Ai.Qimia
Ai.Qimia is a Swiss cloud-native quantum chemistry platform. Its ConstruQt, ReaQt and InteraQt modules generate physics-grounded molecular ensembles, descriptors and reaction simulations for synthetic and process chemists feeding machine learning pipelines. Access is arranged by contacting the vendor.
What is Ai.Qimia?
Ai.Qimia is a Swiss vendor of a cloud-native quantum chemistry platform delivered as SaaS, summed up on its own pages as physics-based AI for synthetic chemists. Instead of describing molecules through 2D fingerprints or rule-based heuristics, the platform computes molecular ensembles with quantum mechanics and ranks them by Boltzmann-weighted energies, so the numbers a downstream model consumes are grounded in physics rather than in topology.
The offer is organized into three named modules. ConstruQt, the product the home page is devoted to, produces what the vendor calls trusted molecular ensembles: Boltzmann-weighted conformer sets, 2D, 3D and electronic descriptors, and provenance metadata recording how each result was obtained. It aims at reducing label noise in machine learning datasets, capturing the tautomers, protomers and stereoisomers that matter in medicinal chemistry, and detecting inconsistencies in experimental databases automatically. ReaQt extends the approach to complete reaction systems under realistic laboratory conditions; an April 2025 announcement adds multi-phase behavior, catalyst dynamics, temperatures up to 200 degrees Celsius and high-pressure flow. InteraQt maps molecular interactions and transition states, the ground on which mechanisms are elucidated.
The stated use cases run from machine learning input data, experimental quality assurance and model validation through to route optimization, condition screening, learning from negative data - the failed experiments usually thrown away - and scale-up forecasting. The vendor's own framing is that rival tools stop at the synthetic route and leave yield, by-products and scale-up unanswered; it compares them to a road map and its own platform to a GPS. It claims throughput across thousands of cores and millions of molecules per day, and a success rate above 98 percent for its semi-empirical and quantum mechanical calculations. Neither figure is supported by a named customer or a case study.
Integration is described as API-first: a JSON-RPC API, or the vendor's own web workspace where projects, libraries, molecules, reactions and trials are organized. Public material is thin, however - five pages and a single news post, with no API reference, no pricing and no legal documents. The leadership team and an advisory board are named on the About page, and Swiss public-innovation supporters appear as logos on the home page.
What it does
- Generate Boltzmann-weighted molecular ensembles computed with quantum mechanics
- Produce 2D, 3D and electronic descriptors for machine learning and predictive modeling
- Capture the tautomers, protomers and stereoisomers that matter in medicinal chemistry
- Cut label noise in training datasets and flag data inconsistencies automatically
- Simulate complete reaction systems under realistic laboratory conditions with ReaQt
- Map molecular interactions and transition states with InteraQt to elucidate mechanisms
- Screen solvents, catalysts, temperatures and reagents, and forecast scale-up to pilot and production volumes
When to use Ai.Qimia / When not to
A quick filter to help you decide if Ai.Qimia is the right fit.
When to use Ai.Qimia
- Synthetic and process chemists who need forward-synthesis prediction that goes past the route itself to yield, by-products and scale-up behavior
- Medicinal chemists whose work depends on tautomers, protomers and stereoisomers that flat 2D representations quietly flatten away
- Machine learning and cheminformatics teams that want physics-grounded input features and provenance metadata rather than topological descriptors alone
- Data curation and quality assurance specialists cleaning experimental chemistry databases, cutting label noise and hunting inconsistencies
- Computational chemistry groups benchmarking or validating models against quantum-mechanical reference standards, from a pilot set up to millions of molecules
When not to use Ai.Qimia
- Buyers who need a published rate card and self-service checkout: no price of any kind appears on the site, and the only way in is a sales conversation
- Procurement, legal and compliance teams with document requirements: no terms of service, privacy policy, legal notice or data processing agreement is published
- Developers expecting to integrate unaided: the JSON-RPC API is claimed and live, but not one page of API documentation is published
- Chemists and analysts without computational chemistry grounding: SMILES notation, isomer handling and quantum chemistry concepts are assumed from the first screen
- Anyone wanting a mobile app, a browser extension or a ready-made connector: the product is web and API only, and no third-party integration is named
How to use Ai.Qimia
A typical end-to-end flow, from setup to results.
- Read the home page and the process page first, to check that ConstruQt, ReaQt and InteraQt actually cover the chemistry you need to model
- Email info@aiqimia.com to discuss trial licenses and pilot projects; the published contact page carries no form, no address and no email of its own
- Open the application from the Try Now button on the home page, which leads to the sign-in screen at app.aiqimia.com
- Register an account through the application's self-service form, which asks you to accept a terms of use and privacy notice it does not link to
- Create a project and a molecule library to hold the structures you want to study
- Enter molecules in SMILES notation, either by typing them in text mode or by using the built-in drawing editor
- Select the generation options you need: tautomers, conformers, protomers, zwittomers and stereoisomers
- Run the staged procedure, from Stage 0 through Stage 5, and follow each calculation status: in progress, draft, complete, failed or validated
- Review the resulting ensembles, descriptors and provenance metadata, and mark them as validated, or move on to the dedicated ConstruQt and InteraQt screens
- For pipeline integration, call the JSON-RPC 2.0 API at api.aiqimia.com - and expect to agree the details directly with the vendor, since no API documentation is published
Pros & Cons
Pros
- A physics-first method: quantum mechanics and Boltzmann-weighted energy ranking instead of 2D descriptors or heuristics, which is a genuine methodological difference rather than a repackaging
- End-to-end ambition: the vendor claims its simulations reach past the synthetic route to yield, by-products and scale-up, and states on its process page that 90 percent of new molecules are de-risked before scale-up and scale-up time is 40 percent faster
- Negative data - the failed experiments most labs simply discard - is explicitly treated as training signal, which few tools in this space bother to do
- API-first design intended to slot into existing machine learning and cheminformatics pipelines rather than replace them
- Provenance metadata shipped alongside results, which is exactly what dataset curation and audit trails need
- Scale claimed by the vendor: thousands of cores and millions of molecules per day, with a success rate above 98 percent on semi-empirical and quantum mechanical calculations
- Verifiable surroundings: a named leadership team, an advisory board drawn from Evotec, SpiroChem and Qilimanjaro Quantum Tech, and Swiss public-innovation supporters shown on the home page - Innosuisse, Canton de Vaud, Biopole Lausanne and FIT
Cons
- No pricing is published at all: no amount, no tier, no currency, no pricing page, and the probed pricing paths return genuine 404 errors, so budgeting is impossible before a sales call
- No legal document is published: no terms of service, no privacy policy, no legal notice, no data processing agreement - and this is an absence of publication, not proof that such documents do not exist internally
- The application's sign-up form asks users to accept a terms of use and privacy notice that are reachable nowhere on the site or in the application itself
- Nothing is published on data hosting, retention periods, sub-processors, or whether customer submissions are used to train the models
- API-first integration is the central sales argument, yet no API documentation is published and no third-party integration is named anywhere
- The contact page is published but empty: it renders the word Contact and nothing else - no form, no address, no email - although the earliest archived version of the site did carry a working contact form
- The public footprint is very thin: five pages, one news post from April 2025, no named customer, no case study, no testimonial, and headline percentages that are not even rendered in the HTML
Pricing & Plans
No price is published. Ai.Qimia operates no pricing page, and a complete pass over the site, both sitemaps and the web application bundle returned no amount, no currency and no billing unit; the probed pricing and plans paths return genuine 404 errors. No free plan and no free trial are announced either: the home page offers "trial licenses and pilot projects" to be arranged by getting in touch, which describes negotiated licenses rather than a self-service trial. The application bundle does carry unpriced tier labels - a freemium tier, an Individual Tier and an Enterprise Tier - but these are internationalization strings that are never rendered to any user and carry no amount and no condition, so they cannot be read as a published price list. No payment processor is present anywhere in the code. Commercial access runs entirely through contact with the vendor at info@aiqimia.com, and prospective buyers should expect a quotation rather than a rate card. This record accordingly leaves the starting price, the currency and the billing unit empty, because none of the three is public.
Data, GDPR & hosting
A consolidated view of how Ai.Qimia handles your data.
GDPR overview
There is no mention of the GDPR anywhere: not one occurrence across the site pages or the web application bundle, no cookie banner, no privacy policy, no data protection officer and no Article 27 representative. The Swiss Federal Act on Data Protection, the vendor's own home regime, is not mentioned either. No sub-processor is disclosed and no hosting location is stated. Ai.Qimia neither claims nor denies compliance, which is why this record leaves the compliance field empty rather than recording a negative. What is documented here is an absence of published information, not evidence of non-compliance: an EU buyer will have to request the vendor's data protection terms directly, because none can be read before making contact.
Who owns the data?
No published document from Ai.Qimia addresses ownership of the data. A full inventory of the site - the WordPress REST API, both sitemaps and a probe of the usual legal paths - returns five pages and one news post, with no terms of service, no privacy policy and no data processing agreement. The web application nevertheless asks each new user to tick a box reading "I have read and I agree to the terms of use and privacy notice", a string carried in the application bundle with no link and no URL behind it. Ownership of the molecules, reactions and computed results a customer submits is therefore governed by no document anyone can read before signing up, and by nothing this record can verify.
Reuse rights
Nothing published states what Ai.Qimia may do with submitted data, and nothing states what a customer may do with the results. There is no usage policy, no license grant and no re-use clause, because no terms of service exist in public. Targeted searches for opt-out, do-not-train, disable-training and zero-retention wording found no match across the six site pages or the web application bundle. In practice the workflow sends molecules in SMILES notation and reaction definitions to a JSON-RPC API, and returns ensembles, descriptors and provenance metadata; whether those outputs may be redistributed, published in a paper or fed into a commercial model is left entirely undocumented. Nothing here says the vendor claims rights over customer chemistry - only that no public text grants, restricts or clarifies anything. Any team planning to re-use results should obtain a written license before uploading proprietary structures.
Data retention & training
Hosting summary
Ai.Qimia publishes nothing about where customer data is hosted or processed. No country, no region, no cloud provider and no sub-processor is named on any page of the site or in the web application bundle, and there is no privacy policy or security page in which such a statement could sit. The hosting fields of this record are therefore left empty rather than inferred. What can be observed from outside is infrastructure, not a commitment: the aiqimia.com domain resolves to 205.196.222.37, an address operated by DreamHost (AS26347, New Dream Network, LLC) in the United States, while the application is served from app.aiqimia.com and the JSON-RPC endpoint from api.aiqimia.com. Where a website is served says nothing about where calculation inputs, molecular structures or computed results are stored. The vendor is established in Switzerland, which would place it under Swiss data protection law, but the site never states this and never identifies the jurisdiction governing customer data. Any organization with a hosting or data residency requirement will have to obtain the answer in writing from the vendor.
Things to keep in mind
Risks and trade-offs to weigh before adopting Ai.Qimia.
- You are asked to agree to a contract you cannot read: the application's sign-up form requires accepting a terms of use and privacy notice that exist nowhere in public, so nothing documents what you have actually signed
- Data governance is a blind spot: no privacy policy, no hosting location, no retention period and no statement on whether submissions train the models - a serious gap when the material uploaded is proprietary molecular structures
- Performance claims are unverified: the 90, 40 and 100 percent figures on the process page are vendor assertions that are not even rendered in the HTML, and the 30 percent accuracy gain in the ReaQt announcement is credited to anonymous early adopters, with no named customer or case study anywhere
- Over-reliance is a real risk with predictive chemistry: ensembles, yields and mechanisms returned by the platform are model output, not experimental fact, and treating them as results - or letting junior chemists skip the reasoning that used to precede a synthesis - erodes exactly the judgment the tool is meant to support
- Support is a single mailbox: the published contact page is empty, so info@aiqimia.com is the only usable channel, with no stated response time and no escalation path
- The central sales argument has no manual behind it: API-first integration is claimed while no API documentation is published, which means vendor-dependent onboarding and real exposure if the relationship ends
- The presence is recent, static and outside the EU: the domain was registered in April 2025 and the only news item dates from the same month, the vendor is Swiss so the GDPR does not apply automatically with no equivalent published commitment, and the earliest archived version of the site advertised contract research and consulting or development services that no longer appear - worth confirming what is still on offer
Setup & Integrations
Technical difficulty
Moderate, and the real barrier is domain expertise rather than IT. Account creation is self-service on app.aiqimia.com, and basic use means entering molecules in SMILES notation by text or drawing, choosing isomer options, submitting and tracking status. That assumes fluency in SMILES and in quantum chemistry concepts such as tautomers, protomers, conformers and transition states. The advanced route, a JSON-RPC 2.0 API, carries no public documentation, so integration effectively requires a conversation with the vendor. No tutorial, quick-start guide or user manual is published.
Deployment
Supported languages
Behind Ai.Qimia
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does ConstruQt actually produce?
How is the approach different from ordinary cheminformatics descriptors?
Which modules does Ai.Qimia offer?
How do you integrate it with an existing pipeline?
Is there API documentation?
How much does Ai.Qimia cost?
Is there a free plan or a free trial?
Is there a mobile app?
What guarantees are given about the data you upload?
Who publishes the tool and how do you reach them?
Should you pick Ai.Qimia?
Ai.Qimia is a precise, well-defined technical proposition wrapped in an unusually thin public presence. The science is stated clearly: quantum-mechanical ensembles ranked by Boltzmann weighting, descriptors and provenance metadata for machine learning, then reaction and interaction simulation through ReaQt and InteraQt. The people behind it are named and checkable, from the leadership team to an advisory board drawn from Evotec, SpiroChem and Qilimanjaro Quantum Tech, and Swiss public-innovation bodies - Innosuisse, the Canton de Vaud, Biopole Lausanne and FIT - appear as supporters on the home page.
Commercially, though, everything points to an early stage. There is no price, no API documentation, no terms of service, no privacy policy, no named customer and no case study. The headline percentages on the process page - 90 percent of molecules de-risked before scale-up, 40 percent faster scale-up, 100 percent of negative data captured - are vendor claims that are not even rendered in the page HTML, and the 30 percent accuracy gain in the ReaQt announcement is credited to unnamed early adopters. None of that makes the claims false; it makes them unverified. The domain itself was only registered in April 2025, and the site's single news item dates from the same month.
The practical consequence is clear. Everything a serious evaluation needs - contractual terms, data handling and hosting, retention, model training, pricing and API specifications - has to be obtained by contacting the vendor, because the site publishes none of it. For a computational chemistry or machine learning team with a real need for physics-grounded molecular data, a pilot project is a reasonable next step. For anyone who needs documented governance before uploading proprietary structures, the paperwork has to come first.
- Choosing a selection results in a full page refresh.
- Opens in a new window.