Intake methodology

Most knowledge-base products ask you to start from a blank form. KlyHub does the opposite: an AI conversation walks you through four phases, asks questions in plain language, and stores the answers as structured knowledge in the right ontology layer.

The methodology is called the 4-phase intake. The phases are: Foundation, Context, Exploration, Optimization.

Why phases (and why this order)

The phases match the order in which a knowledgeable advisor would think through a new company:

Foundation first — until you know what the company is, you can't usefully ask about its market.
Context next — market shape is the lens through which strategy decisions make sense.
Exploration third — only once we agree on market shape can we dig into how the company actually operates within it.
Optimization last — gaps and priorities only become tractable when the previous three phases have surfaced enough context to anchor them.

Every phase outputs facts into one or more layers of the 5-layer ontology. The intake never asks you to think in terms of layers — it asks plain questions and routes the answers for you.

Phase 1 — Foundation

Goal: establish the company's identity. Outputs into the Core layer.

Sample questions you'll see:

What's the company name and the one-sentence elevator pitch?
What problem do you solve, and for whom?
What's the legal entity (LLC, Inc., Ltd., etc.) and the founding year?
Who are the founders, and what's each one's role today?

Foundation is the shortest phase — usually 5–10 minutes. It exists to anchor every subsequent question.

Phase 2 — Context

Goal: map the company into its market. Outputs into the Market layer.

Sample questions:

Who are the top 3 competitors, and where do you differentiate?
What category does the company belong to (industry, sub-industry)?
What's the addressable market — by geography, by buyer persona, by use case?
What's the dominant pricing model in your category, and where do you sit relative to it?

Context can take 15–30 minutes if you're being thorough. You don't have to finish it before the AI client can use what's there — partial Context is still queryable.

Phase 3 — Exploration

Goal: capture how the company actually runs. Outputs into the Motion and Operations layers.

Sample questions:

What's your primary go-to-market motion (PLG, sales-led, channel-led, hybrid)?
Walk me through your acquisition funnel — which channels, what conversion rates?
What does the customer onboarding flow look like, end to end?
What's your team structure and headcount by function?

Exploration is the deepest phase. KlyHub branches the conversation: each motion you have generates its own follow-up thread, so a company with both a PLG motion and a sales-led motion gets two parallel sub-interviews.

Phase 4 — Optimization

Goal: surface gaps and priorities. Outputs into the Memory layer (with cross-references back to earlier layers).

Sample questions:

What's the biggest constraint on growth today?
Where in the funnel are you losing the most customers, and why?
What changed in the last 30 days that an AI assistant should know about?
What's a question you'd love to ask the company that you don't currently have an answer to?

Optimization is also where ongoing maintenance lives — every month or so, KlyHub asks you a short Optimization update, and the new answers append to Memory with timestamps.

Re-running the intake

The intake is resumable, branchable, and rewindable. You can pause mid-phase and pick up next week; you can fork a phase to capture a new motion or a new market segment; and you can edit any saved fact directly from the workspace UI.

For the Pack/Prompt/Template architecture that makes this possible, see docs/METHODOLOGIES.md in the KlyHub repo.

What the AI client sees

Every fact captured during intake is queryable via the klyhub.query MCP tool. The tool takes an ontology layer (or * for cross-layer queries), an optional filter, and returns structured JSON with citations.

For the data shape itself, head to Ontology.

Intake methodology

On this page