Skip to content
Contact us

Services

Industrial Data and AI

We take industrial AI from the sensor stream to an answer an engineer can trust, and we run it in production. The same team builds and operates ID3, the drilling data platform used on more than 150 rigs.

rigs running ID3 at the same time, in more than 35 countries
150+
less time preparing daily drilling reports once AI drafts the time log, on four offshore rigs
>50%
resolution of the rig time breakdown built from sensor data
5 s

AI that engineers can trust

Industrial operations produce more data every year, and most of it sits unused. We put it to work with data pipelines, models and AI agents built for industrial environments and grounded in engineering, so that decisions about wells, plants and assets rest on evidence rather than habit.

We build for the conditions industrial data actually presents: sensors drift, reports arrive late or incomplete, labelled examples are scarce, and every rig, plant or fleet behaves differently. So every alert names the signal driving it, every AI answer cites the data behind it, and any action waits for a person to approve it.

We hold our tools to a strict low-false-alarm discipline, because a tool that cries wolf gets ignored. Performance is measured on data the models have never seen and compared with a simple statistical baseline. Our methods are published in industry journals and conferences: see Research.

What we build

  1. Real-time data integration

    We connect to rig and plant data as it is produced, over WITSML 1.x and 2.x (ETP), WITS level 0 and OPC UA, and from LAS and DLIS files and REST interfaces. Units are normalised to the Energistics standard, so data from different contractors and vendors can be compared.

  2. Automated data quality

    Checks run every day on live wells. Where a finding is ambiguous, an AI agent explains it and quotes the exact evidence, and every correction waits for a person to approve it. Clean data is the first condition for any model worth trusting.

  3. Anomaly detection

    Deep-learning detectors flag unusual behaviour in mudlogging sensor data as it arrives, so engineers see a problem developing instead of reading about it in the next day's report.

  4. AI-drafted drilling reports

    From rig data, AI drafts the time log of the daily drilling report, with activities, IADC codes and non-productive time. Engineers review instead of typing, and on four offshore rigs report preparation time fell by more than half.

  5. AI assistants on your data

    Engineers ask about wells, KPIs and reports in plain language. Answers come from a fixed set of tested queries rather than free access to the database, and each user sees only the wells they are allowed to see.

  6. Documents into data

    Reports that arrive as PDF, such as daily drilling reports, become structured data that can be searched, compared and analysed next to the sensor data.

Built for production

  • Runs where your data lives

    Containerised services on Kubernetes, on AWS, on Azure or on your own servers, deployed with infrastructure as code.

  • Your choice of AI provider

    OpenAI, Azure OpenAI, Anthropic or models running on your own servers, whichever your IT department approves. Changing provider does not change the application.

  • Your data stays yours

    Your data is not used to train AI models, and each client's data lives in its own databases.

  • Access you control

    Single sign-on with OIDC or SAML, roles, and permissions down to the single well.

  • Guardrails on AI

    AI features have usage limits and an off switch for each client, and the assistants only read data, never change it.

  • Monitored in operation

    Services, data flows and AI usage are monitored in production, with alerts to our team.

How a project runs

  1. 1

    Data review

    We look at your data sources, their quality and the decision you want to improve, and agree how success will be measured.

  2. 2

    Pilot on historical data

    We prove the approach on past wells or assets in a few weeks, against a baseline you can check.

  3. 3

    Integration

    We connect live data, deploy on your cloud or servers, and set up access and monitoring.

  4. 4

    Operation

    We run and improve the solution with your team, with regular reviews of the results.

Case studies

How can we help you?

Tell us about your project, asset or data challenge. Our consultants and engineers will get back to you.

Get in touch

Quick links

↑ ↓ to navigate ↵ to open All results

Send us a message

We will get back to you shortly.