Not a report. Not nine hours of video. A structured body of your organization's operating knowledge, written down at the level of decisions rather than steps, in plain files any tool can read.
You own it. You can read it without us, index it with whatever you buy next, and still open it in ten years.
Your existing documentation already covers the steps. What is missing is why the steps are what they are, and what the specialist looks at before choosing between them. That is what we go after.
The real branching logic behind a judgment call. What the specialist checks first, what makes them change course, and which inputs they quietly distrust. Written as an explicit model somebody else can follow, not as a paragraph of advice.
The signals experienced people read without noticing. A sound, a temperature drift, a number that is technically in range but wrong for this line. These almost never appear in a manual, and they are usually the first thing lost.
What goes wrong, how it announces itself, what the wrong fix looks like, and what it costs. Including the failures that were designed around years ago, which is why nobody remembers the design was deliberate.
The step that exists because of a customer contract. The sequence that quietly prevents a damage claim. The reason nobody trusts Monday's cycle count. A newcomer cannot tell these apart from sloppiness, so they get optimized away.
For the legacy systems nobody can fully explain: why it was configured this way, which behaviour is intentional, what was tried and rejected, and which rules it silently enforces. This is what a replacement business case needs and never has.
Every entry records who it came from, when, and in what context. So a reader two years out knows whose judgment they are relying on and how current it is, instead of trusting an anonymous wiki page.
The vault ships as Markdown with YAML frontmatter. Tags in the frontmatter carry the structure, so the same files are readable by a person and parseable by a machine. A JSON index ships alongside them.
That means no proprietary container, no viewer to license, and no vendor whose pricing decisions become your archival risk. The files are static, portable, and diffable, so you can put the whole vault under version control and see exactly what changed and when.
We are a services firm. We are not trying to make you dependent on software we happen to sell.
Machine-readable is not a buzzword here. It is the difference between an asset and a folder, and it is decided by structure, not by file type.
Whatever retrieval or AI system you buy in three years, it will need clean, tagged, consistently structured source material. That is the expensive part and it is the part nobody can generate after the fact. Get the structure right once and every future tool can use it.
Because entries follow a consistent shape, you can see what is missing. Which decisions have no failure patterns recorded. Which systems have no rationale. An unstructured pile of documents cannot tell you what it does not contain.
Nobody reads a two-hundred-page handover document. They have a specific question at 4pm on a Tuesday. Tagged, granular entries can be retrieved and cited. A narrative report cannot.
A general model pointed at your file share has nothing to retrieve on the tacit material, so it invents something plausible. Grounded in a real vault with real provenance, the same tool can cite a named specialist instead of guessing.
The vault is the source. These are the ways your people and your systems use it.
Delivered into Databricks, Snowflake or Microsoft Fabric as structured tables or documents their tools can read. Your data shows what happened. The vault, sitting beside it, holds why.
A standalone search tool over the same open files, run on your own systems or hosted by us if you prefer, so the remaining team and new hires can ask a question and get an answer that cites who it came from.
Procedures and training material written from the vault for successors, with the reasoning kept, not just the steps.
The replacement hire. Someone competent but new, three weeks in, who needs to know why the process is shaped this way before they change it. They can read the vault directly. It is prose, not code.
The remaining team. The people who inherited the questions. They get an answer that cites a named source and a date instead of a shrug and a guess.
Your own systems. Whatever you point at it. The structure is designed so a retrieval layer, a search index, or an AI assistant can consume it without a preprocessing project first.
An outside party, when you choose. A systems integrator scoping a replacement. An acquirer doing diligence. An auditor. You decide who sees it, because you own it.
Most firms are vague about this until the contract arrives. We would rather be specific up front, because it is the part that decides whether this is worth doing at all.
Everything you give us and everything your people say stays yours at all times. Recordings, transcripts, documents, drawings, exports. We hold a limited licence to do the work and nothing more. This is not conditioned on payment, on completion, or on anything else. If the engagement ends badly on day three, your material is still yours and you get it back.
Unconditional. Not contingent on the engagement.
The work product we build from that material, meaning the structured, tagged, indexed vault, transfers to you on signature and payment in full. Until then we hold it. That is the honest position for work sold on a time-and-materials basis, and it is stated plainly in the agreement rather than buried.
Conditioned on payment. Then it is yours outright.
Nothing you tell us trains anything shared. Your knowledge is never used to train, fine-tune, or improve any model or product made available to another client or to the public. Model adaptation, where it happens at all, happens only for you.
If we do not convert, we delete. Where a scoped pilot does not proceed to a full engagement, we delete the work product within 30 days and certify the deletion in writing on request.
Both are time-and-materials engagements. Neither is a subscription, a seat count, or a licence. What separates them is scope and how much of your organization is in the room.
Project based, scoped to a defined body of knowledge. Typically one specialist, one legacy system, or one process area where you can already name the person and the deadline. We scope it, run the elicitation on site, and hand over the vault for that scope.
This is the right starting point for most organizations, including the ones that intend to do more later. It proves the method on something that matters without committing the whole programme.
Billed at $150 per hour on time and materials, with a 40 hour minimum. A typical engagement runs 3 to 5 days on site, with structuring and drafting done remotely.
A programme rather than a project. Multiple specialists across several sites or functions, run as a rolling schedule against a knowledge risk register, usually because the retirement exposure is not one person but a cohort.
Includes the governance that scale needs: a prioritised register, review cadence with your leadership, consistent structure across sites so the vaults compose rather than fragment, and handover into whatever systems you already run.
Same $150 per hour on time and materials. Programme floor and duration are set during scoping.
Both start the same way, with a conversation about who is leaving and what they carry. Scope comes out of that, not out of a package you pick from a table.
Tell us who is leaving, what they run, and roughly when. That is enough to have a useful first conversation about scope.
Applied Continuity LLC dba KnowledgeBricks. Indianapolis, Indiana.