Skip to content

Platform

AI that drafts engineering work for people to review

Agentic AI workflows across an all-in-one engineering platform. AI agents inside the analysis modules draft failure modes, fault trees, bowties, study findings and requirements from your own documents and models, cite their sources and wait for you to accept the result.

Overview

Most of the effort in safety and reliability work goes into producing a first complete draft: listing failure modes, building a tree, filling a worksheet. AI can speed that up, but only if engineers stay in control and can see where each suggestion came from.

Each module's AI assistant works on its own method. Proposals arrive as drafts with the evidence behind them, and you accept, edit or reject them. Grounding draws on up to four sources: asset data, uploaded documents, the Reference Library and reviewed Knowledge Repository examples, each set to auto, always or never.

Approved terms from the Terminology module steer AI generation in several modules. AI use is metered in credits: every plan includes a monthly allowance, from 1,000 credits on the free Explorer plan, and extra credits can be bought and never expire.

Capabilities

FMEA agent

Drafts failure modes per asset from its condition and context, removes duplicates and shows the evidence for each row.

Fault tree agent

Drafts skeleton or full fault trees, expands selected nodes and answers questions in chat.

Bowtie agent

Generates complete bowties with barrier details from a hazard and top event.

Event tree agent

Generates event tree proposals and revises them stage by stage in chat.

HAZID and HAZOP suggestions

Suggests study findings from retained source documents, citing the excerpt behind each one.

STPA agent

Generates a first STPA study or works step by step, and proposes unsafe control actions.

Requirements extraction

Extracts requirement proposals from documents with page and passage citations.

Systems and behavior agents

Draft system structures from documents and behavior from linked parts.

Case Desk triage

Turns free-text reports and photos into proposed triage sheets with confidence per field.

Asset document extraction

Reads documents into a review queue of asset values, condition items and maintenance.

Process and plan drafting

Drafts BPMN processes and project plans from documents, with citations.

Report writing and slides

Drafts report summaries and builds PowerPoint decks from reports on your template.

AI assistance

  • Grounding in asset data, uploaded documents, the Reference Library and reviewed Knowledge Repository examples, each set to auto, always or never.
  • Citations the AI claims are checked on the server against what was retrieved, and invented ones are dropped.
  • Reviewed examples influence future runs only after a person promotes them.
  • The in-app Agent helps you navigate, answers help questions and runs module commands, and supports voice.

AI proposals are drafts: you review and accept them before they enter your model.

Who uses it

  • Safety and reliability engineers who want faster first drafts
  • Teams that need traceable, reviewable AI output
  • Consultants scaling analysis work
  • Students learning methods with AI support

Works with

Frequently asked questions

Does AI change my analyses without asking?
Module AI assistants produce drafts and proposals that you accept, edit or reject before they enter your analysis. Some low-risk actions of the in-app Agent, such as creating a registry snapshot, run directly when you ask for them.
Is our data used to train AI models?
No. ModelGraphX uses AI models for inference only and does not fine-tune them on your engineering data.
How is AI use charged?
AI operations use credits. Every plan includes a monthly allowance, for example 1,000 credits on Explorer and 120,000 on Pro, and you can buy more credits, which never expire.
Can the AI work from our own documents?
Yes. Grounding can draw on uploaded documents, the Reference Library, asset data and reviewed Knowledge Repository examples, and drafts cite the sources they used.

Try it on a real project

Explorer is free, and new accounts get a 30-day trial of the paid capabilities.

All modules
© Archer Futura Technologies AB