Apollo did not reach the Moon merely because its engineers could solve difficult equations. It succeeded by turning a distant ambition into a mission architecture of explicit objectives, simulation, verification, and repeated correction. AI now faces a similar transition: frontier models can solve difficult tasks once the problem, tools, and success criteria are specified, yet consequential real-world challenges rarely arrive in an executable or verifiable form. We introduce Apodex Discovery, a framework for building and evaluating discoverative AI through the heavy-duty solver, a system comprising a foundation model, harness, tools, and control policies that pursues extended, stateful, verifiable investigations. It has three core components. First, a problem-scouting process surveyed 561 industries across 16 sectors, assembled 423 high-value real-world problems, and selected 20 for the initial release. Second, a common environment-task-episode abstraction provides data, tools, constraints, feedback, trajectory recording, and verification of intermediate artifacts and final submissions. Third, HDS6 evaluates Tools, Repair, Alternatives, Coherence, Evidence, and Scope independently of final-task success. In AAV capsid design, Apodex surpassed the published state of the art by 7% across viability, tropism, structure prediction, and generative design. In drug repurposing and reformulation, a task-specific biomedical environment improved the mean normalized prediction score of GPT-5.5 and GPT-5.6-sol by 2.5 and 7.6 points over the same closed-book backbone. Controlled ablations show that the fixed TRACES episode interface enables attribution of performance differences to specific solver components. Apodex Discovery moves AI evaluation beyond predefined benchmarks toward verifiable investigations aimed at genuine discovery.
Alex: Welcome to another episode of ResearchPod.
Sam: Today we're looking at a study that asks a pointed question: why can advanced AI systems pass difficult exams, yet struggle when given a genuinely open-ended scientific problem?
Alex: That gap seems important. What's actually going on?
Sam: The short answer is that these models are missing what the researchers call "mission architecture." Think of it like the difference between a student who's good at multiple-choice tests and a scientist who can run a months-long experiment—designing it, troubleshooting it, and knowing when the results are trustworthy. Current AI is much better at the first than the second.
Alex: So the researchers set out to build something that could actually do the second thing?
Sam: Exactly. They created a framework called Apodex Discovery. The name draws deliberately from the Apollo space program—the idea being that genuine discovery requires the same kind of structured mission control that got humans to the moon. You don't just point a rocket and hope.
Alex: So what does that actually look like for an AI?
Sam: Imagine you want an AI to help design a gene therapy treatment. You can't just type the question into a chatbot and expect a useful answer. Instead, the researchers built what they call a "solver"—a complete working system that includes a foundation model, a set of tools it can actually use, a memory layer, and control policies that govern how it behaves. It's less like asking a question and more like giving a scientist a fully equipped lab.
Alex: And the lab matters as much as the scientist.
Sam: That's the core insight. The solver operates inside what they call a "reality-based environment"—think of it like a high-fidelity flight simulator. The AI gets real instruments, real constraints, and real feedback. It can't just hallucinate an answer; it has to interact with the environment and see what happens.
Alex: How do they grade what comes out of that environment?
Sam: They use two complementary checks. The first looks at the final result—did the AI actually produce something that works? The second, which they call process verification, evaluates how the work was done, step by step. That second check matters because in real science, the outcome alone doesn't tell you whether the reasoning was sound.
Alex: How do you grade a process without already knowing the right answer?
Sam: They developed a six-capability rubric they call TRACES—short for Tools, Repair, Alternatives, Coherence, Evidence, and Scope. Each capability gets scored on a scale. So instead of just asking "did it get the right answer?", you're asking: did it use its tools correctly? Did it ground its claims in actual data? Did it consider alternative explanations? It's closer to how a PhD supervisor would evaluate a student's lab notebook than how a teacher grades a multiple-choice test.
Alex: So the quality of the thinking is being assessed, not just the conclusion.
Sam: Precisely. And that rubric does double duty. When a trajectory—a recorded sequence of the AI's actions—falls short, the same evaluation generates a targeted repair note. It tells the system specifically what went wrong, without revealing the hidden answer. The AI then reruns the work with that guidance.
Alex: Does the repair loop actually help?
Sam: Across hundreds of deficient trajectories tested in the study, the repaired reruns scored higher on average. Most improved rather than declined. It's a meaningful result, though the authors are careful not to overstate it—performance still depends heavily on the quality of the verification setup.
Alex: They tested this on a real biological problem, right?
Sam: They did. The domain was adeno-associated virus capsid design—that's the engineering of protein shells used to deliver gene therapy into cells. It's a genuinely hard problem with measurable outcomes. The system, operating inside a domain-specific environment with the right tools and biological constraints, surpassed previous expert methods across several key metrics: viability, structural prediction, and generative design.
Alex: So putting the AI inside a well-built environment with the right tools made a measurable difference to the actual science.
Sam: That's what the paper suggests. The environment isn't just scaffolding—it's doing real work.
Alex: What's the main thing holding this approach back from being applied more broadly?
Sam: Building these environments is extremely labor-intensive. You need domain experts to set up the constraints, the tools, the hidden verification checks—and those checks have to be leak-resistant, meaning the AI can't accidentally see the answer it's supposed to find. Right now, that level of careful engineering limits the approach to a small number of carefully curated domains.
Alex: So the bottleneck isn't computing power—it's human effort in designing the tests themselves.
Sam: For now, yes. Which is a significant constraint. But the authors argue it's a solvable one. As more of these environments get built, the framework becomes more broadly applicable. And when you zoom out, the direction is notable: we're moving toward autonomous systems that can run multi-step research campaigns across disciplines—not just retrieve information, but actually generate and test new knowledge.
Alex: So instead of answering questions, future AI might actually run the experiments.
Sam: That's the long-term trajectory. By combining structured problem framing with rigorous, multi-layer verification, the goal is to push AI past pattern matching into something closer to genuine scientific discovery. The Apodex framework is one early attempt to show what that might look like in practice.
Alex: It's a meaningful step—and an honest one, given how clearly the authors flag what still needs to be solved. Thanks for walking through it.
Sam: Thank you for listening to ResearchPod.