AI agents are already finding ways into quantum computing research. As these results become more common, researchers need ways to verify the results and starting points to understand the ideas that made them possible. At Zapata Quantum, we believe the need for fully leveraging AI and automation for scientific research is best served by a community that can 1) examine discoveries through open experimentation and collaborative discussion, and 2) co-evolve with a publicly shared evaluation infrastructure that is compatible for ingesting automatically generated contents. As AI expands the possibilities for scientific exploration, the way we organize participation and share knowledge should become part of the research itself.

‍

That belief is central to ForShor, a new open challenge Zapata developed in collaboration with the University of Maryland. Using Shor’s algorithm for integer factorization as a prototypical example, ForShor brings AI-assisted research into a well-defined mathematical problem that arises from algorithm implementation, together with formal verification, interactive documentation and direct participation in an evolving research community.

Designing for Participation from the Beginning

An open research effort is only as strong as the decision around motivating the participants. ForShor starts with a specific problem within Shor’s algorithm, providing contributors a shared starting point before branching off with additional refinements. Although the general structure of Shor’s algorithm is well established, finding more efficient ways to implement its components remains an important research task. Different constructions can perform the same computation while requiring different quantum resources.

‍

Our team has identified a well-defined “policy search” problem within the design of recursive arithmetic circuits for Shor’s algorithm (see paper for details). Roughly speaking, each “policy” determines the structure of the recursion for the quantum circuit. Participants can submit candidate policies to the ForShor site and the harness will automatically validate and integrate into the pool of existing solutions.

‍

The framework also makes participation possible without requiring every contributor to reconstruct the entire algorithm and its correctness argument. Participants can focus their efforts on the exposed problem while following the supporting material (paper and doc site) as deeply as they choose.

‍

To facilitate discussion and community interaction, we have also set up a Discord server for anyone interested to join. Top contributions will be recognized as citable artifacts and select participants will be invited to collaborate with us on further dedicated research topics at the intersection of quantum computing and agentic automation.

Building a Shared Repository of Understanding

Formal verification plays an essential role in this environment: it establishes that an implementation satisfies specified mathematical requirements under stated assumptions.

‍

ForShor’s surrounding framework connects the problem exposed to participants with the correctness of the larger algorithm. A proposed replacement must satisfy the conditions required for that connection to hold. This allows researchers to build on the existing formalization when evaluating contributions.

‍

But a proof and an explanation serve different purposes. A result that passes formal verification may still require substantial work before a human can understand its structure or recognize an idea worth extending.

‍

That is why we have designed verification, documentation and community participation together. The formal framework provides a basis for establishing correctness. The supporting explanation of the challenge problems (ForShor site, paper and doc site) intends to help participants understand the constructions submitted by others. The  Discord forum provides participants a place to ask questions, compare interpretations and identify what needs further explanation.

‍

Each supports the others. Questions from participants may reveal where documentation needs more detail. Examining a solution may suggest another problem worth formalizing. A clearer understanding of the algorithm may help the team define a more productive challenge.

‍

We see this relationship as central to useful AI-assisted research. The community helps turn a checked contribution into something other researchers can understand and develop.

Letting the Research Environment Evolve

ForShor is intended to grow over time as new discoveries are made. The challenge site may expand to include additional subroutines or additional variants of the same challenge. This creates a continuing place to follow the work, explore its explanations and understand how successive contributions relate to one another.

‍

In the present state, ForShor contains two challenges (PhaseProduct and PhaseTripleProduct, see the challenge site) concerning subroutines that can be used in two different quantum factoring algorithms (Shor’s and Regev’s algorithm, respectively). The challenge problems, namely policy search, are intentionally narrow to favor clarity of understanding for early participants over breadth of search space to find the best construction. However, over time as the current search space becomes saturated, we may update the harness at the next checkpoint when the current challenges end (see FAQ on the challenge site).

‍

As the community learns, we expect its questions and contributions to inform how the harness, the software environment that guides and evaluates submissions, develops. The initial challenge is planned to run for six weeks, with opportunities for contributors to become more deeply involved with the organizing team at Zapata for further research. We want participation to contribute to a shared scientific outcome, with a community forming early enough to help develop the work.

An Invitation

ForShor gives us a practical setting in which to develop and evolve this open research paradigm with others. Shor’s algorithm is a starting point, and the challenges let participants experience how a research topic can be advanced by open collaboration in the era of AI automation.

‍

The experience will also inform our broader work at Zapata. Formal verification is planned as part of Quantum Pilot™, our platform for quantum application discovery and development. ForShor gives us a place to explore verification techniques openly and learn from their use by a wider research community.

‍

There will be further questions about which problems to expose to agentic AI. Working through those questions together is part of the purpose of this initiative. Visit ForShor.ai to explore the challenge. Follow the paper and proof documentation for the mathematical details, use the available tools to begin participating, or join the community discussion to learn alongside the team.

‍

Bring your agents, your expertise and your questions. Help us make the next discovery something the community can understand and build on.