Documentation
Platform Contact v1.0
Getting started / Overview

Getting started with energyOS

energyOS is an AI-powered intelligence layer built for energy professionals who need to think faster, find relationships in complex data, and develop better-informed market positions. It is not a data dashboard. It is a reasoning environment — one that uses your data and curated market signals to surface insight, test hypotheses, and support critical decisions.

The platform is used by:

  • Trading desks: From proprietary desks to commodity merchants running directional strategies across gas, power, and petroleum.
  • Grid & operations teams: Utilities, ISOs, and independent power producers managing dispatch and load forecasting.
  • Enterprise logistics: Midstream operators and physical commodity firms managing scheduling, nominations, and arbitrage.

The partnership model

Access to energyOS is not a software subscription. It is a working relationship. Every team that joins the platform does so through a direct engagement with REDR Labs — we handle onboarding, configuration, and ongoing support as an extension of your analytical team, not as a help desk.

What this means in practice Partnership model
Onboarding
Led session — not a video walkthrough. Configured around your specific market focus and workflows.
Ongoing access
Direct line to the REDR Labs team. Questions, configuration changes, and new use cases are handled collaboratively.
Platform evolution
Partner feedback directly shapes feature development. Your use cases inform the roadmap.
On expectations

energyOS is a tool for critical thinking, not a display of pre-packaged answers. The platform rewards users who bring specific questions, hypotheses, and their own data. The more precisely you engage with it, the more useful it becomes.

Introductory concepts

Before using energyOS, it helps to understand how the platform's core components relate to each other. Three concepts underpin everything.

concept / 01
The Pipeline
All market intelligence flows through a five-stage automated pipeline. Data enters at stage one and is continuously processed through to the Intelligence Interface. Inferences are calculated at query time, not stored.
concept / 02
EOS Signals
Signals are the output of each pipeline stage — structured objects that capture a market observation, model output, or anomaly flag. The LLM interface is grounded entirely in these signals.
concept / 03
Session Context
Each analyst session maintains a context window. Signals, uploaded data, and conversation history are combined so that every response is grounded in current data. Nothing from your session persists after it ends.
concept / 04
Regime State
The platform continuously evaluates whether the market is behaving within its historical norm or has entered a structural break. Regime state is surfaced prominently in all LLM responses when a shift is detected.
concept / 05
Data Recency
Signals reflect data as of the last integration cycle — not a live price feed. The platform is designed for insight development and relationship-finding, not tick-by-tick monitoring. Freshness timestamps are visible on all signals.
concept / 06
Scalar Correlation
When you upload your own data, the EOS Scalar engine scores it against all active signals — surfacing hidden relationships between your internal datasets and the broader market environment.
Note

energyOS Chat will not speculate beyond its signal context. If a signal is unavailable or stale, it will say so rather than generate an unsupported answer. This is intentional behavior, not a limitation. The platform is built for analytical integrity.

Platform architecture

energyOS is built entirely on AWS infrastructure, running within a dedicated, isolated environment. All computation, inference, and storage operates within this boundary. No data from your sessions is shared, exported, or retained outside of the infrastructure described below.

The five-stage pipeline

stage / 01
Data Ingestion
Market feeds, satellite data, and NLP sources are pulled on a scheduled cycle and written to the internal signal store.
stage / 02
Normalisation
Raw data is cleaned, aligned to common time horizons, and structured into typed signal objects.
stage / 03
ML Inference
Fair value models, LSTM forecasters, and the autoencoder regime detector run against the normalised signal store.
stage / 04
Anomaly Scoring
Outputs are ranked by anomaly severity and deviation from historical baseline. Regime flags are set here.
stage / 05
Intelligence Interface
Scored signals are injected into the session context window and made available to energyOS Chat and the pipeline monitor.

Infrastructure layers

layer / 01 Data sources
Market price feeds Satellite imagery Weather & climate data NLP / news signals Regulatory filings Your uploaded data
↓ ingested on scheduled cycle
layer / 02 AWS compute & storage
Signal store (S3) ML inference (EC2 / SageMaker) Audit trail (CloudTrail) Config recording (AWS Config) Security Hub
↓ inferences calculated at query time, not stored
layer / 03 Intelligence interface
energyOS Chat Pipeline monitor Data ingestion panel Session context window
Key design principle

Model inferences are computed at query time and returned within your session context. They are not persisted to storage. This means each session is a fresh reasoning environment — and your proprietary data never touches a persistent layer.

Get access

Access to energyOS is established through a direct engagement with the REDR Labs team. There is no self-serve signup. Onboarding is fast — most teams are operational within one business day of first contact.

  1. 1
    Reach out directly
    Contact us via the contact page or email info@redrlabs.com. Include your organization name, team size, and primary use case. We'll respond the same business day.
  2. 2
    Scoping call
    A short call with a REDR Labs analyst to understand your market focus, team structure, and the specific questions you want to answer. This shapes how your workspace is configured before you ever log in.
    Typically 30 minutes
  3. 3
    Workspace provisioned
    Your workspace is configured — feed selection, role structure, and initial signal set — and credentials are delivered securely. Provisioning is automated once configuration is agreed.
    Automated provisioning
  4. 4
    Onboarding session
    A REDR Labs analyst walks your team through the platform — pipeline orientation, first chat queries using your actual market questions, and a data upload walkthrough with a real dataset. Not a demo. A working session.
    Included with all plans

Platform onboarding

Onboarding is designed to be fast and practical. The goal is for your team to leave the session with real, actionable output — not just an orientation to the interface. A REDR Labs analyst runs every session.

Before your session

ItemNotes
Primary market focuse.g. Henry Hub gas, ERCOT power, WTI crude
A sample internal datasetCSV or Excel — used in the data upload walkthrough
2–3 analytical questionsQuestions you want the platform to help answer — the more specific, the better
Team roles to provisionAnalyst, trader, admin, read-only

What the session covers

  1. 1
    Pipeline walkthrough
    Your analyst walks through each pipeline stage against the current state of signals in your configured market verticals. You'll see what data is available, what is pending, and how freshness timestamps work.
  2. 2
    First chat session — your questions
    You run your first energyOS Chat queries using the questions you prepared. The analyst shows how the platform constructs its reasoning from the signal context, and where it will say "I don't have enough signal" versus returning a grounded answer.
  3. 3
    Data upload walkthrough
    Upload your sample dataset through the Data Ingestion Panel. You'll observe the automated analysis pipeline run in real time and see how EOS Scalar Correlation scores your data against real signals. The analyst interprets the first set of correlations with you.
  4. 4
    Workflow configuration
    The analyst helps configure feed prioritisation, anomaly thresholds, and role-based access for your team.
  5. 5
    Agreed follow-up
    At the end of every onboarding session, the REDR Labs analyst agrees a set of follow-up items with your team — typically additional feed configurations, questions to explore in your first week, and a check-in date.
    Partnership standard

Your first analysis

energyOS is designed for questions, not keyword searches. Open energyOS Chat and ask a market question in plain language. The platform reasons over the current signal context and returns a grounded answer — citing the signals it used and flagging anything it cannot support with available data.

Example queries to get started

QueryWhat it surfaces
What is the current fair value for Henry Hub front-month?Fundamental model output, storage vs. norm, weather adjustments
Are there any active anomalies in ERCOT real-time prices?Anomaly flags from the analysis pipeline, nodal LMP z-scores
Is the market in a normal or shifted regime?Autoencoder regime state, reconstruction error, last shift timestamp
What relationships exist between my uploaded position data and current storage signals?EOS Scalar Correlation scores, named signal alignments, directional inference
Summarize the key signals from the last 24 hours.All active pipeline outputs ranked by anomaly severity and recency
Tip

The platform rewards specificity. Asking about a named contract, hub, or ISO yields a more targeted response than a broad market question. Where the platform lacks sufficient signal, it will say so explicitly rather than generate a low-confidence answer.

Uploading your data

The Data Ingestion Panel is one of the most powerful parts of the platform. It lets you bring your own proprietary datasets into the session context and have them reasoned against market signals. Once uploaded, the automated analysis pipeline runs within seconds and results are injected into your active chat session.

  1. 1
    Open the Data Ingestion Panel
    Click the upload icon in the chat toolbar, or drag a file directly onto the interface. Accepted formats: .csv, .xlsx, .xls. Maximum 500 rows per file.
  2. 2
    Automated analysis pipeline
    The system runs descriptive statistics, trend classification, 3/7/14-day forecasting, and anomaly detection on every numeric column. No configuration required. Processing completes in under 200ms for standard datasets.
    < 200ms processing
  3. 3
    EOS Scalar Correlation
    Your dataset is scored against all active EOS signals. Strong alignments are surfaced in your next chat response — each with a named correlation, alignment score between 0 and 1, and a directional interpretation. This is where the platform starts making your data work for you.
  4. 4
    Ask questions about your data
    Once ingested, query the platform to analyse, forecast, or cross-reference your data against signals. Ask about relationships you suspect, patterns you've noticed, or hypotheses you want to stress-test.
Privacy

Uploaded files are processed in-memory only within the AWS environment. No proprietary data is written to persistent storage or retained after session end. Your data does not leave the infrastructure boundary at any point.

Next steps by role

Depending on your role, different parts of the platform will be most relevant to your day-to-day work.

Analyst / Trader
Focus on the Intelligence Interface, signal interpretation, and data upload workflows. Start with the example queries, then work toward your specific market questions and hypotheses.
→ energyOS Chat
Operations / Grid
The pipeline monitor and nodal LMP feeds are your primary tools. Anomaly detection and regime state signals are particularly relevant for dispatch and load forecasting decisions.
→ Grid use case
Admin
Configure workspace settings, provision users, set role-based access controls, and review the audit log. Start in Workspace Settings after your onboarding session.
→ Security & governance

Security & governance

energyOS is built on a continuous compliance architecture. Rather than point-in-time audits, the infrastructure generates real-time evidence against SOC 2 Trust Services Criteria and ISO 27001 requirements. Security controls are not bolted on — they are part of the AWS infrastructure layer from day one.

Infrastructure security controls

Control / 01
Configuration recording
AWS Config records every infrastructure state change in real time. All configuration history is stored in an isolated audit vault and is available for forensic analysis by date, region, and resource type.
Control / 02
Continuous security monitoring
AWS Security Hub runs AWS Foundational Security Best Practices continuously — covering IAM governance, storage encryption, public access controls, and network security group rules.
Control / 03
Evidence aggregation
AWS Audit Manager aggregates evidence from Config, Security Hub, and CloudTrail against the SOC 2 framework — Security, Availability, and Confidentiality criteria. Evidence is available for export at any time.
Control / 04
Identity governance
All permissions follow a Group-Based Access Control model. No permissions are assigned directly to users — they are inherited via groups. Deployment access is managed through a dedicated service account with scoped permissions only.

Data handling principles

PrincipleImplementation
No persistent client dataSession data and uploaded files are processed in-memory and discarded at session end. Nothing is written to permanent storage.
All compute within AWS boundaryInferences are calculated within the AWS environment. No data is sent to external model endpoints.
Encrypted in transitAll traffic is SSL-enforced. Non-encrypted HTTP access is denied at the bucket policy level.
Audit trailAll analyst actions and system events are written to an immutable audit log accessible to admin users.
Least privilegeAccess controls follow least-privilege principles. Admin functions are scoped to named groups, not open to all authenticated users.
Compliance posture

The platform maintains continuous SOC 2 alignment. If your organisation requires formal compliance documentation as part of vendor onboarding, contact the REDR Labs team — evidence exports are available on request.

Troubleshooting

Most issues encountered during early use fall into a small set of categories. If something isn't behaving as expected, check here first — and reach out to the REDR Labs team directly if the issue persists.

  • energyOS Chat says it doesn't have enough signal to answer my question Chat
    This is expected behavior when a signal is stale or unavailable for the specific contract, hub, or time horizon you're asking about. The platform will not speculate beyond its signal context. Check the Pipeline Monitor to see the freshness timestamp on the relevant signal category. If signals are outdated due to a feed delay, the REDR Labs team can advise on expected refresh timing.
  • My uploaded data returned no scalar correlations Data ingestion
    Scalar correlations require numeric columns with sufficient variance and row count. If your file contains primarily categorical data, text fields, or fewer than 20 data points per column, the correlation engine may return no results. Try uploading a subset of your data focused on the numeric time-series columns most relevant to your market question. If you're unsure how to structure your data for best results, this is worth covering in your onboarding session.
  • The pipeline monitor shows a stage as pending Pipeline
    A pending stage indicates that data for that stage has not completed its latest processing cycle. This can occur due to upstream feed delays from the source provider. Pending states are normal for low-frequency signals (e.g. weekly storage reports). If a stage remains pending for more than the expected cycle duration shown in the monitor, contact the REDR Labs team — we monitor pipeline health proactively and are likely already aware.
  • I can't see the Workspace Settings or Audit Log Access
    Workspace Settings and the Audit Log are restricted to users in the admin role. If you need access, contact your organisation's designated workspace admin to have your role updated. If you are the admin and cannot see these sections, contact REDR Labs — this may indicate a provisioning issue with your role assignment.
  • Chat responses seem to ignore my uploaded file Chat · Data ingestion
    Uploaded data is injected into the session context at the time of upload. If you start a new session after uploading, the file will not be present in the new context — it must be re-uploaded. Within a single session, if Chat does not appear to reference your uploaded data, try explicitly referencing it in your query: "Using my uploaded position data, what correlations are present with current storage signals?"
  • I need to update my team's feed configuration Admin · Configuration
    Feed configuration can be updated by admin users via Workspace Settings. For significant changes — such as adding a new commodity vertical or adjusting anomaly detection thresholds — we recommend looping in the REDR Labs team, who can advise on configuration best practice for your use case. Configuration changes take effect at the start of the next pipeline cycle.
Still stuck?

Reach out directly to the REDR Labs team at info@redrlabs.com. As a partner, you have a direct line — not a support ticket queue. We aim to respond to all platform issues within one business hour during market hours.

What to read next