Image Capture
Rebuilding the desktop tool embryologists use to photograph eggs and embryos — so the image is the interface, capture stays fast, and quality is caught before it ever reaches the AI.
Future Fertility uses AI to assess egg and embryo quality from microscope images taken in IVF and egg-freezing labs. FF Capture is the tool at the front of that pipeline: the desktop app embryologists use at the microscope to photograph each oocyte or embryo. Those images are what the AI reads — and what later becomes the VIOLET and MAGENTA reports clinicians walk patients through. Capture the image, assess it, report it: this is the capture end.
The app worked and labs trusted it, but it had been built feature-by-feature by engineers and looked a decade behind the science it served. I led the redesign as the sole designer — running discovery interviews with embryologists, mapping the real lab workflow, and prototyping a modern, image-first experience — working closely with the CTO and dev team to keep it buildable, and with the internal embryology team for fast, iterative feedback.
The image was the smallest thing on screen
FF Capture was built by engineers, for a job it did reliably — and the labs that used it had genuine affection for it. But it had grown the way dev-built tools do: dialog by dialog, toolbar by toolbar, until the screen was a wall of controls, tables, and settings that looked a decade behind the AI it fed. The one thing the app exists to do — take a clear picture of an oocyte or embryo — was crammed into a small preview in the corner, surrounded by everything else.
The clutter wasn't only cosmetic. Capture happens under real time pressure: every second an egg spends outside the incubator is a second a lab works to minimize. A dense interface taxes exactly the moment that needs to be fastest and calmest — and it made the product harder to learn for every new clinic onboarding to it.
Eight-plus conversations at the bench
Before redrawing anything, I ran discovery interviews with 8+ representative embryologists — internal and external — to learn how capture actually happens: the sequence, the friction, the workarounds, and the things they'd never give up. I mapped the end-to-end workflow from those sessions, which made the priorities legible: the core capture loop dominated real usage, while much of the visible UI served edge cases that didn't need to live on the main screen.
Two findings anchored the redesign. First, lab efficiency is the metric that matters — specifically, minimizing the time eggs spend outside the incubator — so the capture loop had to get genuinely faster, not just better-looking. Second, the existing mental model was an asset: hundreds of clinics already knew this app. The job was to streamline and modernize that model with a scalpel, not replace it with something theoretically nicer that would strand power users.
Make the image the interface
The redesign gives the image the screen. The live oocyte or embryo fills the canvas, capture becomes a single obvious action, and everything else — settings, secondary tools, less-used controls — was regrouped semantically and moved into accessible, out-of-the-way places. It reads as calm and modern without changing the fundamental flow embryologists already had in their hands.
Quality is surfaced where the decision is made. Because capture quality gates the downstream AI — there's a minimum threshold for highlights, shadow, and focus — the redesign makes quality control glanceable: a simple green / amber / red read on each image, so embryologists get it right the first time and hand a clean image to the reporting pipeline. The product identities (VIOLET, MAGENTA, ROSE) and a hand-tuned light/dark theme carry through from the report system, so the whole suite feels like one product.
The decisions that mattered
Three calls shaped the redesign — each a deliberate trade-off in a tool hundreds of clinics already depend on.
- 01
Streamline the existing paradigm — don't reinvent it
A guided, wizard-style flow was tempting and might have demoed beautifully. But the app's existing mental model was a hard-won asset across hundreds of clinics worldwide; re-flowing it would have stranded power users to court newcomers. I chose scalpel-precision: keep the paradigm embryologists already know and modernize it underneath — faster for experts, easier for the next clinic onboarding, jarring for no one.
- 02
Decide what earns the screen
Modernization here was editing, not decoration. Using the discovery findings, I ranked what mattered moment-to-moment and gave it prominence — the image, the capture action, the quality read — then regrouped everything else into accessible menus. Showing less at once is a real cost for the occasional power-user task; the bet was that focus on the core loop serves the lab, and onboarding, far more than density does.
- 03
Optimize for incubator time, and catch quality at the source
The north-star constraint isn't clicks — it's how long an egg is out of the incubator. So the capture loop was tuned to be fast and unambiguous, and image quality was made glanceable (green / amber / red on highlight, shadow, and focus) right at capture. Getting quality right upstream is what lets a clinic generate the reports they need downstream without rework — Capture and the Cloud reporting app are two ends of one pipeline.
A trusted tool, modernized — without breaking it
The redesign rolled out to clinics with the embryology team leading adoption, and early response has been strongly positive — the clinics love it. The modernized experience lifted the perception of the product and eased onboarding for new clinics, all while preserving the workflow labs already relied on and catching image quality at the source for a cleaner handoff into the reporting pipeline. Harder numbers (post-rollout NPS) are being gathered.
- 8+
- Embryologists interviewedInternal & external, in discovery
- Hundreds
- Clinics served worldwideThe global install base whose mental model the redesign preserved
- Loved
- Response at rolloutQualitative, per the embryology rollout team; quantitative NPS in progress
