Elephant Hawk — Opportunity Scorer

A transparent, multi-criteria decision tool. Drag the sliders to score an opportunity across all layers; the verdict, method comparison and charts update live. Hover any termiHover the dotted terms anywhere in the tool to see a plain-language definition. for its meaning. Add your own criteria in the Query tab; see how the numbers are built in Methodology & Sources. SEEKER (the companion finder) and the calibration harness share this same engine.

What are you evaluating?

Scoring as: — the Deal & Fit layers and AI guidance adapt to whose seat you're scoring from.

Type the opportunity above and click Run live research & scoring. In a normal browser it pulls real FDA / ClinicalTrials.gov / PubMed records and — with an API key set in AI engine settings below — scores every layer with sourced values. Inside the Claude app it hands off to Claude's own research tools. (The in-app preview can't fetch external data — open the downloaded file in a real browser.)

AI scores every layer and draws the graphs for you. You only move the sliders if you want to fine-tune — let me adjust the scores manually.

AI engine settings — use your own API key (works in any browser, no Claude needed)

Your key is remembered in this browser only (localStorage) and used solely for a direct browser call to that provider — never sent anywhere else. Requires opening the file in a real browser.

Hard gates (uncheck = fails): CapabilityiDo we plausibly have, or can deliberately fund, the core technical competence to build this? A "no" is fatal regardless of how attractive the market is. Regulatory pathwayiIs there an achievable route to clearance/approval for this device class and intended use (e.g. FDA 510(k)/De Novo/PMA, EU MDR, CDSCO)? Focus-area fitiDoes it sit inside, or adjacent to, one of our chosen focus areas, where effort compounds and we build credibility? Impact overrideiExceptional clinical/mission impact lever. Use only to consciously bypass a failed gate or weak score — never as an automatic pass.

Focus multiplier iWeights up opportunities sitting firmly in a focus area (where time compresses and effort compounds) and down those outside it. 1.00 = neutral.: 1.00

Headline score — TOPSISiTOPSIS measures how close the opportunity's profile is to an ideal (all-100) versus the worst case (all-0). It rewards balance, so a spiky profile scores lower than an even one at the same average. Used as the decision driver. (gated)

—

Decision-method comparison (MCDA)iMulti-Criteria Decision Analysis: the same layer scores run through four aggregation methods. Agreement = robust call; divergence = the answer depends on method, so use judgement.

Same scores, four methods. WSM is the intuitive number; TOPSIS drives the verdict; Maximin/WPM expose weak links.

India · US · GlobaliCompares how favourable it is to develop and commercialise this in India, the US, and Globally — dimension by dimension. India favourability feeds the composite; the three-line radar shows where each geography wins.

weight 7%

—

Why it's here: Where a programme is built changes its cost, speed, regulatory path, manufacturing and market. This layer makes the India-vs-US-vs-global trade-off explicit instead of assuming a single base. Each dimension is scored for India (red), the US (green) and Global (blue).

All layers — single-profile radariEvery layer's score plotted on one polygon — the overall shape of the opportunity. A big, round shape = strong and balanced; a spike pulled toward the centre = a weak link (a red flag the average can hide).

All 14 layers summarised as one shape. Updates live as scores change.

Saves a standalone .html snapshot with the current scores, all tabs, graphs and clickable source links — open or share anywhere, works offline.

Add your own criterion / query layer

Have a factor the framework doesn't cover? Add it as a custom layer. It folds into the composite, the verdict and the MCDA methods exactly like the built-in layers.

Criterion name

Meaning / note (shown on hover)

Weight %

Score 0–100

No custom criteria yet.

Adding criteria proportionally rescales all weights so they still sum to 100%. Remove a criterion any time; the model recomputes instantly.

Why these scores — section-by-section justification

Populated when you score with AI. Each layer gets the reasoning, the hard facts / figures behind the numbers, and sources to verify.

Verify before you rely on it. These justifications and citations are AI-generated — treat them as leads to check, not final references. Confirm any hard number (market size, IRR, regulatory class, CPT code) against its primary source before using it in a decision.

Score with AI to populate per-layer justifications.

Alternative solutions (evidence-based)

Other ways to solve the same clinical problem — devices, drugs, procedures, and traditional remedies — included only where there is real scientific evidence. Each is rated by strength of evidence (R/Y/G) with a source. This is the "compared to what?" check: a new product must beat the best existing alternative, not just exist.

No alternatives listed. Run live research to populate.

Competitor landscape

Structured competitive set — the "compared to what?" map. Derives the four Competition-layer axes from real rivals instead of a guess.

Positioning map — where the white space is

Two-axis map. Each dot is a rival; the red dot is us. The dashed box is the emptiest quadrant — open ground.

IncumbentChallengerAdjacent / braceSubject (us)

Feature comparison

Each rival on the dimensions buyers ask about. green favourable · amber mixed · red weak · grey TBD.

Where we can win

Hard losses to avoid

Player-by-player

Market context

Reimbursement constraint

Live evidence — real public data, no Claude needed

Runs automatically a moment after you type the opportunity. Pulls real records from US FDA device databases (510(k), PMA, classification, recalls), ClinicalTrials.gov and PubMed, lists them with source links, and auto-sets the data-backed sliders. Each set value shows its basis and reference below.

Search term (device / technology / condition)

If you see network/CORS errors, open this file directly in a browser (double-click the .html) — the in-app preview blocks external data calls. Data: openFDA & ClinicalTrials.gov, public APIs.

Methodology & Sources

How the numbers and logic are built, with references for further study.

Transparency note. The per-opportunity scores are judgement calls (yours, or an assistant's) on a 0–100 scale — they are not measured constants. What is grounded in the literature below is the framework logic: which layers matter, how criteria are aggregated, and the regulatory/financial/clinical definitions behind each axis. Default layer weights are deliberate starting points for calibration, not empirical truths — tune them against deals you already have opinions on (use the calibration harness).