Alpaca Analysis

The film

The pre-audit, in an afternoon

What used to take weeks of analyst time and a seven-figure engagement now takes a few hours, with your own people in the loop.

The pre-audit, then and now

Then With Alpaca

Time to findings Weeks to months A few hours

Analyst effort Many hundreds of hours An afternoon of review

Outside spend A seven-figure engagement Kept in-house

Data freshness A quarter old on delivery Refreshed on your schedule

Next year Pay again Re-run on a schedule

How it works

From raw records to closed actions

Five stages, each one reviewable. Alpaca does the work; your people make the calls.

01 Hooks
02 Snapshots
03 Analyses & reports
04 Scenarios
05 Actions

Hooks

Connect to the records where they live.

A hook is a saved, reusable connection to a source of records. Alpaca sets it up with you in conversation, including the filters people will pick from each time it runs.

EAM and ERP databases, read live: work orders, assets, requisitions, purchase orders, cases, downtime
Any REST API, with credentials registered once
Connected apps such as Jira, Confluence and SharePoint
Spreadsheet uploads, plus a document library for procedures and reports
Filters by site, rig, date window or code list, with rolling dates like "the last 7 days"

Snapshots

DBA-grade data work, done for you.

Every hook run lands as a real table in your own SQL Server. From there Alpaca does the data preparation a good database administrator would do, and keeps the receipts.

Typed columns, with a preview and total row count before anything is saved
Derived tables written in SQL: joins, aggregates and flags across snapshots, indexed where it counts
Every derivation is stored and replays without the AI, so the numbers are reproducible
Replace, append or upsert refreshes on a daily, weekly or monthly schedule
Lineage on every table: which hook, which parameters, every refresh and its row counts
Refreshes fail loudly when a source changes shape, instead of quietly filling gaps

A derived snapshot, as Alpaca writes it (illustrative)

-- Overdue safety-critical items by rig and system
SELECT  d.Rig, e.System, COUNT(*) AS OverdueItems
INTO    reports.Snapshot_OverdueByRig
FROM    reports.Snapshot_OpenDeferrals d
JOIN    reports.Snapshot_Equipment e
        ON e.EquipmentId = d.EquipmentId
WHERE   d.DueDate < SYSUTCDATETIME()
GROUP BY d.Rig, e.System;

CREATE INDEX IX_OverdueByRig_Rig
    ON reports.Snapshot_OverdueByRig (Rig);

Stored with the snapshot and replayed on every refresh, without the AI.

Analyses & reports

Findings with the evidence behind each one.

Alpaca works the problem in steps, bringing in Specialists (expert profiles your team writes) where the judgment matters, and writes a report you can defend.

Multi-step analysis plans, run in order, with any failed step retried on its own
Report templates that set the required sections, checked after every run
Governed process flows with review points: the run waits for a person to accept or return it
Scheduled re-runs that produce a new dated report and can email it
Recommendations marked in the report, ready to become actions

Scenarios

Tools, not just reports.

Ask for a tool and Alpaca builds one: an interactive page with filters, sliders and charts, working from your snapshots. Reports get read once. Tools get used every day.

Planners, risk explorers, living assessments and audit explorers
Live mode queries your data on demand; baked mode refreshes on a schedule
Published with approval and shared with a link
An Alpaca beside every tool that explains the page and checks the numbers

Actions

Closure is the point.

Recommendations become actions with an owner and a due date. Every action links back to the report section that produced it, so the audit trail writes itself.

Routed to a department and, optionally, a named person
Open, assigned, complete or cancelled, with completion time stamped automatically
Due dates, overdue flags, working comments and full status history
Everyone sees their own actions; leaders see the whole list

Governance

Built to stand up to an auditor

The same controls you would expect from a well-run analytics team, applied to every run.

Approval to publish

Reports and tools reach readers only after an approved publisher signs off.

Review points

Governed analyses pause for a named reviewer, and every decision is logged.

Full lineage

Every table, refresh, analysis step and action status change is recorded.

Sandboxed database access

Alpaca writes only to its own reporting schema, as a restricted database user.

Your database, your tables

Snapshots are tables in your own SQL Server. Analysis steps send the models only the rows they need.

Enterprise AI

Enterprise-grade AI models served through Microsoft Azure, with the right model chosen for each analysis step.