Hey everyone, I'm looking to automate a process in purchasing at a wholesale/distribution company and I'm trying to figure out if I can automate a very manual part of our order process using Claude's API. I have zero dev experience but I'm willing to learn - just want to know if this is actually doable before I commit.
**The current process (very manual):**
We receive large wholesale orders from customers, usually as Excel files. Each order has three columns: product code, description, and quantity.
For every single line on the order, someone has to:
Look up that product in our system
Find all available stock batches in our warehouse
Apply a set of rules to pick the best batch - things like expiry date, warehouse location, and minimum stock thresholds (anything below a certain level is classed as unavailable even if it shows stock)
Adjust quantities so they match full box/tray sizes (we can only ship in multiples of a tray quantity)
Note anything that's out of stock separately - we have a weekly out-of-stocks list with expected return dates supplied by purchasing
The output we need is basically: product, best batch, expiry date, quantity after tray adjustment, and a separate section for anything that's out of stock or unavailable.
**What I'm imagining:**
A simple internal tool where I upload the customer's Excel order file, it queries our system (which runs on SQL), applies all of the above rules, and spits out a completed report. No one needs to manually touch the ERP. I want Claude to handle the logic - things like "pick the earliest expiry batch above the minimum threshold, round down to the nearest tray size, flag everything else as OOS."
The out-of-stocks list would be uploaded separately once a week by purchasing, and the tool would cross-reference it automatically.
**My questions:**
- Is this a realistic project for someone learning to code? What stack would you recommend?
- How would Claude API fit into this - is it the right tool for applying business rules to structured data, or is there a better approach?
- The system runs on SQL - how straightforward is it to query a SQL database as part of a pipeline like this?
- Any gotchas I should know about before I start?
Appreciate any advice - happy to answer questions if I haven't explained something clearly.