Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for erp consultants

Bulk Upload Dry-Run Agent

A bulk load that goes in clean, with counts verified at every step

Bulk Upload Dry-Run Agent: what goes in, what the agent does and what you get

What it does

Bulk uploads fail halfway or load bad data into a live system. This agent validates the file against the target rules, such as required fields, formats and allowed values. It runs a dry run in the system, fixes what it can safely fix, such as trimming spaces, and lists the rest for a person. It then loads in small batches and checks the record counts after each one. If a batch fails the check, it stops. A person approves the live load. Edge case: an ID already exists in the system, so the agent holds those rows as updates and asks whether to overwrite.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueApprovedYes, continueNoNo 1 STARTS WHEN Upload file is ready 2 USES A TOOL Validate each row against the target rules 3 DOES Apply safe fixes such as spacing and format 4 DOES List rows with problems that need a person 5 USES A TOOL Run a dry run in the system 6 CHECKS THE RESULT Does the dry run accept all valid rows? If not: correct the cause and validate and dry-runagain. Back to step 2. 7 YOU APPROVE A person approves the live load 8 USES A TOOL Load in small batches 9 CHECKS THE RESULT Do record counts match the expected counts aftereach batch? If not: stop loading and report the difference. Back tostep 8. 10 RESULT Load report with counts
Read the steps as a list
  1. Upload file is ready
  2. Validate each row against the target rules
  3. Apply safe fixes such as spacing and format
  4. List rows with problems that need a person
  5. Run a dry run in the system
  6. Does the dry run accept all valid rows?If not: correct the cause and validate and dry-run again. Back to step 2.
  7. A person approves the live loadThe agent waits here for your OK.
  8. Load in small batches
  9. Do record counts match the expected counts after each batch?If not: stop loading and report the difference. Back to step 8.
  10. Load report with counts

How it decides

A row is loadable when it passes all rules. Safe fixes are formatting only. Anything that changes meaning goes to a person.

  • Apply only formatting fixes automatically
  • Hold duplicate IDs as a question
  • Load in batches of a set size
  • Stop on any count difference

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Batch size (default: 500)
  • Safe fix list
  • Required fields
  • Duplicate handling

What keeps you in control

It always asks you first

  • A person approves the live load and any overwrite

Hard limits

  • Never loads live data without approval
  • Never overwrites existing records without a decision

It stops when

  • Done: all rows are loaded and counts match
  • Stop: a batch count differs and the load is halted

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensA file of 5,000 contacts had 140 rows with bad phone formats and 22 with existing IDs. The agent fixed the 140 formats and held the 22 for a decision. The dry run accepted 4,978 rows. A person approved the load. Batch 3 of 10 was short by 4 records, so the agent halted and reported the 4 rows that failed.

More agents for erp consultants