F13 raises $5M to make AI-generated charts answer to the data

Why it matters

F13 is betting that accurate, editable charts and diagrams are a distinct market within AI image generation. Its $5M round funds that test as established design tools and newer vector models pursue similar workflows.

Gregory Janik and Ingimar Tomasson have emerged from stealth with F13 and $5 million in pre-seed funding for AI that generates editable vector graphics. Their wager is that a chart or scientific diagram has to meet a stricter standard than an illustration: its numbers, proportions and labels must survive inspection. Tech.eu reported the round on September 23rd.

Janik, F13's co-founder and CEO, has built software in markets where outputs have practical consequences. He co-founded SaaS-financing business Vitt and led engineering at education startup Knowunity. Brickwise AI lists him as a co-founder and engineering lead. Tomasson also worked at Knowunity and co-authored research published at the 2025 International Conference on Machine Learning on predicting optimal-transport calculations. That paper establishes his research background; it does not establish which methods F13 uses.

F13 is registered in Berlin as F13 GmbH, according to its company imprint, which names both founders as managing directors. The $5 million round was led by Credo Ventures and Point Nine Capital. The named angel participants include Carles Reina of Baobab Ventures and Jack Richardson of mainframe. F13 says it will spend the funding on compute, data and hiring as it brings its first model to market.

An editable image is only the starting point

F13's website shows the kind of work Janik and Tomasson want to take on: an air-quality chart, a unit-circle explanation of sine and cosine, a labeled skeleton and a service-uptime gauge. F13 says its models generate SVGs from text, images or existing graphics. Unlike a flat image, an SVG can be resized and edited as a collection of visual elements. That gives a designer room to correct a label, move a shape or adapt a layout without starting over.

Editability does not make a graphic accurate. An incorrect axis can remain incorrect in a perfectly clean SVG. A skeleton with movable labels can still identify the wrong bone. F13's harder claim is that its models can follow the underlying data and specifications while producing a usable design. Janik told Tech.eu he did not want to build a model that was "impressively expressive and quietly wrong." F13's site makes similarly specific promises about fidelity to data and design requirements. Those are the capabilities on which a professional user would have to judge it.

The proposed uses extend from editorial infographics and educational diagrams to maps, presentations and brand-consistent marketing materials. They do not all present the same test. A marketing layout might be useful after a designer fixes its spacing; a data visualization must also put the right quantities in the right places. If F13 can make that distinction reliably, its developer API could let other software products generate information-bearing graphics within a larger workflow. That is an inference from F13's stated use cases and API plan, rather than a demonstrated deployment.

F13 is in early access through a developer API and a web platform for generating, editing and sharing graphics, according to Tech.eu. Its public website invites visitors to join a waitlist, and F13 plans a broader release later in 2026. The visible examples show what F13 is pursuing, but they do not establish how consistently its model handles unfamiliar datasets or demanding specifications.

The vector market is already occupied

Janik and Tomasson are entering a field with working products. Recraft offers an API for generating and editing vectors, including style controls. Adobe's June Illustrator update added AI-assisted vector creation and editing features. QuiverAI announced an $8.3 million seed round in February for vector-native models and tools; it released Arrow 2 on September 7th with editable SVG output and a focus on diagrams, assets and animation. F13 cannot rely on editable vectors alone to stand apart.

That competition sharpens the purpose of the pre-seed round. F13 says it will invest in its first model and the resources needed to train and deliver it. The business case depends on whether its emphasis on factual and spatial fidelity holds up in real work, especially when a graphic contains figures or technical relationships that a polished result could obscure. The investors have backed that proposition at pre-seed; F13's wider launch will put it in front of users who can check the answer against the brief.