Which experiment comes next?
A world of possibilities. Opsight helps scientists decide where to look next.
98,423,325 possible experiments
Loading research data
An interactive star field of 150 model-evaluated photocatalysis experiments chosen by Opsight. Stars range from black for the lowest hydrogen evolution rate to pink for the highest.
Inside the experiment
The 10 inputs of hydrogen production
Hover or tap a star to inspect it. Click a star to keep its details open; press Escape or click empty space to close them. Scroll or pinch over the plot to zoom. Drag to pan when zoomed. With a star focused, use arrow keys to explore experiments, Enter to keep details open, plus or minus to zoom, and 0 to reset the view.
Recorded hypothesis
Preparing the research sequence.
— µmol h⁻¹ · hydrogen evolution
Loading
JavaScript is needed for this animation. The source dataset is available in the research details below.
About this illustration
Each star is one of 150 experiments chosen by Opsight in an optimisation run from our study published in Digital Discovery (2026). The problem, photocatalytic hydrogen evolution, was first studied with our autonomous robotic chemist: ten inputs and 98,423,325 possible experiments. Opsight proposed every experiment with a named hypothesis and learned from each result, reaching 28.37 µmol h⁻¹, the highest rate possible in this problem.
Brighter, pinker stars produced more hydrogen: colour runs from black at the lowest rate to pink at the highest. Star positions trace how the search spread out and then concentrated, rather than coordinates in the ten inputs. The spider chart shows the ten inputs used in each experiment. The opening stars are decorative and are not part of the research data.
Hover or tap a star to see its hypothesis and result. Scroll or pinch over the field to zoom, and drag to explore.
Read the paper
Good experimental evidence is rarely in one place.
Past results, published research and scientific judgement can all inform what to try next.
↗ Past results What happened before “ Published research What is already known ∴ Scientific judgement What might explain it ? Open questions What remains uncertain Opsight Context for
One search. More of the science in view.
Turn research context into an inspectable search.
Choose an experiment from the published run to see its hypothesis, inputs and result together.
001 Baseline
031 New direction
117 Refined search
HYPOTHESIS
Baseline
MEASURED H₂ 9.20 µmol h⁻¹
EXPERIMENTAL INPUTS
A clear thread from the question to the evidence.
Less wasted effort.More useful knowledge.
In a published photocatalysis benchmark, Opsight's hybrid approach improved average best result over Bayesian optimisation alone.
0 10 20 30 µmol h⁻¹
Bayesian optimisation alone
11.6 ± 4.4
BORA hybrid o3 + Bayesian optimisation
25.3 ± 2.9
Cissé et al. · Digital Discovery , 2026 · Table 3 · Read the paper
Privacy notice
Last updated: 10 October 2026
Opsight helps research teams explore what to test next. When you visit our website or start a conversation with us, we aim to be equally clear about what happens to your personal information.
This policy explains what we collect through our public website and inquiries, why we use it and the choices available to you. It does not cover information processed within the Opsight application or datasets shared for a pilot or study. Those activities will be covered by separate information and agreements, as appropriate.
What information do we collect?
When you submit an inquiry, we collect your first and last name, email address and a description of your challenge or use case. You may also provide your role or job title and organization, and tell us whether you are requesting a demo, discussing a pilot or exploring a retrospective study. If you are interested in a retrospective study, you may provide a short description of an existing dataset.
If you contact us by email or continue a discussion with our team, we keep the relevant correspondence and any additional information you choose to provide.
You do not need to create an account or upload a dataset to make an inquiry. Fields marked as required are needed to respond meaningfully; other fields are optional.
Please do not include confidential datasets, sensitive personal information or unpublished technical details in the inquiry form. If a discussion progresses, we will agree an appropriate way to share further information.
Why do we use your information?
We use inquiry information to respond to you, understand your needs and assess whether there may be a useful fit with Opsight. This may include arranging a demo, discussing a pilot, retrospective evaluation or research collaboration, and managing related follow-up.
Where applicable, we also use relevant information to operate and secure the website, manage our records, meet legal obligations and establish or defend legal claims.
Making an inquiry does not subscribe you to marketing emails. If we offer an optional subscription, its purpose and choices will be explained separately when you sign up.
What allows us to process it?
Our main basis for handling inquiries is our legitimate interest in responding to people who approach us and managing potential business or research relationships. We consider the purpose of the processing, its necessity and the impact on your rights and interests.
Where applicable, we may instead process information to take steps at your request before entering into a contract, or to perform a contract with you. We may also process information to comply with legal obligations.
Where an activity requires consent, we will request it separately. You can withdraw that consent at any time; this does not affect the lawfulness of processing carried out before withdrawal.
How long do we keep it?
We keep personal information only for as long as it is needed for the purposes described in this policy, taking account of any continuing relationship and applicable legal requirements.
If an inquiry develops into a pilot, study or other agreement, relevant records may need to be retained under the arrangements for that relationship. We will delete or anonymize information when it is no longer needed, subject to any necessary legal retention.
If you have a concern
You are welcome to contact us with any concern about how we handle your information. You also have the right to request deletion of information or restriction of its use.
Other websites and policy updates
Our website may link to research publications, event pages or other external websites. Those sites have their own privacy practices, which are not covered by this policy.
We may update this policy when our website or information-handling practices change. The latest version will appear here with its publication date. Where required, we will bring material changes to your attention.
