ORIA ORIA ORIA · AI agent

An AI agent that does the work. And shows every step.

Give ORIA a goal. It plans, uses real tools, checks the evidence, and hands back a verdict, a fix or a finished report. Built by NET4NINJA, NIDSARK's parent company, and the agent inside NIDSARK's NSWEB browser.

Agent run · Phishing triageORIA
ORIA chat: a user asks whether an Outlook Web App login link is phishing. ORIA shows its plan, then six completed steps: checked threat intelligence, read the page, searched the web, ran Python, saved to memory, and built a PDF.
How it works

Give it a goal. Watch every step.

Each tool call appears as it happens, with its arguments, its result and how long it took. Nothing runs out of sight.

01

Plans

Reasons about the goal and lays out the steps before it touches a tool.

02

Acts

Searches, reads pages, runs Python, checks threat intelligence and calls connected tools.

03

Checks

Ties each claim to a source or a tool result, and says when the evidence is thin.

04

Delivers

Hands back the verdict, the fix, or the Word, PowerPoint, PDF or Excel file you asked for.

Runs Python on the evidenceORIA
ORIA's Python step expanded: code comparing the lookalike domain with the real one, and its result showing one swapped character and a similarity of 0.79.
Delivers the verdict and the fileORIA
ORIA's answer: credential phishing, an evidence table, three actions to take now, a one-page PDF incident note ready to download, a saved memory, and sources from CISA and NIST.
ORIA Cyber

Investigations that show their evidence.

Paste hashes, URLs, domains or IPs, or ask in chat. ORIA Cyber checks them against VirusTotal, opens sandbox behaviour and relationships, and keeps every lookup on a timeline.

Six indicators, triaged at onceVirusTotal demo data
ORIA Cyber Investigations: six indicators pasted in, five returned malicious and one with no detections, with totals for each verdict.
VerdictDemo data
A malicious file's detection summary: 41 of 69 engines flagged it.
Sandbox behaviourDemo data
The sample's process tree from provider sandboxes: the payload starts cmd.exe, PowerShell and a scheduled task.
RelationshipsDemo data
A relationship graph linking the sample to dropped files, contacted domains and contacted IP addresses.

Screenshots use VirusTotal demo fixtures, marked as demo data in the product. A live deployment uses your own VirusTotal key.

Toolkit

One agent. The tools to finish the job.

Switch tools on or off per conversation. ORIA picks the ones a task needs.

Research with sources

Live web and X search, page reading, and multi-agent research where several agents search at once. Citations on every answer.

Python on the evidence

Writes and runs code for maths, parsing and data, and shows you the code and its output.

Finished documents

Builds Word, PowerPoint, PDF and Excel files, illustrated on request, ready to download.

Memory and skills

Remembers what you ask it to keep, and loads your instruction packs when a task calls for them.

Live voice

Talk to ORIA and hear it answer, with the same tools available while you speak.

In the NSWEB browser

ORIA reads pages, compares tabs and helps fill forms, and asks before anything sensitive. Never passwords or payment.

Guardrails

Autonomy you can audit.

An agent is only useful if you can check its work. ORIA is built to be checked.

Every step visible

Tool calls, arguments, results and timings stay on screen, and the reasoning is one click away.

Evidence, not hope

"No detections" is never reported as safe, and community comments are treated as claims, not findings.

You approve actions

Submissions, rescans and browser actions that matter wait for your go-ahead.

A full audit trail

Lookups, evidence and actions are recorded per investigation, with where each fact came from.

Put ORIA on your hardest ticket.

See it plan, investigate and report on a real case from your environment.