Qiskit Fall Fest 2026, hosted by QuantumXRegister now
DsynQ Research Preview - beta version, for research and evaluation purposes.
AI-Powered Design for Quantum Photonics

Physics-constrained AI for quantum photonic circuit design

Transform photonic Quantum device requirements into fabrication-ready chip designs using AI-assisted workflows.

◈gdsfactory layout engine◇AI intelligence layer

DsynQ Research Preview

This platform is currently a beta version and intended for research and evaluation purposes. Generated designs, simulations, and AI analysis are provided as engineering aids and should be validated before fabrication.

◈How it works

From requirement to tape-out in four steps

Step 01

Define Requirements

Describe your device in plain language or structured fields - wavelength, platform, objective.

Step 02

AI Design Assistant

The AI engine translates intent into a validated, structured photonic design specification.

Step 03

Photonic Layout Generation

gdsfactory synthesizes real chip geometry with an interactive SVG preview.

Step 04

Export Manufacturing Files

Download fabrication-ready GDSII, SVG, JSON, and an engineering report.

◈Supported components

The building blocks of photonic circuits

Beam Splitter

Available

Directional 1×2 power splitter

Splits an input mode into two outputs at a configurable ratio. The workhorse of interferometric quantum photonic circuits.

Waveguide

Available

Single-mode routing element

Low-loss channel that confines and routes light across the chip. The foundation every other component connects to.

Directional Coupler

Available

Evanescent 2×2 coupler

Two waveguides brought close so light couples between them - the basis for tunable splitting and switching.

Grating Coupler

Available

Fiber-to-chip interface

Diffractive grating that couples light between an optical fiber and an on-chip waveguide for I/O.

Mach-Zehnder Interferometer

Available

Tunable interferometric switch

Two splitters bridged by two arms - the programmable unit cell of quantum photonic processors.

◈See it in action

From requirement to a reviewed, simulated design

Representative examples of DsynQ output

50:50 Beam Splitter

Silicon photonics - 1550 nm - balanced split

Efficiency
96.4%
Insertion loss
0.18 dB
Bandwidth
42 nm

Mach-Zehnder Interferometer

Silicon nitride - 1310 nm - max efficiency

Efficiency
94.1%
Insertion loss
0.31 dB
Bandwidth
35 nm

Grating Coupler

Silicon photonics - 1550 nm - max coupling

Coupling
-2.9 dB
1 dB band
48 nm
Fab score
88 / 100

Example Simulation

Analytic-FDTD - power budget at 1550 nm

Transmitted96%
Reflected2.4%
Radiated1.6%
Estimate vs simulation error1.3 pts

Example Candidate Comparison

Three evaluated variants, ranked

Cand.ScoreEff.Loss
ABEST0.9296.4%0.18 dB
B0.8895.1%0.14 dB
C0.8593.7%0.22 dB

Example AI Design Review

Grounded, structured - quotes only figures from the design

“Candidate A meets the balanced-split objective at 96.4% efficiency with 0.18 dB insertion loss across a 42 nm band. The design is robust to typical fabrication tolerances (fab score 88/100).”

Risks

  • ◆ Coupler gap sensitive to +/-10 nm lithography drift
  • ◆ Bandwidth narrows at the band edges

Next steps

  • ◆ Sweep coupler length for tolerance margin
  • ◆ Confirm with a full-wave simulation before tape-out
◈Benefits

An AI-native photonic design platform

Faster Design Iteration

Go from requirement to layout in seconds, compressing days of manual CAD work.

AI-Assisted Engineering

An intelligence layer that recommends parameters and writes your engineering summaries.

Fabrication-Ready Outputs

Standards-compliant GDSII streams that drop straight into your tape-out flow.

Quantum Hardware Focused

Built around the interferometric building blocks of quantum photonic processors.

Design your first photonic chip today

No sign-up required - generate a fabrication-ready layout in under a minute.

Start Designing