Quaisr ads

Proud to have presented alongside BHP on a problem geoscientists know well. Many plausible geological models can fit the same data; looking at just one can hide that uncertainty. For teams deciding where to invest, that can mean taking on risks they haven’t accounted for. We shared how to both construct and leverage an ensemble of models to guide subsequent modelling, geological interpretations and decision-making. For exploration teams like BHP's, this means better understanding of the space of possibilities, less time second-guessing a single model, and more confident investment decisions. #Geoscience #AI #Mining #Quaisr #BHP #DAIMNA26

Quaisr was proudly featured at the Chemical Engineering Research Showcase 2026 by Department of Chemical Engineering at Imperial College London. A full afternoon of brilliant minds sharing research on everything from green solvents to AI-driven drug discovery. Events like this remind us why we do what we do. The conversations, the ideas, the people... it's genuinely exciting to be in a room where the future of engineering is being shaped. Thank you to Omar Matar, Incoming Dean of Engineering at Imperial and Quaisr Co-founder, and the whole Department for putting together such an inspiring day. #ChemEngShowcase26 #ImperialCollege #Quaisr #digitaltwins #research #simulation #tech #modelling #ai #ml

"Whoa... he just did the proof of concept as an example. And I get the impression you did it this morning, which is freaking me out." $130bn Global Mining Major We’ve all gotten used to a certain rhythm at work. You ask for a complex data integration or a new model setup, and you mentally log it for next month. We just adapt to the waiting. When we heard this feedback from a partner, it was a good reminder of why we started building Quaisr. It wasn’t about trying to show off or prove we could work faster than anyone else. It was just about what happens when the digital plumbing actually works the way it's supposed to. Geologists and engineers shouldn't have to spend weeks manually wrangling satellite feeds and old physics models just to see if a plan is viable. When you clear the friction out of the background and let those systems connect naturally, work that used to drag on for months can finally happen in a morning. We’re incredibly grateful for partners who trust us with their workflows, and it feels rewarding when we can give them some of their time back. #ResourceExploration #Geology #MiningEngineering #DataWorkflows #Quaisr

The $2M drill hole that hit nothing. A geology team builds a model from survey data, the model looks defensible, the board signs off, and the drill comes back empty. Nobody did anything wrong; the model was a reasonable interpretation of the data. It just wasn't the only one. That's the part usually missing from the process: any given magnetic survey is consistent with more than one underground structure, depending on which assumptions you started with. Most workflows pick one model and move forward, because generating even a single inverted model is slow enough on its own, manual, iterative, and usually stuck on whichever specialist's laptop ran it. Quaisr changes this. Instead of one model taking days of manual iteration, the same survey data and geological priors can be run through the inversion process many times over, enabling probabilistic methods to be utilised to generate an ensemble of models, with attached uncertainties. Pictured: airborne magnetics survey (top), which is inverted probabilistically to generate an ensemble of models for better drill targeting (bottom). Anonymised illustrative example, not real survey data. The value isn't the ensemble itself, though. It's what clustering across it reveals about the uncertainties: structures that persist across priors , versus regional uncertainties highlighting missing constraints.. That distinction is what gets handed to borehole planning, not a guess dressed up as certainty, but a target that survived being checked against its own alternatives. Drilling decisions should get made with certainty, probabilistic modelling provides the uncertainty analysis to do this. This ensures drilling decisions are against every reasonable version of the model first, not just the one that happened to be built. #GeoPhysics #MiningTech #MineralExploration #GeologicalModeling #DataDrivenMining #ExplorationGeology #SubsurfaceModeling #DrillTargeting #ResourceExploration #MiningInnovation #GeoAI #AppliedAI #ProbabilisticModeling #Quaisr #DigitalGeoscience #MagneticSurvey #UncertaintyQuantification #FutureOfMining

The hardest balance to strike in platform engineering is finding the right level of abstraction. Make a tool too visual, and your developers reject it. Make it too code-heavy, and it becomes a black box to stakeholders and scientists. But today, there’s a new layer of complexity: your platform doesn't just have human users anymore. You now have to design for AI agents. When you watch the workflow above, you aren't just seeing a chat interface, you are watching multimodal interaction solve this abstraction problem in real-time. Notice what happens when the user asks about the effect of cooling temperatures during crystallisation. The user doesn't hunt through repositories for a script. They simply state their intent. Because the system is built with Model Context Protocol (MCP), the internal LLM natively "tool calls" the exact approved simulation workflow. When the AI recommends a Bayesian Optimisation approach over a pure gradient method, it doesn't just return a text response. It physically pulls the optimization block onto the canvas and wires it to the simulation. The logic isn't buried in code, it is transparent, interpretable, and completely audit-ready for scientists and architects. When the user asks for a summary, the underlying infrastructure executes the complex solvers in the background. The user gets a clean, visual scatter plot, while the developers maintain complete control over the API and code layer without needing to refactor a single line. A shared system doesn't require a shared interface. True velocity happens when you stop forcing users into a "single pane of glass" and instead meet your developers, your scientists, and your AI exactly where they work. #PlatformEngineering #DataScience #AIAgents #DigitalTwins #SystemsArchitecture #Quaisr

Four years ago, Quaisr pitched at Imperial College London's ChemEng Enterprise Day for the first time. Last week we were back, represented by Assen Batchvarov, PhD, this time as an established spinout, sharing the room with 20+ ventures building some genuinely important things. From DyeRecycle's breakthrough in textile recycling, to ElectroPET turning plastic waste into green hydrogen, to Sangcyte's nanomedicines for targeted therapy. The pipeline coming out of Imperial is remarkable. We also heard reverse pitches from investors, a format that flips the usual dynamic and makes for a much more honest conversation. Great perspectives from Tanel Ozdemir and Leigh Brody at Albion Capital Group LLP, Alexis Zervoglos at Science Capital, Michelle Robson at Odyssey Ventures, and the teams at TSP Ventures and Parkwalk Advisors. The day closed with a lecture from Nalin Agarwal, CEO of Climate Collective, a useful reminder that the most important climate innovation isn't always happening where you'd expect. Proud to still be part of this community. More to come. #ImperialEnterpriseDay #DeepTech #ChemEng #Spinouts #Startups #CleanTech #ClimateTech #Innovation #VentureCapital #Imperial #London #NetZero #Sustainability #Entrepreneurship #ScienceAndTech #GreenTech #CircularEconomy