ResearchOS

The description states that ResearchOS is a system that turns a research question into a structured, verifiable record across five surfaces:

1. Literature — GPT-5.6 reads uploaded papers and extracts claims, each pinned to an exact quoted span from the source.
2. Knowledge Graph — verified claims become an evidence graph of materials and properties; numeric disagreements surface as red dashed contradiction edges.
3. Hypotheses — drafted only from verified evidence, and required to be measurable by an experiment the system can actually run.
4. Experiment Planner — runs real physics: an ASE + MACE-MP-0 substitution-energy screen on CPU, not a mock.
5. Research Replay — because every action is an event, the entire project replays from the first question, with provenance behind each step.

The core rule is that AI proposes (amber), humans verify (green), and only verified evidence counts. There is no free-text chat window; the model produces bounded, structured proposals.

Inference This is a research workflow automation tool with strong emphasis on verifiability and reproducibility through structured AI outputs and human validation.

The author positions ResearchOS as a system that moves beyond AI fluency to focus on provenance and verification, where trust comes from traceability, not just language generation. The product is described as:

• Not a chatbot or general-purpose AI assistant.
• A tool for structured research workflows with human-in-the-loop validation.
• Designed to support scientific rigor in hypothesis formation, experiment planning, and result interpretation.

Inference The positioning has evolved from a generic AI-powered research assistant to a structured, evidence-based research platform that emphasizes reproducibility and scientific rigor.

The description does not explicitly state who the target customer or ideal customer profile (ICP) is. It implies use by researchers working in domains like materials science, particularly hydrogen storage, but no names, roles, or organizational affiliations are mentioned.

Inference The likely ICP includes researchers or research teams working in materials science or related fields, especially those needing structured workflows for hypothesis testing and reproducibility.

No evidence of a business model or pricing structure is provided. The description does not mention monetization, licensing, subscriptions, or any commercial arrangement.

Inference There is no evidence of a business model or pricing strategy. This is likely an early-stage prototype or proof-of-concept.

The system is built using:

• AI stack: GPT-5.6 (with multiple specialized models), OpenAI API, Codex.
• Backend: FastAPI with async SQLAlchemy, PostgreSQL 16 + pgvector, Redis with arq for task queuing.
• Frontend: Next.js with sigma.js graph and SSE event stream.
• Compute: MACE-MP-0 substitution-energy screen on CPU using ASE.
• Domain knowledge: Encoded in a “hydrogen-storage domain pack” that includes property registry, unit aliases, comparability transforms, and conflict rules.

Inference The system is built with a modern, scalable stack for AI + data processing and scientific simulation. It shows technical maturity in handling structured outputs, grounding, and domain-specific logic.

There is no evidence of traction or adoption beyond the author’s own development. The team size is listed as 0, and there are no mentions of customers, users, revenue, or usage metrics.

Inference This is an early-stage prototype or proof-of-concept. There is no evidence of product-market fit, user feedback, or real-world deployment.

The description does not mention any competitors or direct market comparisons. It focuses on the unique aspects of ResearchOS (human-in-the-loop, grounding, conflict detection) but does not situate itself in a competitive landscape.

Inference There is no evidence of competitive positioning. It’s unclear whether this product addresses an unmet need or competes with existing tools in the space.

1. No traction or adoption: The project is described as self-built by one person, with no team or users.
2. Unproven business model: No evidence of monetization or revenue streams.
3. High technical complexity without real-world validation: The system uses advanced AI and scientific simulation but lacks real-world testing or feedback.
4. Self-reported only: All claims are unverified; there is no third-party corroboration.
5. No domain-specific commercial use case: While it includes a hydrogen-storage pack, there’s no indication of broader application or market demand.

1. What is the actual research domain you're targeting beyond hydrogen storage?
2. Have you tested this with real researchers or labs? What feedback did you get?
3. How do you plan to scale the human verification process for larger datasets?
4. Are there any plans for monetization or commercial partnerships?
5. What are the key technical challenges that remain unresolved in production use?