Personalization
How Clarity learns a member's positioning — a book of conviction theses, kept current by a feedback loop — and uses it to personalize every answer.
By default, Clarity answers the same way for everyone: grounded in evidence, cited, honest about gaps. Personalization layers a member's own positioning on top of that, so the same engine can explain what a development means for this person — without ever loosening the citation rules.
Personalization is built from three things Clarity keeps about a member: their suitability (goals, risk appetite, time horizon, values), their convictions (what they actually believe about live market questions), and their thesis book (the positions they hold and why). This page explains the model and how it stays current.
Personalization is enabled per organization. With it off, Clarity behaves exactly as documented elsewhere — grounded, cited, and impersonal. With it on, the same answers gain a suitability- and thesis-aware layer.
The thesis book
A member's positioning is not a single statement — it's a book of conviction theses, and they differ by holding period.
A long-horizon thesis is the member's core: the things they hold for years ("broad markets compound" → a global equity fund, an index, steady names).
A shorter-horizon thesis is a tactical tilt: a view they're acting on right now ("AI demand keeps compounding across the supply chain").
"Core" isn't a separate kind of thing — it's simply a long-horizon conviction. Every entry in the book is the same shape:
A plain-language statement — the core idea in a sentence a non-investor understands.
A narrative tree — the idea broken into the pillars that have to hold for it to be right (for an AI thesis: memory → chips → networking → data centres → the names that express each), each pillar grounded in cited evidence.
An allocation sleeve — what share of the member's positioning this thesis represents, and which holdings express it. The same holding can appear in more than one thesis.
Its supporting and contradicting convictions — the market questions the member has weighed in on that argue for or against it.
A lifecycle — every thesis is versioned, so "how my view changed" is answerable: a thesis is proposed, endorsed, and later superseded by a revised version, with the history kept.
Clarity proposes, the member endorses
A member never starts with a blank page. Clarity proposes a thesis — grounded in the evidence it can cite, shaped to the member's suitability — and the member confirms or edits it to endorse it. Only an endorsed thesis becomes part of the book and starts shaping answers.
The split is deliberate: Clarity suggests, the member decides. Clarity is the analyst that drafts and re-drafts the thinking; the member owns what they actually believe. A firm can require an advisor to review a proposal before the member sees it.
The feedback loop
A thesis book that never updates goes stale. Clarity keeps it current with a loop between the member's convictions and their theses.
Convictions are real signals, not a survey. When a development bears on something a member holds or believes, Clarity surfaces a concrete market question tied to it and records the member's answer with their confidence.
A conviction can challenge a thesis. When a new answer argues against a thesis the member has endorsed, Clarity re-evaluates that thesis against the latest evidence and, if its view has materially changed, drafts a revised version — keeping what still holds, changing what doesn't. The member is never overwritten; the revision waits for them to endorse it.
Endorsing a thesis invites new convictions. Committing to a thesis is itself a signal: Clarity can surface fresh conviction questions about that thesis's drivers, so the member keeps weighing in on the things that matter to their book.
Theses are re-checked on a cadence that fits their horizon. A short-horizon tilt is revisited often; a long-horizon core rarely — a quiet thesis isn't churned for no reason.
Explanation vs. advice
Personalization changes what Clarity explains, not whether it gives orders. The honesty rules are unchanged everywhere: every personalized point still rests on cited evidence, and nothing is invented to fit a member's thesis.
Suitability-aware explanation (the general, single-stock, and portfolio personas). Clarity frames a development for this member — which of their theses it touches, how it sits with their goals and risk appetite — while stopping short of telling them to buy or sell. It explains; it doesn't instruct.
The advisor persona goes further for firms that enable it: a top-down, personalized read that works from the member's endorsed theses and holdings. Even here, every recommendation is tied to a cited catalyst and the member's suitability, never to price action alone, and is subject to the firm's review and compliance gates.
What a read can act on depends on the asset class. Which asset classes carry through to acting on a position is gated by what the firm has enabled for a member. Equity positioning is explanatory and tracked only — a read on a stock is advice for review, never an instruction that executes. Only enabled executable classes — currently crypto tokens — can carry through to acting on a position. By default a member is scoped to tokens, with equities added as a tracked, advisory class per the firm's configuration.
Related
Surfaces & analyst personas — the personas this layer rides on
Pipeline — how Clarity grounds and answers
Portfolio context — the holdings model behind a member's positions
In the admin panel — where a firm enables and reviews
Last updated