%
- accountable partner in place of several outsourced suppliers
- 1
Questions before you start
Show us what gets typed by hand.
On a strategy call we go through one data flow with you: where it starts, where it ends, who types it. If the solution doesn't fit, we'll tell you on the call.