Use agents to reconcile files and flag bad records
A developer set up agents to load a legacy file into a database, compare it with active work, and draft clarification messages for problems.
- What they did
- The developer built a workflow for a specific industry file that usually belongs in a legacy system. They used a low-cost model to read the file, turn it into database records, and launch agents that compared those records with work already underway. Other agents reached out to the relevant people when data looked wrong, while additional agents kept checking new updates against many items.
- How it went
- They say the system can stand in for a service that costs more than $500 an hour and can spot errors as they happen across languages and cultures. There were no measured benchmarks shared.
- Tips
- Expect a lot of tuning and test runs before it behaves reliably. Ongoing maintenance mainly means keeping the tests up to date and reviewing anything that might send a message or change a record.
Try it yourself with Claude Code
I get a [file type] from [source] every [time period]. Build a workflow that imports it into a local [SQLite/Postgres] database, compares each row with my [active work source], and flags mismatches or suspicious records. For each issue, draft a clear message to the responsible person and list the exact question to ask. Stop before sending, submitting, deleting, or changing anything, and wait for my approval before taking any irreversible action.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
1 more report of this
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Screen supplier invoices for duplicates and GST gaps
An AP workflow checks incoming supplier bills against past records so a person can review problems before approval.
Audit your mail DNS and add DMARC
An agent reviewed a domain’s email DNS, spotted a missing DMARC record, and a single TXT update raised the score from 80 to 96.
Combine homework submissions into one daily report
A teacher used a no-code agent to pull homework status from several places and merge it into a single check-in report.
Use a background agent with inbox-style approvals
They used a desktop app to run agents in the background and review results later in separate inboxes for finished work and decisions.
Use a voice agent for oral exams
A professor used a voice agent to run oral finals, then had an LLM panel grade the transcripts and flag shared weak spots.
Use a text grid to let agents handle web forms
They made a browser view that turns pages into numbered text so an agent can work through applications without image models.