Reporting
- Today
- Someone copies figures from spreadsheets and systems into a monthly report.
- With automation
- The report is assembled and drafted for you; you review and send.
AI that works for your business and stays inside it.
We automate the reporting, customer replies and admin that eat your team's week, and build the private AI systems behind it: knowledge assistants, AI agents and local AI servers, with the cybersecurity and backups to run them safely.
Everyday work we automate
You do not need to be a science company or have an IT department. If your team spends hours each week on tasks like these, AI can take the first pass and leave the decisions to your people.
Customer service
- Today
- The same questions are answered by hand, one email at a time.
- With automation
- Replies are drafted from your own policies and past answers; a person approves anything sensitive.
Accounting and admin
- Today
- Invoices, payments and records are matched line by line.
- With automation
- Routine matches are made automatically and the exceptions are flagged for a person.
Consultancy and professional services
- Today
- Every proposal and client report starts from a blank page.
- With automation
- First drafts are built from your previous work, in your style, ready to edit.
Finding information
- Today
- Answers are buried in shared folders, old emails and one colleague's memory.
- With automation
- Ask a question in plain language and get the answer with the document it came from.
Enquiries and follow-up
- Today
- Leads and requests wait in an inbox until someone has time.
- With automation
- Enquiries are sorted, summarised and routed, with a suggested reply waiting.
Why this matters for your business
AI is only worth it if it changes how the working week goes. These are the results we design for.
Hours back every week
Routine drafting, copying and checking is done before your team sits down, so their time goes on customers and decisions.
Knowledge that stays
What your best people know is captured in systems everyone can ask, instead of leaving when they do.
Fewer routine mistakes
Repetitive checks are done the same way every time, and anything unusual is passed to a person.
You own it
The systems, the data and the website are yours. No lock-in, and no bill that rises with every new user.
What we build
Seven services that fit together. Most clients start with one, usually a private knowledge assistant or a backup review, and add the rest as confidence grows.
Local AI servers
In plain terms
An AI assistant like the well-known chat tools, but running on a computer in your office instead of in someone else's data centre.
Capable language models running on hardware you own, on your premises or in a UK data centre you control. Your documents, prompts and outputs never train someone else's model.
- Hardware sizing and model selection
- Installation, hardening and monitoring
- Private chat for your team, and a connection (API) for your other software
Knowledge assistants (RAG)
In plain terms
The AI looks up the relevant pages in your own documents before it answers, so it works from your facts instead of guessing, and shows you where each answer came from.
RAG stands for retrieval-augmented generation.
Ask questions of your own reports, procedures, contracts, lab notebooks and correspondence, and get answers that cite the document they came from.
- Ingestion of PDFs, Office files, email and databases
- Access control that follows your existing permissions
- Answers with references you can check
Connections to your tools (MCP)
In plain terms
Think of it as a universal adaptor: it lets an AI assistant look things up and carry out tasks in the programs you already use, such as email, your customer database or your accounts, and only where you allow it.
MCP stands for Model Context Protocol, an open standard for connecting AI to other software.
Your AI assistant stops being a separate chat window and starts working inside the systems you already use, reading and acting with exactly the permissions you grant.
- Connectors for email, files, customer records (CRM), accounting, planning (ERP) and laboratory (LIMS) systems
- Custom servers for in-house tools and databases
- A log of every action an assistant takes
AI agents and automation
In plain terms
A chatbot answers one question at a time. An AI agent is given a goal, works through the steps itself, and checks with a person before anything important.
Agents that carry out multi-step work: compile a report, triage an inbox, reconcile records, watch a process and flag what needs a person.
- Workflow mapping to find what is worth automating
- Human approval at the steps that matter
- Measured by time saved, not by demos
Cybersecurity
In plain terms
Locks, alarms and house rules for your computers and your AI tools, so the wrong people stay out and your data stays in.
AI tools add new ways in. We secure them together with the rest of your systems, and write the rules your staff will actually follow.
- Network hardening, key-based access and minimum necessary permissions
- Safeguards against prompt injection (hidden instructions that trick an AI) and data leaks
- Practical security and AI use policies
Backup and recovery
In plain terms
Automatic spare copies of everything important, kept in more than one place and tested, so you can get back to work quickly after a failure or an attack.
Encrypted, tested, off-site backups, so a failed disk or a ransomware attack is an inconvenience rather than a crisis.
- 3-2-1 backup design: three copies, two kinds of storage, one off site
- Automated, encrypted and versioned
- Written recovery procedures, rehearsed
Websites you own
In plain terms
We build your site and put it on a small server that is yours, so you are not renting it from a website platform month after month.
A fast, modern website built for your business and hosted on a small server you control, for a fraction of what subscription website services cost to run. This site is built and hosted the same way.
- Design, build and copy, in one language or several
- Self-hosted: no platform subscription or per-feature fees
- Secured, backed up and easy to update
AI should make your people better at their jobs, not replace them.
We build AI to work under human direction. Your team's judgement, experience and relationships are what your customers value; our job is to give those people more time and better information.
People decide
AI prepares, drafts and checks. Anything that matters, such as a reply to a customer, a payment or a report, is approved by a person before it goes out.
Your expertise guides it
Each system is built around how your best people already work, and is adjusted as your team corrects it.
More capacity, same team
The aim is for you and your staff to get more done, and to spend less of the day on work nobody enjoys.
Use AI without handing over your IP.
Public AI tools are useful, but every prompt and uploaded file leaves your control. For companies whose value sits in their know-how and intellectual property (IP), a private system is the safer way to get the same benefits.
Your data stays where you put it
Formulations, process data, client files and unpublished results are processed on machines you control. Nothing is sent to a third party unless you decide it should be.
Costs you can plan
Hardware you own replaces usage bills that grow every time another colleague starts relying on AI.
Cloud where it makes sense
A hybrid setup lets sensitive work run locally while routine tasks use cloud models, under rules you set and can audit.
Internal R&D programme
SkAInet: warning of unplanned plant stops months ahead
Our in-house research programme: a machine learning system that aims to warn chemical and food processing plants of unplanned stops months ahead, running entirely on site.
Read about SkAInetBuilt by scientists, for businesses with knowledge worth protecting
We come from chemical engineering and industrial R&D, so we understand confidential data, regulated processes and the cost of getting things wrong. The same care applies to any business that wants AI to take on real work, whatever its size or sector.
- Accountancy, legal and consulting firms
- Service businesses with customers to answer and reports to send
- R&D, engineering and technology companies
- Laboratories and testing services
- Manufacturers and process industries
How we work
A clear sequence from first conversation to a system your team relies on. You can stop after any step and keep what has been delivered.
- 1
Assess
We review how your team works, where your data lives and which tasks cost the most time. You get a short written report with priorities and rough costs.
- 2
Design
We choose the models, hardware, integrations and security controls that fit your budget and risk, and agree what success looks like.
- 3
Build and integrate
We install the system, connect it to your tools and test it on real work with the people who will use it.
- 4
Secure and back up
Access control, monitoring, encrypted backups and recovery procedures go in before anything goes live.
- 5
Train and support
Your team learns to use it well. We stay available for updates, new workflows and the questions that come with growth.
A good place to start
AI Opportunity Review
A short, focused review of how your team works and where the time goes. You get a written report: the tasks most worth automating, what each would take, rough costs, and whether it should run locally, in the cloud or both.
Scope, duration and price are agreed with you beforehand, because they depend on the size of your company and how you work. No obligation to continue.
About
Pro Scientia is a UK company that began as a scientific consultancy in process development, modelling and industrial R&D. It now applies the same discipline to AI: understand the system first, measure what matters, and build things that keep working after the demo.
Dr Nicolás Schulz
Founder and director
Chemical engineer with a PhD in Chemical Engineering and more than 16 years of experience across academia and industry, including Dow, Honeywell and several start-ups. He combines a background in electrochemistry, process engineering and industrial R&D with hands-on work in data science and AI.
- Credentials: PhD in Chemical Engineering; IBM-certified data scientist; twice a Marie Curie Fellow; DAAD Fellow
- Ventures: co-founder and scientific advisor of Hypergen
- Artificial intelligence: AI agents, MCP integrations, RAG systems and local LLMs, implemented in industry
- Machine learning and data science: predictive models and data analysis with Python
- Process engineering: process development, intensification and scale-up, from laboratory to pilot plant
- Research: decarbonisation, electrochemistry and catalysis; 11 peer-reviewed publications, an international patent and articles in The Conversation
Ignacio Cuñado
Associate, chemical engineer
Chemical engineer with experience in process engineering and pharmaceutical process development, combining industrial engineering know-how with the quality and traceability demands of regulated industries. Graduate of Universidad Nacional del Sur and recipient of the Ing. Isidoro Marín Award from Argentina's National Academy of Engineering.
- Process engineering: PFDs and P&IDs, engineering calculations and simulation in HYSYS, Aspen and DWSim
- Pharmaceutical development: formulation development, engineering batch design and manufacturing methods
- Quality and compliance: GxP practices, batch records, deviation management and continuous improvement
- Artificial intelligence: LLM-based systems, RAG and locally run models; machine learning with Python
Tell us what takes your team the most time.
Start with a short call. We will tell you honestly whether AI can help, what it would take, and whether it should run locally, in the cloud or both.
- info@proscientia.ai
- Phone
- +44 (0) 203 916 6151
- Based in
- United Kingdom