What this measures
Buyers researching Personal Injury in New York City increasingly consult ChatGPT, Claude, Gemini, or Perplexity before they ever open Google. The AI assistant returns a short list of names — often three to five — with a recommendation tone. That list has become the new shortlist generator for high-stakes buyer decisions.
This report measures who occupies that shortlist right now. We asked each of the four AI assistants a versioned set of buyer-style questions about New York City Personal Injury, parsed the named firms out of every response, and matched them against a pre-harvested pool of New York City Legal entities. The output is a live leaderboard with reproducible methodology, honest confidence intervals, and full verbatim-response traceability. Every ranking on this page traces back to a specific AI response.
The leader — and the gap behind them
In the June 2026 snapshot, Gair, Gair, Conason, Rubinowitz, Bloom, Hershenhorn, Steigman & Mackauf holds the #1 position in New York City Personal Injury AI recommendations with a 8.0% share of citations across all four assistants. The gap to #2 is narrow (1.34× the runner-up's share) — the top of this vertical is contested.
Top-ranked firms — Jun 26, 2026
The top 10 firms by AI recommendation share across all four assistants tested this snapshot. Share is computed from citations in the ranking layer of our prompt set (see methodology).
| Rank | Firm | Share | AI models cited by |
|---|---|---|---|
| #1 | Gair, Gair, Conason, Rubinowitz, Bloom, Hershenhorn, Steigman & Mackauf | 8.0% | 2/4 |
| #2 | Block O'Toole & Murphy | 6.0% | 2/4 |
| #3 | Sullivan Papain Block McManus Coffinas & Cannavo P.C. | 5.5% | 2/4 |
| #4 | The Perecman Firm, P.L.L.C. | 5.1% | 4/4 |
| #5 | Rosen Law Firm | 5.1% | 3/4 |
| #6 | Rheingold Giuffra Ruffo Plotkin & Hellman LLP | 4.3% | 2/4 |
| #7 | Cellino Law Injury Attorneys | 3.2% | 2/4 |
| #8 | Parker Waichman LLP - Personal Injury Accident Attorneys | 3.1% | 1/4 |
| #9 | KDLM | Kramer, Dillof, Livingston & Moore, LLP | 1.9% | 1/4 |
| #10 | The Law Offices of Michael S. Lamonsoff, PLLC | 1.7% | 3/4 |
The full leaderboard with confidence intervals, per-assistant breakdown, and every recorded citation lives at the live personal injury leaderboard.
Cross-assistant coverage
A firm named by all four AI assistants is a structurally different signal from one named by only one. Cross-assistant coverage in the current snapshot:
- 1 firms cited by all four assistants (ChatGPT, Claude, Gemini, and Perplexity). These are consensus-tier — the AI assistants agree.
- 3 cited by three of four.
- 10 cited by two of four — half the market surfaces them; half doesn't.
- 23 cited by only one of the four — visibility with a specific assistant but invisible on the others.
The single-assistant tier is where most of the tail sits. Being "recommended by ChatGPT" is not the same as being recommended across the whole AI shortlist — buyers who consult a different assistant get an entirely different list.
What this means for a New York City attorney
If your firm appears in the leaderboard at all, you're in the top 37 firms the AI assistants know exist for personal injury. If you're not in the leaderboard, the AI assistants aren't currently surfacing you when New York City buyers ask.
Two variables move ranking. The first is what the AI can quote about you — your website's ability to answer buyer questions in short, factual, paragraph-level chunks that AI assistants can lift verbatim. The second is where you're already cited — press mentions, directory listings, and peer citations that the assistants used at training time and continue to weight.
Both are actionable. The Viclaro Market Audit ($1,500) walks through the five prioritized content gaps for a specific firm, with named source pages the AI is citing instead, and publish-ready draft copy your team can adapt.
Methodology (short version)
Versioned prompt set of buyer-style questions organized into five conversation-state buckets (Process, Specific Need, Conversational, Style / Approach, Broad Authority). Each question sent to ChatGPT, Claude, Gemini, and Perplexity with multiple samples per (question × assistant). Every response parsed into a JSON array of named firms; names normalized and matched against a pre-harvested New York City pool. Share = citation count / total ranking-layer responses. Immutable snapshots — every ranking is reproducible.
Full methodology, including bucket definitions, sampling design, and confidence-interval calculation, is at viclaro.app/leaderboards/methodology.
Related
- Live personal injury leaderboard — full rankings with confidence intervals and per-assistant breakdown
- Construct validity of AI citation audits — the underlying research paper
- Viclaro Atlas methodology — full technical documentation