AI-ready data & exports

Every report comes with the structured data behind it, ready for a model or an integration to read as it stands, and whatever you measure can leave the platform as a file too.

1 · What it is

Behind every report on the platform is a structured JSON payload: the same numbers the report shows, in the form a machine reads. You can hand that payload to a large language model, an ERP or a BI tool and have it read directly, without the data-engineering cleanup that usually comes first. It means you can ask your own questions of the data, rather than only the ones a dashboard was built to answer.

2 · What’s in it

Alongside the data, the payload carries a short piece of embedded guidance: what the fields mean and how they relate, so a model doesn’t have to guess at the shape of your data. And every figure traces back to the meter it was measured on, so an answer built on the payload is built on measured data rather than an estimate.

3 · What you can do with it

The payload supports three levels of use, from something any user can do today to integrations you or Augos can build on top.

Ask a question today

Download a report’s JSON, give it to a secure model, and ask something specific. A financial controller hands over twelve months of half-hourly data and asks what the peak demand was in July, and what a different tariff would have charged for that single spike. An engineer uploads a week of power-factor data and asks for every period it dropped below 0.85 for more than fifteen minutes, matched against when the chillers started.

Automate a report

A scheduled agent can take the JSON at set intervals and produce a written summary, the kind of thing an analyst does by hand. One reads last week’s consumption and cost data every Monday and drafts a three-line summary of the most expensive events before stand-up; another turns a month of minute-level data into a Scope 2 carbon table on the first. These are integrations you build on the export, not features switched on out of the box.

Connect it to your systems

Further out, an agent can act on the data: drafting a maintenance work order in your CMMS when it sees a sustained phase imbalance, or signalling your building management system to shed non-critical load as demand approaches your notified maximum. These are advanced, build-on-top scenarios that show where AI-ready data leads, rather than current platform features.

4 · Other formats

If you would rather work with the data yourself, every report and dataset also exports as a file. Pick by what you are doing with it:

• CSV or Excel, for your own analysis, raw or aggregated down to the minute.
• PDF, a fixed record of what a report showed, for sharing and sign-off.
• JSON, the same machine-readable payload above, for a model or an integration.
• The API, when you want a live feed straight into your systems rather than a file.

5 · Access and privacy

Getting the data out doesn’t widen who can see it. The payload obeys the same point-level permissions as the platform: you can only pull the points you are cleared for, and you cannot extract a portfolio-wide payload without portfolio-level access. See Roles & access.

One caution when using external models. Don’t upload raw Augos data to a public, consumer-tier AI: consumer models may keep what you upload to train future versions, and a minute-level load profile can reveal production volumes and shift patterns. Use an enterprise model with a zero-retention agreement, or one hosted on your own infrastructure, and strip out sensitive metadata such as internal codes before you pass the data on.

6 · Common questions