Runflow

Ship AI image features. Without the complexity.

5 followers

Ship AI image features. Without the complexity.

5 followers

Production-ready AI image APIs with built-in quality evaluation. Background removal, on-model editing, headshots, and more. Every image scored before delivery. Go live in hours, not months.

Free Options

Launch tags:API•SaaS•Photography

Launch Team / Built With

LightfieldAI-native CRM that builds itself and does work for you

Promoted

BetterPic

Maker

📌

Hey Product Hunt 👋 I'm Ricardo, co-founder of Runflow. Here's the honest origin story. Two years ago, we were scaling BetterPic, an AI headshot product. At peak, we were running 100K+ AI jobs a month. Our gross margin was sitting at 40%. That's not a business, that's a treadmill. So we built our own infrastructure layer: workflow orchestration, output quality filtering, cost routing. Margin went to 89%. Same product, same customers, radically different unit economics. Then we realized: every team building on AI image APIs is solving this exact problem from scratch. Stitching together RunPod, fal.ai, Replicate. Writing their own retry logic. Manually QA-ing outputs. Hoping the model doesn't hallucinate a sixth finger on a hand. That internal infrastructure became Runflow. For Builders, we offer Image Solutions API's without the complexity of hiring AI Engineers. For AI Engineers, we turn your ComfyUI workflows into a production API endpoint in one click. With uptime, scaling, and Sentinel, our quality evaluation layer that scores and filters outputs before they ever reach your users. We sit in the managed middle. Not a raw GPU provider. Not a black-box API. You keep control of your pipeline. We handle everything that breaks at 3am. We're early. BetterPic is our Customer Zero. First external customers are live now. If you're building anything with AI image pipelines, I'd genuinely love to hear what's breaking for you. Drop a comment or find me here. And if you find this useful, an upvote goes a long way. 🙏

Report

5mo ago

BetterPic

Maker

CTO here. If you've ever shipped an AI image feature to production, you know the real work starts after the model call returns. Queue, retry, fallback, eval, cost routing, observability, autoscaling.

And it gets worse the second you want custom workflows. Suddenly you're deploying ComfyUI, packaging containers, orchestrating GPU workers, handling cold starts. Months of infra work before a single user sees an image.

Every team I've talked to is rebuilding the same plumbing. That's what we obsessed over.

Two things worth flagging:

Sentinel is the piece I'm proudest of. It detects the use case, picks the right preprocessors (OpenPose, face similarity, OCR, etc), runs a dynamic rubric, and gives a verdict before the image ships. Most platforms skip eval entirely and push that cost to you.
Solution APIs let you stay high-level, and our 500+ building blocks let you drop down and compose custom workflows yourself, in a simple to use UIX. Same infra, your call on abstraction.

Ask me anything technical, infra, eval, routing, ComfyUI in prod, whatever! :)

Report

5mo ago

Reviews