Research that shows its work

See what matters before you move capital.

Prism turns a ticker or portfolio into a scored brief so you can spot concentration risk before it costs you, see which assumptions are carrying the thesis, and know exactly where confidence stops.

Open Workspace View Example Brief
Get notified when new features ship No account required.
Spot concentration risk early Know what is driving the target Stop confusing narrative with evidence Expose ETF overlap fast Confidence limits stay visible
Portfolio risk Spot concentration, sector crowding, and ETF overlap before the written commentary starts sounding confident.
Thesis discipline Know exactly which assumptions are carrying the case and which weak metric is keeping the call from upgrading.
Coverage honesty When evidence is thin, Prism cuts the target, lowers conviction, and says so directly instead of filling space.

Example brief

What a Prism brief actually looks like.

This is a live-rendered sample of the output Prism is designed to produce: a scored read, plain-English evidence, and a visible confidence limit.

Sample output · illustrative

NVDA · NVIDIA Corporation

HOLD
Current price context$934.20
Research score7.4 / 10
Strongest signalBusiness quality
Weakest signalValuation discipline
Business quality 9.0

Demand, margins, and platform position still carry the case.

Valuation discipline 4.7

The multiple leaves less room for a normal reset than the story suggests.

Risk profile 6.3

Execution is strong, but the stock is still exposed to expectation compression.

Evidence depth 8.2

Prism has enough price and company context to defend a structured read.

Evidence summary

  • Data center demand is still the main reason the score holds up.
  • Margins remain high enough to keep business quality above narrative-only hype.
  • Valuation is the weak point because the market is already pricing a lot of the upside.

What would change the call

  • A cleaner reset in valuation without a break in demand would improve the setup.
  • Margin compression or evidence of order digestion would weaken the brief quickly.
  • Prism would only widen the target range if support improves in both fundamentals and price.
Confidence limit: Prism will not publish a stretch target here because valuation support is still thin and the next leg of demand durability is not fully proven.

The method

How Prism scores, what it uses, and where it stops.

Prism is not trying to sound like an analyst. It is trying to keep the analysis structured enough that you can challenge it quickly.

Inputs

Specific inputs, not vague “AI context.”

  • Live quote data, 52-week context, and key valuation ratios.
  • Company profile data, sector classification, and public market identity checks.
  • Portfolio holdings math including weights, crowding, ballast, and ETF overlap.
  • For public names, filings and fundamentals shape the read; for private names, Prism stays explicitly lower-confidence.
Score construction

Four dimensions decide the read.

Business quality
35%
Valuation discipline
25%
Market behavior
20%
Risk profile
20%

Evidence depth is a guardrail, not a bonus. It constrains how far the verdict and target can go.

Evidence limits

“Confidence limits” means Prism knows when to stop talking.

  • If market identity is weak, the brief stays limited or does not run.
  • If the score is supported but the target is not, Prism can keep the rating and hide the range.
  • If portfolio breadth is thin, the diversification score explains why instead of pretending the book is balanced.
What Prism does not do

Honest constraints are part of the product.

  • No guaranteed price predictions.
  • No blind buy or sell signals.
  • No fake precision when the evidence cannot defend it.
  • No selling or sharing of your analysis data.

Workspace

Open the desk when you want the real work surface.

The app route is built for doing the work fast: one active tool, tighter inputs, live market context, and briefs that stay structured enough to inspect or copy.

Analyze a ticker Compare two names Check portfolio health Run company diligence Rank curated themes Save names on-device
Open Workspace Get Product Updates

Trust

Built like a research tool, not a finance toy.

Serious products earn trust with clear limits, live maintenance, and a real person behind the work. Prism now shows all three.

Founder

Why Kaiden Leung built Prism

Kaiden built Prism after seeing too many research tools swing between generic AI prose and false certainty. The goal was a tighter desk: show the score, show the weak point, and keep the limit on the call visible before conviction gets ahead of the evidence.

Contact GitHub
Product status

Beta, live, and actively updated

Current status: v1 beta. Updated April 2, 2026 with clearer method copy, a live sample brief, stronger social metadata, and tighter workspace UX.

Landing page upgrade New
Added a value-first hero, concrete sample brief, and a plain-English method section.
Trust architecture New
Added founder context, privacy language, product status, and a visible changelog.
Portfolio calibration Updated
Diversification scoring now treats focused portfolios more fairly and gives breadth credit for sector and geographic spread.
Policies

Clear promises, clear limits

Privacy: We do not store or sell your analysis data. Recent briefs and saved names stay on this device unless you choose to copy them.

Disclaimer: Prism is a research tool, not financial advice. All outputs are for informational purposes only.

Stay close to the product

Want new releases without checking back?

Join the early updates list and Prism will open a prefilled email to request access or feature notifications.

Opens your mail app with the request prefilled to Kaiden.