How data gets into buffers.ai
buffers.ai runs on your sales history. This page is the honest, specific version of "how does it connect to what I already run" — what's required, what's been done in production, and how a rollout gets scoped.
Evaluate with a file. Go live with a feed.
To evaluate: upload a file
The live benchmark at buffers.ai/benchmark/webinar takes a CSV or XLSX export of your own sales history and runs it through the same five-algorithm backtest used in production — no ERP work required to see how it performs on your catalog.
To go live: batch data exchange
Production rollouts run on a recurring batch feed from your existing system — FTP-based exchange with the customer's ERP has been used in deployment (see the RAI case study) — configured around your specific stack rather than a fixed set of named connectors.
buffers.ai does not currently publish a list of named, pre-built ERP or POS connectors. What's actually true, based on completed rollouts: the team configures batch/FTP data exchange directly against a customer's existing system, including catalog-specific handling like fractional units, weight-based products, and vendor-specific naming. If your evaluation requires a named-connector answer for a specific platform, ask directly — that's a faster, more accurate answer than anything published here.
Three columns are required
One row per SKU per day. Everything beyond the three required columns is optional and depends on your catalog.
Required
- date
- Calendar day, YYYY-MM-DD
- sku
- Product / SKU identifier
- sale
- Units sold that day (0 on no-sale days)
Optional, if you have it
Never required, but the more consistently they're filled in, the more the forecasting models can learn from them.
Working with your existing stack
On RAI's rollout — a grocery retailer managing 6,000–7,000 SKUs — buffers.ai's team worked directly with RAI's internal developers to configure FTP-based data exchange with their existing ERP, handle fractional and weight-based units, localize product naming for vendor communication, and resolve unit inconsistencies across suppliers, with ongoing technical support as the integration matured. That's the model: configuration around your system, not a requirement to adopt a new one.
Full write-up: How RAI saved 3 hours per day per store manager.
Frequently Asked Questions
Tell us your stack, we'll scope the integration
Every rollout starts with a conversation about your specific ERP, POS, or WMS — that's a faster, more accurate answer than a generic connector list.