Skip to content

Getting Started — Overview

Time to first query end-to-end : 30 min walltime. Time to first impression (no install) : 2 min.

Welcome. This 9-page guide takes you from zero to your first end-to-end query on AKKO. Whether you are a data engineer, an analyst, a steward, or an admin, every step is short, copy-paste friendly, and validated against the live demo every release.


What is AKKO?

AKKO (Analytics Kernel, Keep Ownership) is a sovereign data and AI platform you deploy on your own infrastructure and operate forever. It is layer-first : every capability you need to run a modern lakehouse is shipped as a pluggable layer, wired together with single sign-on, governance, and observability out of the box. Bring your servers, your laptop, or your sovereign cloud — AKKO runs the same way everywhere.

The 10 Lego layers

AKKO is built as ten composable layers. You can swap any of them without rewriting the rest.

# Layer What it does
1 Compute Distributed transformations on lakehouse tables, batch and streaming.
2 Query Engine Federated SQL across every catalog, with role-aware row and column filtering.
3 Storage S3-compatible object storage that holds raw, curated, and serving zones.
4 Catalog Open table format + REST catalog that exposes governed, versioned tables.
5 Orchestration Pipeline scheduler with lineage emission, retries, and SLA alerts.
6 Lab In-browser notebook IDE with Python, R, Julia, Quarto, and VS Code.
7 BI Self-service dashboards and ad-hoc SQL Lab with shared assets.
8 AI Natural-language query, RAG, model gateway, evaluation, and inference.
9 Governance Identity provider, policy engine, audit trail, and steward review surface.
10 Observability Metrics, logs, traces, alerts, and a single dashboards portal.

Why ten layers, not one stack? Each layer has a published Protocol contract. You can swap the default engine for a partner engine by changing one Helm value. No re-platforming.


Who is this guide for?

Persona What you will learn
Data engineer Spin up the platform, wire a catalog, run distributed transforms.
Analyst Ask a question in natural language, drill into a generated dashboard, share it.
Steward Review AI-suggested tags, govern PII, run audit reports.
Admin Manage roles, scope data access, enforce identity policy.
Prospect Tour the live demo in 2 minutes with no install.

Two ways to follow this guide

Path A — Live demo (instant)

No install. Open https://demo.akko-ai.com, log in with a seeded persona, and complete the tour in your browser. Start with chapter 01 — Quick start (live demo).

Path B — Local k3d (15 min)

Deploy AKKO on your laptop with Docker Desktop or Colima. You own every layer, every byte stays local. Start with chapter 02 — Install on your laptop.


Chapter map

# Chapter Time Persona
1 Quick start — Live demo (no install) 2 min any
2 Install on your laptop (local k3d) 15 min engineer / admin
3 First login and the 5 personas 5 min any
4 First query — Natural-language SQL 5 min analyst / engineer
5 First dashboard — BI surface 5 min analyst / viewer
6 First data source — Add a catalog 5 min engineer / admin
7 Governance tour — Platform vs Data RBAC 5 min admin / steward
8 Troubleshooting reference any
9 Next steps 2 min any

Conventions

Every chapter follows the same shape :

  • Time estimate at the top (Time : 5 min).
  • Persona badge at the top (Persona : alice (admin)).
  • A short Pre-requisites box.
  • Numbered steps with copy-paste commands and an Expected result for each.
  • Key screenshots at decision points.
  • A Troubleshooting section listing the 3 to 5 most common issues.
  • A Next link that points to the following chapter.

All commands and URLs use a DEMO_HOST placeholder you can override. The default is https://demo.akko-ai.com (the public sandbox). For a local install replace it with your own domain (for example https://demo.akko.local).

Ready? Open chapter 1.