XWAI
The AI agent built into S-Cube's platform. Ask XWAI to navigate your inversion sequences, compare job parameters, pull cost and runtime data, render model slices, and answer methodology questions — all in plain language, grounded in your actual project data from inversion jobs running on AWS.
Your projects. Instantly queryable.
The conversation on the right is a real XWAI session on an active project — navigating an inversion sequence, filtering jobs, comparing parameters, and rendering model slices, all through chat.
- Sequence steps, job lists, and parameters — retrieved instantly
- Side-by-side velocity model slices rendered in the chat
- Cost and runtime pulled from live job telemetry
From job groups to per-iteration cost.
The same copilot drills into a subsalt update — opening job groups, building parameter tables, comparing model slices across iterations, and breaking job cost down to the per-iteration level.
- Job groups opened and turned into parameter tables
- Model slices compared across iterations
- Run cost split per iteration, density and mutes diffed
From first slice to final convergence.
A longer session on the same field — isolating reflector forward models, comparing inline slices across iterations, linking straight out to dashboard shot gathers and trace-fit plots, and plotting trace-fit convergence right in the thread.
- Job groups isolated and described in plain language
- One-click deep-links into dashboard shots and trace-fits
- Trace-fit convergence plotted inline across iterations
Everything you need, one conversation away
XWAI is connected to your projects, jobs, and knowledge base — so every answer is grounded in your actual data, not generic guidance.
Project & Sequence Navigation
Lock onto any project and instantly retrieve its full inversion sequence — AWI, RWI, TSL-AWI steps and all. XWAI reads your project description so you never have to hunt through docs.
Job Discovery & Filtering
Find, filter, and cross-reference jobs across complex sequences and job groups in plain language. Narrow by functional type, frequency, mask, or any parameter — and refine on the fly.
Parameter Comparison
Ask XWAI to compare a set of jobs and it generates a clean parameter table showing only what differs — frequencies, functionals, masks, densities, mutes, tiletraces, and more.
Cost & Runtime Reporting
Pull cumulative runtime and AWS cost — total or broken down per iteration — for any job or selection. Know exactly what a run cost before planning the next one.
Visual Model Comparison
Request any inline or crossline slice of one or more velocity models and XWAI renders them side-by-side — across sequence steps and iterations — so you can see how the model evolved.
Shot & Gather QC
Inspect predicted versus field data at any shot. XWAI surfaces shot gathers and predicted-data panels so you can QC the fit in the data domain, not just the model.
Convergence & Trace-Fit Charts
Plot trace-fit convergence across iterations right in the chat. Compare how competing jobs converged and spot stalls or gains at a glance — no spreadsheet required.
Dashboard Deep-Links
When a view needs the full platform, XWAI hands you a one-click deep-link straight to the right dashboard page — the exact job, shot, or trace-fit plot, already loaded.
Knowledge-Base Q&A
Ask technical methodology questions — AWI, RWI, TSL-AWI, cycle skipping — and get answers grounded in S-Cube's internal knowledge base with cited source documents.
Elastic compute, spent once, not twice
The inversion jobs behind every session on this page run on AWS. That is what makes production FWI possible at this scale — and it is also why finding a previous result matters as much as computing a new one. On the cloud, compute is elastic but never free, so every workflow you don't have to run again is money back.
Inversion runs at hyperscale on AWS
Full Waveform Inversion is embarrassingly parallel by nature, and our jobs run on AWS to exploit that — sub-jobs distributed across regions and availability zones on Amazon EC2 Spot capacity, with Amazon S3 delivering the throughput a petabyte-scale seismic dataset demands. Elastic capacity is what makes semi-global multi-parameter FWI practical at production scale.
XWAI reads that estate in place
Every answer XWAI gives traces back to a job that ran on AWS. It reads the live run record — job groups, parameters, captured metrics, cost and runtime telemetry, model artifacts — rather than a parallel copy, so what you interrogate in chat is what actually executed.
Re-use instead of re-run
This is where the cost saving lands. Inversion compute is the expensive part of the workflow, and the most expensive failure mode is launching a fresh run because an existing result was too hard to find or trust. When a practitioner can establish in a single question whether a result already exists, what parameters produced it, and why those parameters were chosen, the default shifts from re-running the workflow to reusing it — and the copilot costs a fraction of the jobs it prevents.
Try XWAI for yourself
Open the live preview, explore the saved sessions above, and start your own conversation with XWAI.
Free preview · takes 20 seconds to start