CASE STUDY

From 40% margins to 87% how BetterPic turned AI infrastructure into a competitive advantage

Reliable, automated, and 2x more profitable. A Runflow case study.

87% Gross Margin
40% → 87% in 12 months
Zero Infra Team
BetterPic×Runflow
BP

BetterPic

AI-powered professional headshot platform

Product

4K AI headshots for individuals & teams, with AI editing studio

Revenue

$2.3M/yr

Pricing

Starting from $35 per headshot set

Team

Lean engineering team, no dedicated ML or DevOps headcount

Challenge

GPU unreliability causing 24h customer wait times and margin compression at ~40%

THE PROBLEM

Customers were waiting 24 hours.

BetterPic was growing fast, but their infrastructure couldn't keep up. Customers were paying $35+ for professional AI headshots and waiting up to 24 hours to get them.

GPU jobs were failing silently. No retry logic, no fallback providers, no automatic recovery. When a job failed, it just… didn't complete.

Customer complaints were increasing. Support tickets piling up. The engineering team was spending more time debugging infrastructure than improving the product.

On top of the reliability problems, gross margins sat at ~40%. Every new customer meant more GPU spend with no predictability. They faced an impossible choice: hire an expensive DevOps/ML infrastructure team, or keep firefighting and hope it held together.

"Customers were paying for AI headshots and waiting a full day. That's not a cost problem, that's a product crisis."

~40%
Gross margin
24h
Customer wait time
Silent
Job failures
0
Retry mechanisms
THE SOLUTION

Two endpoints.
No ML, AI, or Infrastructure team needed.

Runflow's infrastructure was born from solving this exact problem internally. The team that built it had lived through the same GPU unreliability, the same firefighting, the same margin pressure.

Jobs stopped failing. Automatic retry, intelligent GPU routing, failover across providers — customer wait times collapsed. GPU quota limits and availability problems that used to cause outages were solved by routing across multiple datacenters. On top of reliability, Runflow's orchestration layer drove down inference costs through model quantization, multi-step workflows combining the best open-source and closed-source models, and intelligent GPU scheduling across providers. Margins climbed from 40% to 87% over 12 months.

BetterPic's developers integrate new AI features by calling a single endpoint. No machine learning expertise needed. No infrastructure knowledge needed. No DevOps team needed.

Reliable GPU Orchestration

Automatic retry, multi-provider failover, intelligent routing. Jobs don't fail, they reroute.

10+ AI Workflows in Production

Headshot generation, quality scoring, clothing swap, background editing, skin correction, and more, all through single API endpoints.

Zero Infra Overhead

BetterPic's developers call endpoints. Runflow handles GPUs, scaling, monitoring, and optimization behind the scenes.

API Request
curl -X POST https://api.runflow.io/v1/flows/headshot/run \
  -H "Authorization: Bearer rf_..." \
  -H "Content-Type: application/json" \
  -d '{"images": ["selfie_1.jpg", "selfie_2.jpg"],
       "style": "professional",
       "quality": "premium"}'
THE RESULTS

Margins nearly doubled. In under two months.

The steepest gains came in the first 3–4 months. The curve tells the story: rapid initial impact from fixing reliability, followed by compounding optimization.

Gross Margin Trend

January 2025 — January 2026

STRONG UPWARD TREND
100%80%60%40%
Gross Margin Trend
Jan '25Mar '25May '25Jul '25Sep '25Nov '25Jan '26
87%
Gross Margin
Up from ~40% in Jan 2025
3 months
Time to Impact
Steepest gains in first quarter
10+
AI Workflows
Distinct inference pipelines
Zero
Infra Team
No dedicated ML/DevOps headcount
Before Runflow
  • ~40% gross margin
  • Customers waiting 24 hours
  • Jobs failing silently
After Runflow
  • 87% gross margin
  • Headshots delivered reliably
  • 10+ AI workflows running on autopilot
"
Working with Runflow changed how we build. Our developers can integrate a new AI feature by calling an endpoint, they don't need machine learning expertise, infrastructure knowledge, or DevOps skills. We just focus on making the best headshot product, and Runflow handles everything underneath.
Thibaut Hennau
Thibaut Hennau
CEO, BetterPic
HOW IT WORKS

From API call to delivered result.

01

Call an endpoint

BetterPic's developers make a single API call, no GPU config, no model setup.

02

Runflow orchestrates

Intelligent routing across GPU providers, automatic retry on failure, cost-optimal allocation.

03

AI workflows execute

Headshot generation, quality scoring, editing tools, all running on managed infrastructure.

04

Results delivered

Processed outputs returned. Reliable, fast, and at a fraction of the original GPU cost.

WORKFLOW SHOWCASE

What Runflow powers for BetterPic.

Not a single model call, an entire product's AI stack across 10+ distinct inference pipelines.

AI Headshot Generation

AI Headshot Generation

Upload selfies, get 4K professional headshots in multiple styles.

Quality Scoring

Quality Scoring

Automated ranking of generated headshots, only the best reach the customer.

Clothing Swap

Clothing Swap

One-click outfit changes, 4 variations generated per edit.

Background Change

Background Change

Transform scene to professional, outdoor, or minimalist settings.

Magic Fix

Magic Fix

Brush-based targeted editing with AI prompt guidance.

Remove Background

Remove Background

Isolate subject with transparent or colored background.

Ready to stop firefighting your GPU infrastructure?

Join BetterPic and other AI companies running on Runflow. Better margins, better reliability, zero infra overhead.

SOC2 Certified
Trusted by BetterPic & more