Spirit Intelligence

In short: Spirit Intelligence is the AI layer of Spirit Sight: a Virtual Master Distiller that acts as a digital twin of your master distiller's palate and identifies which barrels will produce a target flavor profile, a morning Daily Brief written from your own data, anomaly detection across environment and books, maturation and demand forecasting, and Vectrant, a grounded BI engine that answers plain-language questions across the platform's live data. It explains and prioritizes; it never files, posts, or blends on its own. Every underlying number stays deterministic and reconciliation-checked.

Deterministic first, intelligent second

Every proof gallon, dollar, and TTB line in Spirit Sight is computed deterministically and verified by automated reconciliation checks that fail if any figure drifts. Spirit Intelligence reads those verified numbers and adds judgment on top: what changed, what needs attention, what is likely to happen next. AI never writes to the ledger, never files a form, and never moves spirit. If you turn every intelligent feature off, the platform still runs your distillery completely.

The Virtual Master Distiller: a digital twin, not a replacement

The Virtual Master Distiller is a digital twin of your master distiller. It learns their palate from your own tasting records, regauges, and per-barrel environment, and then applies that judgment at a scale no person can: name the profile you are after, the one your last great dump had, and it tells you which barrels are developing toward that same profile, scoring caramel, vanilla, oak, spice, fruit, smoke, leather, and honey on every barrel with a readiness window. No one can taste fifty thousand barrels a season; their twin can, and it reports back so the master distiller decides. The predictions live beside the human tasting journal, and the palate on the floor always outranks the model.

The AI Daily Brief

Every morning the Daily Brief reads yesterday's activity, the live alert queue, upcoming TTB filings, cash position, and what the automations handled, and writes a plain-language briefing grounded in those facts alone. It is supervised and read-only: it drafts words, not actions. The owner starts the day knowing exactly where to look first, without opening six modules to find out.

Anomaly detection and forecasting

Intelligence watches for the things a busy operation misses: an environmental drift that will cost proof gallons, a loss rate out of family with a rickhouse's history, a cost that moved when nothing should have. Forecasting extends the same discipline forward, projecting maturation and angel's share, demand, working capital, and tax liability from your measured history rather than industry averages.

Vectrant: ask the whole platform

Vectrant is the grounded BI engine built into Spirit Sight, and it is the piece leadership feels first. Ask in plain language, which barrels are ready to dump, where margin moved last quarter, how much eight-year stock is in a given rickhouse, and Vectrant answers from the platform's live data with the answer tied back to the underlying records. Because every module writes to one data model, that is full platform intelligence: inventory, production, compliance, and the books, queryable in one place. No report request, no analyst queue, no exported spreadsheet that was stale before it was opened.

Connect your enterprise AI

If your company already runs enterprise Claude or ChatGPT, Spirit Sight meets it where it is: a built-in MCP (Model Context Protocol) endpoint lets your AI connect directly to your install and ask grounded questions, barrels, inventory, upcoming TTB due dates, financial position, readiness assessments, with every answer scoped to the credential's access and strictly read-only. Your team talks to its own data in the AI it already uses, and nothing about your operation leaves your install to do it.

Organized as modules, priced by what you use

Spirit Sight is built modularly: inventory, production, compliance, financials, warehousing, and intelligence are self-contained pieces on one shared data model, so a distillery can adopt the platform a piece at a time next to systems it already runs. For operations wary of migrations, that is the point: no big-bang cutover, run in parallel until the numbers have earned your trust, and expand when ready. You pay for the modules you use: predictable pricing for the areas you select, with no per-barrel or per-seat metering, and you switch on more when you need it.

Your data, your install, your rules

Spirit Intelligence runs against your own operation's data on your own installation, on-premise if you choose. Nothing about your barrels, recipes, or books trains anyone else's model. For a craft in Kentucky, the team that built it is in Lexington and can sit across the table; there is no overseas ticket queue between you and an answer.