Wafer — a Wellspin case study
Wafer builds AI infrastructure for faster, more efficient open-source LLM inference. Raised $40M in Series A. Website live in ten days, then the console and the Series A deck.
- ✦Client
- Wafer (YC S25)
- ✦Stage
- Series A ($40M)
- ✦Industry
- AI infrastructure
- ✦Type
- Design Partnership
- ✦Deliveries
- Website, product design, pitch decks & marketing assets
## Story
Wafer (YC S25) builds AI infrastructure that makes open-source LLM inference faster and more efficient.
We work with Wafer as a design partner across its product and its marketing. Our work spans the website with its technology and case-study pages, the console at app.wafer.ai, investor and customer decks with technical diagrams, and sales documents, social assets and merch.
## What we did
Website live in ten days, then the console and the Series A deck.
## What Wafer says
We first brought in Wellspin to design and build wafer.ai, and they went from kickoff to a live site in just ten days. Since then they’ve redesigned our console at app.wafer.ai and worked with us on our Series A deck.
Same pattern every time: they understood the technical story without hand-holding, pushed back when something oversold, and shipped code directly. If you’re a technical founder who needs design that actually ships, hire them.
## The work · 17 slides
The case page shows the work as slides, one after another: live pages, films and stills. Each one, in order:
- ✦
01 · Logo
The Wafer wordmark in white on black, over a field of dithered pixel clouds.
- ✦
02 · Landing hero
The wafer.ai homepage hero: “Inference that keeps getting better” over pixel-art clouds, a performance line rising across the page and a Start Building button.
- ✦
03 · Palette
Wafer’s colour palette: sage, sand and white, then three near-black greys.
- ✦
04 · Never finished optimizing
Feature cards from the website: “Tailored to your traffic” (a tokens-per-second gauge) and “The whole stack searched” (model, engine, kernels and hardware tuned together).
- ✦
05 · Data visuals
Three technical diagrams in thin white lines on black, made for the site and the decks: how a deployment trades throughput against latency as it is tuned.
- ✦
06 · Case — Neon
A customer case-study page on the Wafer website: Neon, “Voice agents with no awkward pauses” — the overview, the solution, and a chart of client-observed time to first token cut from about 800 ms to about 550 ms.
- ✦
07 · Mobile screens
The website on phones: the homepage hero and the Wafer Manifesto page.
- ✦
08 · Onboarding
The Wafer console (app.wafer.ai): the model catalogue with input and output prices per model, the first-request steps, the account balance and the user menu.
- ✦
09 · Laptop login
The console’s sign-up screen on a laptop: “Welcome to Wafer” with GitHub and Google sign-in, over a pixelated classical seascape.
- ✦
10 · API Keys
The console’s API Keys page with the “User API key created” dialog that shows a new key once, to copy.
- ✦
11 · Team invite
The console’s invitation screen: “You’ve been invited to join a team”, with Accept and Decline.
- ✦
12 · Teams
The console’s Teams page: a table of members with their keys, join dates and roles.
- ✦
13 · Usage
The console’s Usage dashboard: tokens, spend, requests and average request cost for the month, a daily token chart and spend by model.
- ✦
14 · Manifesto
The Wafer Manifesto — a long-form page on efficiency and compute — on a laptop, over a pink evening sky.
- ✦
15 · Posters
Tall banners: “Your models. Your traffic. Your stack. Always improving.”, “Always getting better” and “Inference that keeps getting better — 2T+”, with QR codes to wafer.ai.
- ✦
16 · Stickers
Wafer stickers on a laptop lid: a layered chip (“layer by layer”), a stopwatch (“µs matter”), a dartboard (“cache hit.”) and calipers.
- ✦
17 · Posters — light
Three print posters on a light wall: a wireframe globe, “Built layer by layer” (a stacked server) and “Built to keep going”.
## More projects
Alinea is an AI-powered investing app for the next generation, with 2M+ downloads. Series A ($10.4M) + $22.5M.
- ✦
Omen
Omen puts stocks, perps, prediction markets and AI agents in one fast, institutional-grade trading app.
## Work with us
## For agents
Client: Wafer builds AI infrastructure for faster, more efficient open-source LLM inference. Raised $40M in Series A.
Studio: https://wellspin.design/index.md
Canonical HTML: https://wellspin.design/case/wafer/
Markdown: https://wellspin.design/case/wafer/index.md
Index: https://wellspin.design/llms.txt
Last updated: 2026-11-10