Arena Physica

Arena Physica · Launch Video Breakdown: Hook, Pacing & Motion Design

Advance humanity’s rate of hardware innovation

0:00 · The Hook · Introducing Heaviside 1: The EM Foundation Model
0:00 / 0:00
AI AgentsProductionSeptember 1, 2026

Founder Authority & Product Capability Reveal

9

Cuts / min

2:35

Runtime

22K

Views

135

Likes

27

Reposts

13

Replies

Scene-by-scene timeline & spoken transcript

  1. The Hook

    Introducing Heaviside 1: The EM Foundation Model

    “Hi, I'm Mike, Chief Research Officer at Arena Physica. Today, I'm excited to introduce Heaviside 1, our next-generation EM foundation model. Heaviside 1 takes 3D geometries as input and makes full electromagnetic field predictions in 3D space.”

    On screen
    Michael Frei, Ph.D. CHIEF RESEARCH OFFICER Arena Physica Arena Physica ATLAS FIELDS STUDIO Port: Board: Signal: Dielectric: Ground: 50 Ω 8 mm x 8 mm Copper - 0.035 mm RO4003C - 0.2032 mm Copper - 0.035 mm PREDICTING
    Camera
    Alternates between a static medium shot of the speaker and dynamic 3D renders of electromagnetic simulations, with the camera panning and rotating around the simulated geometries.
    Motion
    Cross-dissolves between speaker and product UI/3D renders. UI elements animate in with subtle fades and slides. 3D simulations feature particle systems representing field vectors.
  2. Problem Agitation

    The Challenge of EM Simulation & Verification

    “This is a dramatic improvement over our previous model, Heaviside 0, which only predicted S-parameters. In software engineering, we have effective verifiers. In electromagnetism, it's much harder. Heaviside 1 addresses this by predicting full 3D near fields across space, representing a richer training ground for our models.”

    On screen
    test_filters.py RUNNING test_cutoff_of_known_lc test_insertion_loss_half_amplitude test_lossless_pass_is_zero_db test_passband_edge_is_inclusive test_stopband_excludes test_cutoff_scales_inverse_sqrt running pytest 8.4 cutoff frequency of this geometry insertion loss at half amplitude lossless pass is 0 dB passband edge is inclusive stopband rejection cutoff scaling with inductance SIX QUESTIONS NOTHING TO RUN THEM >250K DIFFERENT GEOMETRIES
    Camera
    Static medium shots of the speaker are interspersed with close-up views of code editor interfaces and abstract data visualizations. The camera in 3D scenes continues to orbit the simulated objects.
    Motion
    Text elements animate in with a typewriter effect or subtle fades. Data visualizations use animated bar charts and grid layouts to represent complex datasets. Transitions are quick cuts.
  3. Product Reveal

    Heaviside 1: Speed, Accuracy, and Richer Data

    “Heaviside 1 was trained on over 250,000 different geometries, each with full 3D near field data. This allows Heaviside 1 to predict full 3D near fields across space, representing a richer training ground for our models. This is a dramatic improvement over our previous model, Heaviside 0, which only predicted S-parameters.”

    On screen
    Arena Physica ATLAS FIELDS STUDIO Port: Board: Signal: Dielectric: Ground: 50 Ω 8 mm x 8 mm Copper - 0.035 mm RO4003C - 0.2032 mm Copper - 0.035 mm 336 ms 10,000 FIELD VECTORS PREDICTED EMVAL-SP OUT OF DISTRIBUTION Heaviside-1 Heaviside-1 (S-only) RMSE WEIGHTED MAE WEIGHTED MAE REAL + IMAG MAGNITUDE IN DECIBELS PHASE IN RADIANS 0.15 1.0 0.20 0.10 0.75 0.15 0.05 0.50 0.10 0.00 0.25 0.05 0.0 0.00
    Camera
    The camera alternates between the speaker and detailed UI demonstrations of the Heaviside 1 model in action. 3D simulations are shown with dynamic camera movements, highlighting the field predictions.
    Motion
    UI elements animate in with precise timing, showcasing data points and performance metrics. Bar charts and graphs dynamically populate with data, emphasizing improvements. Particle systems visualize electromagnetic fields.
  4. Call to Action

    Explore Heaviside 1 Today

    “We're excited for you to explore Heaviside 1 and its features. Visit studio.arenaphysica.com to learn more.”

    On screen
    EM Benchmark EMVAL-SP IN DISTRIBUTION Heaviside-1 Heaviside-1 (S-only) Heaviside-0 RMSE WEIGHTED MAE WEIGHTED MAE REAL + IMAG MAGNITUDE IN DECIBELS PHASE IN RADIANS 0.08 0.6 0.10 0.06 0.4 0.05 0.04 0.2 0.02 0.0 0.00 PUBLIC PRIVATE PUBLIC PRIVATE PUBLIC PRIVATE (STANDARD) (STANDARD) (STANDARD) DESIGN TEMPLATES COMBINE FILTER Arena Physica ATLAS FIELDS STUDIO GAP-COUPLED MICROSTRIP SHARE Port: Board: Signal: Dielectric: Ground: 50 Ω 8 mm x 8 mm Copper - 0.035 mm RO4003C - 0.2032 mm Copper - 0.035 mm Seed 77329 Line width 0.79 mm Finger width 0.20 mm Interclearance 0.22 mm Hairpin spacing 0.35 mm Leg length 2.38 mm PREDICTIONS Field E H S Pi Port 18.00 GHz Freq Mode Vectors Field Lines Color by E Render Low Medium High -511 -321 -105 0 dB S(1,1) -50 -100 18.00 6 GHz 474 ms 10,000 FIELD VECTORS PREDICTED 10,000 field vectors predicted in 474 ms at F = 18.000 GHz studio.arenaphysica.com Arena Physica
    Camera
    The camera continues to alternate between the speaker and detailed UI demonstrations, including interactive elements and benchmark comparisons. The final shot is a static abstract 3D grid with the website URL.
    Motion
    Interactive UI elements highlight features. Data visualizations animate to show comparative performance. The final call to action features a subtle 3D background animation with text fading in.

Related AI Agents Product Launches

Explore all AI Agents launches →
OpenAI
@OpenAI
AI Agents
@OpenAI
OpenAI@OpenAI
Gopuff
@gopuff
AI Agents
@gopuff
Gopuff@gopuff
Elon Musk
@elonmusk
AI Agents
@elonmusk
Elon Musk@elonmusk