Intelligence for Operational Safety

The AI that understands real work and turns it into management decisions

NormalWork Chat applies Learning From Normal Work (ACTN/LFNW), HOP, Safety II and Human Factors to turn field reports into decision narratives, Impact Reports and Learn Backs, in minutes.

Validated by a specialist with 40+ years in HSE
ACTN/LFNW · Learning From Normal Work HOP Safety II Human Factors Learning Teams
Features

Built to understand the field, not just answer questions

An expert copilot that captures real work, reasons with method and delivers decisions ready for management.

Understands real work

Reads field reports and distinguishes fact, hypothesis, inference, uncertainty and recommendation, without guesswork or blaming people.

Operational drift

Reads the reports and estimates how much margin has already been consumed, not how many rules were broken. It turns the deviation that seemed normal into a signal that the boundary is closer.

Operational resilience

A practical reading of the four potentials – Monitor, Anticipate, Respond and Learn – from real work. It shows where the operation is strong and where the capacity to absorb the unexpected is thinnest.

Reads images, PDFs and Office

Attach photos (with high-resolution OCR), PDFs and spreadsheets. The system extracts and analyzes the content inside the conversation.

Generates ready-to-use deliverables

Impact Report, Learn Back and decision cover pages, in PDF/DOCX with your brand's visual identity, ready to circulate.

Ask by voice

Record the report by speaking; the transcription understands safety vocabulary and field technical jargon.

Curated, secure knowledge

A client-neutral methodological base: it doesn't expose one client's data to another and keeps methodological rigor in every answer.

On desktop and mobile

A modern, responsive interface, with per-client co-branding and usage control, ready for the whole team.

Process Safety

NormalWork Chat as a Process Safety accelerator

PSM is strong at identifying technical hazard and weak at seeing real work. What blocks this integration isn't method, it's channel. The weak signal the operator notices (a fluctuating gauge, a valve that sticks, a procedure that changed informally) dies in the shift and never reaches the HAZOP.

NormalWork Chat closes that gap

It captures the signal in natural language, structures it as scenario, barrier, mechanism and evidence and connects it to your existing HAZOP/JSA. You anticipate barrier degradation before the LOPC and generate a systemic recommendation instead of "more training".

It doesn't replace, it amplifies

It doesn't replace HAZOP, LOPA or SIL. It's the channel that makes those methods see real work.

How it works

From field report to decision, in a single flow

A single, auditable pipeline that turns what happens in real work into learning and action.

Capture real work

The team records the report by text, voice or attachment (photo/PDF). What actually happens in the field becomes a sensor.

Analyze with method

NormalWork Chat applies ACTN/LFNW/HOP/Safety II: it classifies, measures, interprets and flags uncertainties transparently.

Decide and give back

Out comes the decision narrative and the Impact Report, and the cycle closes with the Learn Back, taking the learning back to the field.

Capture Classify Measure Interpret Decide Learn Back
Real result

From field report to Impact Report, automatically

Attach the learning team's note. The system reads it, classifies the findings and delivers a ready Impact Report, in PDF, with your brand.

Field report · ATN
Site: Blue Bay Terminal (illustrative scenario)
Activity: fuel transfer between tanks.
Difficulties: the main pump wouldn't start; the electrical panel had no power due to a breaker fault, only noticed at the time of operation.
Adaptations: on-call electrician called verbally; backup pump used to avoid missing the window.
Error traps: level indicator reading diverging from the real volume.
Solutions: energize the panel; review the procedure and define an electrical contingency plan.
Field report (text)
Impact Report generated by NormalWork Chat
Impact Report generated by the system · PDF
Moisés Costa at talks and events
Expert validation

Moisés Costa

Founder of CSS (Costa Safety Solutions)

It's not "just AI". Every step of the system's creation was validated with the experience and knowledge of a specialist with more than 40 years in Oil & Gas and senior HSE leadership at an industry multinational, ensuring quality and methodological rigor from the very start.

Occupational Safety Engineer 40+ years in Oil & Gas Senior HSE leadership Researcher at PUC-RS ACTN specialist
View on LinkedIn

See NormalWork Chat working with your data

Request a free demo and a personalized quote. Our team responds quickly.

produtos@costass.com.br

Or write directly to produtos@costass.com.br and talk to our team.