openshield: a batch API and a second server for the moderation and fraud APIs

What openshield is

openshield is a hosted API from Sway SRL with two jobs. It moderates images and text: one call answers accept, review or reject, with the categories, scores and thresholds behind the answer. It also scores sign-ups, logins and payments for fraud: one call answers allow, step up, hold or block, with the rules that fired. Both run on open-source engines (openmod and openfraud, Apache-2.0).

It is live today at https://openshield.dev, in beta. You can try it without paying: every organization gets EUR 2.00 of use each month.

What works today

• Image and text moderation with the full answer on every account. Content is checked in memory and never stored.
• A fraud prevention API in beta, with an instance of its own for each organization, on a server in Germany.
• Prepaid, no subscription. An image with the usual workflow costs EUR 0.002, a fraud decision EUR 0.001 (prices exclude VAT and are the beta prices).
• Three SightEngine-compatible routes, an MCP server, and integration guides for Node.js, Python, Go, PHP, Cloudflare Workers, Supabase Edge Functions and Nuxt/Next.

What this campaign pays for

These two pieces are planned and not built yet:

1. A batch API: send a batch of images or texts, read the results as a file, get a signed webhook when it is done.
2. A second server for the API, for more capacity and for resilience. A dedicated server costs about EUR 100 a month.

If the goal is not met, nobody is charged and the work goes on at the pace of the beta.

How backing works

This is a Build fund campaign: your card is charged at the end of the campaign, only if the goal is met. Each reward is balance on an openshield account, usable as soon as it is added. Like any top-up, it stays usable for 12 months.

Who sells this

Sway SRL (Belgium) sells these rewards and builds openshield. Sway SRL also runs fundmysaas, the platform that hosts this campaign, so it has an interest in this campaign doing well.