ChatGPT crowns JuliaHub #1.
Claude only gets there 40% of the time.
20-query audit across ChatGPT, Gemini & Claude. JuliaHub (Dyad) is cited on 34 of 60 responses (56.7%). But the story isn’t the composite — it’s the 35-percentage-point gap between ChatGPT (75%) and Claude (40%). Claude knows JuliaHub, but lags significantly behind ChatGPT on the queries that matter most to buyers.
This report was generated using Xtrusio, an AI visibility and demand intelligence platform that analyzes how companies appear across modern AI systems such as ChatGPT, Gemini, Claude, Perplexity, and other generative engines.
The insights in this page are generated using Xtrusio’s proprietary research and content intelligence framework.
JuliaHub is a two-speed company in a three-platform world.
On ChatGPT, JuliaHub is crowned #1 on 8 of 20 queries — the dominant choice for AI-native physics simulation. On Gemini, it holds solid mid-pack presence with 4 #1s. But on Claude — the platform used by millions of engineers and researchers — JuliaHub is cited on only 40% of queries vs ChatGPT’s 75%. Critically, Claude misses JuliaHub entirely on its two most important buyer queries: “AI agent for physics simulation” and “plain-English modeling tools.” The May 2026 Dyad 3.0 GA announcement has reached ChatGPT’s index but not Claude’s. That 35-point gap costs discovery every day it goes unclosed.
Platform Scorecard
JuliaHub citation rate across AI platforms — and how the competition stacks up
AI Visibility Leaderboard
Who owns the AI conversation in engineering simulation — citations across all 60 responses
AI Positioning Audit
20 buyer-intent queries — click any row to see the exact question
Each query was written from the perspective of a real engineering decision-maker evaluating physics-based simulation and model-based design tools. These personas represent the buyers whose AI search results determine whether JuliaHub gets discovered — or whether Simulink, Dymola, and Ansys win the conversation by default.
| # | Query Topic | Cluster | ChatGPT | Gemini | Claude |
|---|---|---|---|---|---|
| 1 | NL → Physics AI Agent | AI-Native | ✓ | ✗ | ✗ |
Exact question asked across all AI platforms: “Is there an AI agent that can write physics-based simulation models from natural language descriptions?” | |||||
| 2 | Plain-English → Validated Model | AI-Native | ✓ | ✗ | ✗ |
Exact question asked across all AI platforms: “What modeling tools let me describe a mechanical or thermal system in plain English and get back a validated simulation model with derived equations?” | |||||
| 3 | AI Discovers Missing Physics | AI-Native | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “Are there simulation platforms that use AI to help discover missing physics in my system model from experimental test data?” | |||||
| 4 | Best Modern Modelica Alternative | Acausal | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “What is the best modern alternative to Modelica for acausal multi-physics modeling in 2026?” | |||||
| 5 | Physics + NN Surrogates Unified | AI-Native | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “Which simulation tools combine first-principles physics models with neural network surrogates in a single unified framework?” | |||||
| 6 | Fastest Scientific Computing Platform | Platform | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “What programming language or platform is fastest for large-scale scientific computing and engineering simulation — Python, MATLAB, or something else?” | |||||
| 7 | Unit Consistency & Conservation Laws | Acausal | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “Are there modeling languages that enforce unit consistency and conservation laws automatically at the compiler level before simulation runs?” | |||||
| 8 | Git-Based Model Version Control | Workflow | ✓ | ✓ | ✗ |
Exact question asked across all AI platforms: “What modeling platform should our engineering team use if we want true git-based version control and code review for our simulation models?” | |||||
| 9 | Model-Based Design Leaders 2026 | MBD | ✓ | ✓ | ✗ |
Exact question asked across all AI platforms: “What are the leading tools for model-based design and system-level simulation in 2026?” | |||||
| 10 | Digital Twin Platform | Digital Twin | ✓ | ✓ | ✗ |
Exact question asked across all AI platforms: “What’s the best platform for building physics-based digital twins of industrial equipment that can update with live sensor data?” | |||||
| 11 | Automotive OEM Powertrain Simulation | Vertical | ✗ | ✗ | ✗ |
Exact question asked across all AI platforms: “What software do automotive OEMs use for powertrain, vehicle dynamics, and thermal management system simulation?” | |||||
| 12 | Aerospace GNC Development Tools | Vertical | ✗ | ✗ | ✗ |
Exact question asked across all AI platforms: “What tools do aerospace companies use for developing guidance, navigation, and control (GNC) systems for spacecraft and aircraft?” | |||||
| 13 | AI for Hardware Engineering | AI-Native | ✓ | ✗ | ✓ |
Exact question asked across all AI platforms: “How are engineering companies using AI to accelerate hardware engineering, simulation, and model-based design workflows?” | |||||
| 14 | Cloud GPU / HPC for Engineering | Cloud HPC | ✗ | ✗ | ✗ |
Exact question asked across all AI platforms: “What’s a good cloud-based platform for running large-scale engineering simulations and scientific computing workloads on GPUs?” | |||||
| 15 | Multi-Domain Acausal Modeling (E/M/H/T) | Acausal | ✓ | ✓ | ✗ |
Exact question asked across all AI platforms: “What multi-domain modeling tools can handle electrical, mechanical, hydraulic, and thermal systems together in one acausal model?” | |||||
| 16 | Best Simulink Alternative | MBD | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “What is the best alternative to MATLAB Simulink for system-level simulation and model-based design?” | |||||
| 17 | Simulink / Modelica Migration | MBD | ✓ | ✓ | ✓ |
Exact question asked across all AI platforms: “Which modern simulation platforms integrate well with existing Simulink and Modelica workflows during a migration?” | |||||
| 18 | HVAC / Refrigeration Simulation | Vertical | ✗ | ✗ | ✗ |
Exact question asked across all AI platforms: “What is the best software for HVAC system modeling, refrigerant cycle simulation, and building energy analysis?” | |||||
| 19 | DO-178C / ISO 26262 Code Gen | Safety | ✗ | ✗ | ✗ |
Exact question asked across all AI platforms: “What modeling tools can generate DO-178C and ISO 26262 compliant embedded C code from system-level models for safety-critical applications?” | |||||
| 20 | Mature Multi-Physics Component Libraries | Libraries | ✓ | ✗ | ✗ |
Exact question asked across all AI platforms: “Which simulation platforms have the most mature component libraries for aerospace, automotive, and industrial multi-physics applications?” | |||||
| TOTAL | 15/20 (75%) | 11/20 (55%) | 8/20 (40%) | ||
The Claude Gap
Where JuliaHub loses 35 percentage points vs ChatGPT — and why it matters
The gap between ChatGPT (75%) and Claude (40%) isn’t a minor measurement variance. It’s a 35-point structural divergence that traces directly to a single cause: Claude’s training data hasn’t absorbed JuliaHub’s commercial press, product launches, and rebranding from JuliaSim to Dyad. While ChatGPT indexed the May 2026 Dyad 3.0 GA announcement and the Ansys TwinAI partnership, Claude returned academic papers and competitor tools on the same queries.
“Is there an AI agent that can write physics-based simulation models from natural language descriptions?”
“What modeling tools let me describe a mechanical or thermal system in plain English and get back a validated simulation model with derived equations?”
“What are the leading tools for model-based design and system-level simulation in 2026?”
Dyad 3.0 went GA in May 2026. ChatGPT indexed it. Claude partially has — but not on the queries that matter most. On Q1 and Q2, the two queries that define Dyad’s entire positioning, Claude returned academic papers and competitor tools instead of JuliaHub. The fix isn’t more features. It’s amplification on sources Claude indexes: major business press, Wikipedia, structured product documentation, and third-party editorial that references “Dyad” specifically — not just “Julia” or “SciML.”
AI Topic Authority Map
Which JuliaHub product lines are visible to AI — and which are invisible
Each row maps to a JuliaHub revenue product line. Each cell shows whether AI cited JuliaHub on queries testing that product line. Green = cited and leading. Yellow = cited but mid-pack. Red = cited but buried. Grey = not cited. The pattern tells you where JuliaHub’s AI authority is real — and where it’s missing entirely.
5 queries
8 queries
3 queries
2 queries
▹ Cloud HPC / JuliaHub is the only product line with zero visibility across all three platforms.
Methodology
How we conducted this Xtrusio AEO/GEO Audit
This research is based on Xtrusio’s proprietary AI visibility analysis framework.
Recommendations
What JuliaHub needs to do to close the Claude gap, capture cloud HPC, and amplify vertical case studies
The audit identifies three distinct problems requiring three different interventions. Phase 1 addresses the most urgent and highest-impact gap — closing the Claude visibility deficit on Q1 and Q2. Phase 2 extends authority into the cloud HPC category where JuliaHub is currently invisible. Phase 3 converts marquee customer success into vertical-specific AI citations that engineers actually search for.
- Wikipedia: Create or substantially expand the “Dyad (software)” Wikipedia article. Claude’s web search prioritizes Wikipedia for named commercial tools. Currently Dyad has no Wikipedia entry; the old JuliaSim article redirects to JuliaHub. A dedicated Dyad article with product description, launch date (May 2026), and use cases will directly address Q1 and Q2 misses.
- Business press amplification: The Dyad 3.0 GA announcement and $65M Series B were covered by ChatGPT’s sources but not Claude’s. Pitch press coverage to VentureBeat, TechCrunch, The Register, and IEEE Spectrum — sources Claude’s search index reliably ingests.
- Structured product docs: Publish a “What is Dyad?” page on juliahub.com with explicit language: “Dyad is an AI agent for physics-based simulation that converts natural language descriptions into validated engineering models.” One sentence. Exact query match to Q1 and Q2.
- Protect unanimous #1 territory: Q3 (missing physics), Q4 (Modelica alternative), Q5 (physics+ML surrogates) are clean #1s across all three platforms. Publish quarterly SciML blog posts that reinforce these exact phrases in their H1 headlines to defend the position.
- Q14 is a clean 0-of-3 miss — all platforms. JuliaHub-the-cloud-platform doesn’t appear when buyers ask about cloud GPU / HPC for engineering simulation. Rescale, CoreWeave, SimScale, and AWS dominate. Publish a dedicated “JuliaHub Cloud vs Rescale” comparison page and a “Scientific HPC on JuliaHub” landing page explicitly targeting the query: “cloud platform for large-scale engineering simulation.”
- Ansys TwinAI partnership content: The Ansys co-branding deal was cited on ChatGPT (Q5, Q4) but only partially surfaced. Publish a joint case study or co-authored technical post with Ansys that explicitly names JuliaHub as the cloud compute layer — this surfaces JuliaHub on digital twin and cloud HPC queries simultaneously.
- NASA / ASML benchmark content: NASA 15,000x speedup and ASML 50% cycle time reduction are extraordinary numbers. Neither surfaces on any cloud HPC query across any platform. Convert these into “JuliaHub HPC vs on-premise” content with headings like “Running large-scale simulation on JuliaHub Cloud” — exact match to Q14 buyer intent.
- Q11 automotive — zero citations across all platforms. Stellantis VP John Alexander (P1) is exactly the buyer being lost. Boeing GNC 2yr vs 5yr development cycle, Williams Racing 50x accuracy, Instron 500x speedup — none of these convert into automotive or aerospace citations. Commission third-party editorial (not case studies) in Automotive Engineering International, Aerospace America, or SAE publications that name-drops Dyad in vertical context.
- Q18 HVAC — MERL case study not surfacing. Mitsubishi Electric HVAC with <2% error rate is a compelling result. It doesn’t appear on any HVAC simulation query across any platform. Publish a MERL-attributed technical post on juliahub.com with H1: “HVAC Refrigerant Cycle Simulation with Dyad” and submit to ASHRAE Journal or Buildings journal for editorial indexing.
- Pumas standalone audit consideration: Eli Lilly Sr. Director Andrés Olivares-Morales (P3) represents a buyer segment — pharmaceutical pharmacometrics — that was not tested in this 20-question audit. JuliaHub’s Pumas product has a separate AI footprint. Consider a dedicated 20Q Pumas audit targeting PK/PD, MIDD, population pharmacokinetics, and clinical trial simulation queries.
Ready to close the Claude gap?
JuliaHub is cited on 40% of Claude queries. Let’s make it 70%.
This research report was generated using the Xtrusio Company Intelligence Module.
