Technologies

An integrated technology chain

Four building blocks, developed in-house, cover the full life cycle of a molecule: its design, its development, the control of its production and, tomorrow, its manufacture in a micro-factory.

Our value chain

From molecule to production, an integrated chain

Four building blocks developed in-house, from AI design to controlled production. Hover over each step.

01
Design
ALCHEMAI
02
Develop
CONTINUOUS FLOW
03
Control
ALCHEMDRIVE
04
Produce
CHEMPOCKET
01 · Explored space 02 · Generation 03 · Screening 04 · Scoring 05 · Shortlist Constraints applied by stage Target properties Physico-chimiques, biological, reaction-related Regulation & HSE REACH, CLP, PFAS-free, ecotoxicity Synthetic accessibility Retrosynthesis, building blocks, price, feasibility Freedom to operate Brevets, familles chimiques exclues (FTO) Espace chimique large ALCHEMAI generative model Generation under constraints Regulatory screening Scoring & tri Candidat A Candidat B Shortlist to synthesise or order for testing Laboratory validation Measured yield, selectivity, model deviations // mode: conceptual pipeline · scale sampled per project
What sets us apart

Control that adapts continuously

Most industrial processes are regulated by advanced process control (APC). AlchemDrive goes further: where APC holds a known set point, our AI learns to drive the process, even when its model is incomplete or conditions drift.

Conventional advanced control · APC

Regulate around a set point

  • Relies on a process model, often linearised around a nominal set point
  • Computes corrections to hold that point: very effective in steady state
  • Optimises mainly steady state, less the phases where the process changes
  • Assumes the optimal regime is already known and degrades when moving away from it
AI control · AlchemDrive

Learn to drive the whole process

  • Learns the operating strategy from experience, with no pre-established analytical model
  • Trained on the full dataset, it also optimises transient phases: start-ups, set-point changes, production transitions
  • Discovers the optimal regime rather than assuming it known
  • Adapts in real time to drift: input variability, catalyst ageing, unexpected disturbances

Where industry often loses yield and material, in the transients, AI learns to optimise the whole trajectory, not just the set point. APC and AI are not always opposed: AlchemDrive can also complement an existing control system, which remains the master system.

When AI control makes the difference

AI control, when it makes sense

We're direct about it: a simple, stable and well-modelled process runs perfectly well without AI. Learning-based control comes into its own in four situations.

Poorly known or new chemistry

No reliable analytical model of the process: AI learns where the equation is missing.

Transient-rich processes

Frequent start-ups, set-point or production changes: optimisation cannot be limited to steady state.

Drift-prone processes

Catalyst ageing, raw-material variability: yesterday's right setting is no longer right today.

Moving optimum

When the optimal set point moves, a fixed setting lets part of the accessible yield slip away.

Why Alysophil

+

Chemists AND data scientistsControlling a chemical process with AI is not improvised from a spreadsheet. Our teams combine synthesis chemistry, process engineering and data science, not just software developers.

+

Industrial cultureWe know the real constraints of the field: process safety, regulatory requirements, scale-up. Control only has value if it holds within that framework.

In practice

These technologies at work

Our technology blocks come into their own on real projects: AI molecule design, transposition of syntheses to continuous flow, production control. See how they combine in our case studies, or explore our engagement frameworks to work with us.

A chemistry hard to transpose?

Let's talk