Free to build.
Or built alongside you.
Open source forever. The Ark’s design, described on this site, and its software are free to copy and build, and always will be: the answer engine, the tools that download and verify every file, and the catalog of everything on the drive, published under the Apache 2.0 licence. The book and supported deployment are optional: they fund the free version rather than gate it.
The design, the scripts and the reasoning are open and stay open. Some people want exactly that and nothing else. Others have the hardware budget but not the fortnight, and would rather have someone who has already made the mistakes sitting on the call.
Everything here is free, forever.
The design on this site, the software that runs the node, the scripts that fetch and verify every file, the catalog of everything on the drive with its checksums, the printed recovery procedure and the operator’s manual. No account, no key, no telemetry, nothing to renew.
The node software is published. The answer engine, the tools that download and verify the library, and a starter kit are on GitHub, free under the Apache 2.0 licence. On a Windows PC with an NVIDIA card of 6 GB or more, one command takes an empty folder to a first answer that cites its page, in about an hour.
There is a book. The Ark of Noahledge walks the same build in eighteen chapters, starting on the laptop you already own and explaining each idea at the moment it is needed. Free path or guided path — the specification is the same. Read about the book →
Someone has to run itA backup only its builder can run is not a backup. The builder can always operate it, so nothing ever visibly fails — until the day it matters. There is an operator’s manual written for whoever ends up holding the drives rather than for the person who assembled them: every command literal, every expected output shown, and every failure routed to a page that names the symptom. English and Spanish, with a PDF sized to print and keep beside the cold copy. Read the operator’s manual →
One thing to knowThe corpus has two editions — a Community edition, which is this one, and an Institutional edition used in supported deployments. They differ by a handful of components, for licensing reasons rather than capability ones. The reason is set out under Institutions →
Guidance, not a black box.
For an individual, a small team or a local organization with some technical capacity who would rather not do the whole thing alone. You keep the hardware, the drives and the decisions; what you are buying is someone else's scar tissue.
Sold as blocks of consultancy hours — typically five, ten or twenty — drawn down as you need them rather than spent on a fixed schedule. What they usually cover:
- 01A discovery call. Which rung actually fits the site, whether it needs solar at all, and what the room and the power situation impose.
- 02A remote build walkthrough. Working alongside you through acquisition, verification and first boot rather than handing over a document.
- 03Index-build troubleshooting. The step that most often goes wrong unattended, and the one that costs days of compute to redo.
- 04Voice pipeline setup. Including the espeak-ng linkage that silently removes half the Spanish coverage when it is missed.
- 05A staff training session. For the people who will actually stand in front of it — how to ask, how to read a source line, and when to disbelieve it.
Larger than that?Deploying across many sites, or on behalf of a government, city or institution? See Governments & Institutions →