The Molecule Insights Research Assistant (MIRA) Stack

Project Evaluation

Separate from everything above, we’ve built a standalone tool that reads a project’s data room files and answers a question that funders and science teams both care about: where is this project actually at, and what needs to happen next?

This will be displayed on project pages as a ‘Technology Readiness Score’ (TRL); a system invented by NASA and now widely applied to technological inventions. Projects have the opportunity to improve their scores by uploading relevant documentation.
It works as a four-step process. Each step builds on the previous one, and you can run the whole thing in a single session once the relevant documents are available. View the GitHub repository.

Step 1: Where is the science?

A quick read of the project that places it at the right level of research maturity: early idea, early evidence, or approaching clinical readiness, based on what the documents actually show, not how the team describes it. This takes about five to ten minutes and sets the frame for everything that follows.
Output: TRL 1, 2, or 3 classification

Step 2: How strong is it?

This is where the framework does most of its real work. Once a project has been placed at the right TRL, it gets scored against a set of criteria matched to that stage, and each criterion carries a weight that reflects how much it actually matters at that point in a project's life. The weights are not uniform - what separates a strong early-stage project from a weak one is different from what separates a strong late-stage project from a weak one, so the scoring has to reflect that.
Output: Score (0–5) with a breakdown by criterion

The rationale behind these weights is that at the early feasibility stage, the single most important question is whether the mechanism actually works in a lab, which is why therapeutic relevance carries 45%. Hence, a project can have a compelling IP strategy and a wide range of potential indications, but none of that matters if the experimental data doesn't support the underlying hypothesis!

Utility of candidates sits at 20% because identifying specific molecular entities with reproducible activity becomes meaningful at this stage, although it is still secondary to whether the mechanism holds up.

IP position and safety are both weighted at 15%, reflecting that they become consequential by the time a project reaches TRL 2, without yet being decisive.

Therapeutic optionality is deliberately held at 5%, because multi-indication potential is largely speculative this early and over-weighting it would reward storytelling over evidence.

Step 3: What should happen next?

The score maps to a clear recommendation: move forward, hold and reassess, change direction, or stop, directed at potential funders and project evaluators. Each outcome comes with a written rationale grounded in the scoring from Step 2, so the decision has something concrete behind it rather than being a gut call.
Output: Promote / Supplement / Pivot / Pause + written rationale

Step 4: What exactly needs to be done? Optional module.

When a team needs more than a recommendation, this step produces a practical advancement plan built around the criteria where the project scored lowest.

It covers: the specific experiments or milestones needed to reach the next TRL level, what resources and expertise are required, rough timelines, the biggest risks and how to mitigate them, and suggested partners or collaborators where relevant. If projects upload the relevant data, their TRL score improves.

The output is structured so it can go directly into a grant application, an investor update, or the team's internal planning.
Output: Action plan with tasks, resources, timelines, and risks

Adding MIRA to Claude

MIRA can be connected to any Claude project, which means everyone working in that project gets access to all the live tools above without any extra setup per conversation. Setup instructions and the current configuration are in the mira-knowledge repository.

Try MIRA

MIRA is live and ready to use. Head to molecule.xyz to ask it about any project in the ecosystem, explore tokens by research area, or get a full briefing on a specific IPT. For project teams, uploading documentation to your data room is the fastest way to improve your TRL score and unlock the full evaluation stack.

Technical details

These are the tools available through the MIRA server today. Each one is a specific thing MIRA knows how to do - the AI figures out on its own which one to use based on what you ask.