Can this AI predict the side effects of cancer therapies before the first human dose?

Our CEO Fanny Giannou sat down with Startup Magazine after Alithea Bio’s double win at VivaTech Paris — on drug safety, the seed round ahead, and why biology can’t be copied the way software can.

L-R: Fanny Giannou, Prof. Dr. Tim Fugmann .

Not a better filter — a missing dataset

Fanny is clear that HLA-Compass isn’t patching a gap in an existing model — it exists because the underlying data simply didn’t exist before. The HLA peptides Alithea Bio generates in-house, from real human biological material, are exactly the signals the immune system reacts to, at a scale and quality unavailable anywhere else. That data lets drug developers see an individual toxicity risk for a candidate compound in every phase of development, before a trial even starts.

For patients, she argues the impact isn’t just a “better filtered” therapy — it can be the difference between having an immunotherapy option at all, because a risk that used to be estimated can now actually be measured.

Where TGN1412 fits — and where it doesn’t

Asked directly about TGN1412 — the infamous 2006 case where preclinical models completely missed a human immune reaction — Fanny is precise about scope. TGN1412 involved a CD28 superagonist triggering a cytokine storm, a different mechanism than the one HLA-Compass is built for. What Alithea Bio addresses is a related but distinct blind spot: whether a target protein is also presented on healthy human tissue, the core danger for T-cell-based therapies specifically. That’s the gap being closed — not every conceivable form of toxicity.

The biggest lever right now: cancer vaccines

Of everywhere the data gets used — cancer immunotherapy broadly, TCR-based approaches, TCR mimetics — Fanny points to cancer vaccines as the area with the most momentum today. A vaccine only works if it trains the immune system on peptides that are actually sitting on the tumor and not on healthy cells, and the HLA-peptide database is what makes it possible to show, patient by patient, which targets are genuinely tumor-specific.

Two founders, two very different starting points

Fanny doesn’t run the lab — she can’t operate a mass spectrometer, by her own account — and co-founder Prof. Dr. Tim Fugmann is the scientist behind the platform. Her path in was an observation, not a single idea: excellent research often never reaches patients not because the science is lacking, but because research teams rarely know how to turn it into a fundable, market-ready company. She took over as full-time CEO in 2022 to build that bridge.

Bootstrapped to 23 million data points

Alithea Bio has been largely self-funded until now — around €580,000 from grants and smaller programs, with the bulk of growth financed through more than 200 completed client projects and partnerships, including Children’s Hospital of Philadelphia, Deck Bio, and several major pharma companies. With over 23 million immunopeptidomics data points, an in-house lab in Berlin-Buch since 2024, and validated pharma partnerships, the team is now raising its first institutional seed round to take the jump from a profitable data service to a scalable software platform.

The five-year goal, in Fanny’s words: HLA-Compass as the reference clinicians, researchers, and pharma companies all check before capital goes into a trial.

Alithea Biotechnology GmbH · Freiburg & Berlin. Original interview: Startup Magazine, July 31, 2026.