Get your firm
recommended
when clients ask
AI who to call.
Specter gets your firm into those answers, on your existing site, without buying ads.
Six days · No engagement required · You keep the assets
“
Specter Discover is an AI search visibility service for plaintiff law firms. It tracks how often a firm is named in answers from ChatGPT, Gemini, Perplexity and other AI assistants, closes the evidence gaps that keep it out of those answers, and reports share of answer weekly on the prompts that produce cases. It runs on the firm’s existing website, with no ad spend, and it exists because an AI answer names two or three firms rather than returning ten links to compare.
The answer is the shortlist.
Ask a model who to call and it does not hand back ten links to compare. It names two or three firms, and those are the only firms that get called. Nobody asks the question until they already know they need a lawyer, which makes it the highest-intent traffic on the internet, and the fastest-growing.
- 1Hollis Barrera LLP
- 2Trent Alvarado Law
- 3Kaplan Reed
Year on year, across the legal accounts we instrument. Answers roughly quadrupled what they sent, while paid search costs rose.
Visits from AI answers, monthly · trailing twelveAn AI-sourced consult signs at more than twice the rate of a paid-search click, because the model already did the qualifying.
Consults that became signed retainersAgainst the same firm's paid search, in the same practice area, over the same ninety days.
Media and fees per signed caseTurn AI search into a client acquisition channel.
Discover finds what decides which firms AI recommends, closes the gaps in your presence, and keeps improving your visibility across ChatGPT, Gemini, Perplexity and AI search, then counts every stage, down to the retainer read back from your case system.
Questions asked
Real client questions, per jurisdiction.
Answers won
A model names you, with a fact it can verify.
Visits
Traffic traced back to the answer that sent it.
Consults booked
Logged at intake, with the platform named.
Retainers signed
Read back from your case system, with the fee.
Built to drop into the firm you already run.
No new platform, no parallel website, and a named human who owns the outcome.
Works with your website
Schema, attorney entity pages and answer-first content deployed straight into your existing CMS. We never touch your theme, and nothing publishes without an attorney's approval.
Secure by default
Encryption in transit and at rest, least-privilege access and a DPA your general counsel can actually read. We see intake outcomes, never privileged content or matter files.
A named person on the account
One named person who owns your prompt universe, briefs your attorneys and reports to your partners. Not a ticket queue, and not a dashboard you are left to operate.
Find out where you stand this week.
Six days, no engagement required. You keep everything we produce.
What you are choosing between.
Three things get sold under the AI-search banner. They are not the same product.
| Your SEO agency | Horizontal GEO tool | Specter DiscoverBuilt for law firms |
|---|---|---|
| × Optimises rankings, not answers | × Category-agnostic, with no legal sources | → Share of answer on the prompts that produce cases |
| × No work on the evidence a model can verify | × Ignores bar registries and certifications | → The seven directories legal answers actually cite |
| × Legal directories are an afterthought | × Bar rules not modelled at all | → Bar registries, certifications and verdicts as evidence |
| × Bar advertising rules vary by shop | × You operate it, in a dashboard | → Bar rules enforced per jurisdiction, at drafting |
| × Reports traffic and form fills | × Gives you a score, not a case | → Prompt to consult to retainer to fee value |
| “Traffic is up.” | “Our score improved.” | “This signed five cases at $2,190 each.” |
Priced against a signed case, not a seat.
One flat monthly fee, published up front. The size of your prompt set scales with the plan. Ninety days, then month to month on fourteen days' notice.
For firms that want to see the gap before they close it.
Measurement and playbook. You publish the work.
Start with Signal- ✓Core prompt set, swept weekly across five engines
- ✓Share of answer against your real competitor set
- ✓Citation & directory audit, monthly
- ✓Living answer playbook and weekly digest
- –No dedicated support. Email replies in two business days
- –No pages drafted or published for you
Everything on this page, run for you by a named person.
The full engine. Your attorneys approve, we do the rest.
Get my free audit- ✓Everything in Signal, on an expanded prompt set
- ✓Practice-area & case-type pages, twice a week
- ✓Attorney-authored answers drafted weekly
- ✓Citation & directory fixes filed for you
- ✓Published into your existing site, with no rebuild
- ✓Approval queue: nothing ships without a partner's yes
- ✓Prompt to consult to retainer attribution, wired into intake
- ✓A named person on your account, in your channel
- ✓Monthly partner session
Multi-office, multi-jurisdiction, mass tort and MDL.
Priced on jurisdictions and prompt volume.
Talk to us- ✓Everything in Discover
- ✓Separate office & jurisdiction tracking
- ✓Dedicated account lead and backup, on an SLA
- ✓Case-management read-back & fee reporting
- ✓Custom DPA, SSO and audit logs
- ✓Quarterly on-site partner briefing
What managing partners ask first.
What is generative engine optimization (GEO) for law firms?
Generative engine optimization is the practice of getting a firm named in the answers AI assistants give, rather than in a list of blue links. When somebody asks ChatGPT, Gemini or Perplexity which firm to call, the model returns a short answer naming two or three firms, assembled from sources it can verify. GEO is the work of making a firm one of those verifiable sources.
In practice that means the firm's evidence — attorney credentials, practice areas, jurisdictions, case types, results language, directory and citation footprint — is complete, consistent and machine-readable everywhere those models look.
Which AI platforms does Specter monitor?
ChatGPT, Gemini, Perplexity and the other assistants your prospective clients actually use, swept weekly against your own prompt set. The set is built from your practice areas, jurisdictions and target case types rather than from a generic keyword list.
Does a firm need a new website to use Discover?
No. Discover runs on the site the firm already has. Schema, attorney entity pages and answer-first content deploy into the existing CMS, the theme is not touched, and nothing publishes without an attorney's approval.
Is this just SEO with a new name?
No, and the difference is measurable. SEO optimises a page's position in a list of links. AI answers do not produce a list of links. They produce a paragraph naming two or three firms, assembled from sources the model can verify. Ranking first for “Austin injury lawyer” and being absent from the ChatGPT answer to the same question are entirely compatible states, and we see them together constantly.
What carries over: technical hygiene, content quality, authority. What does not: keyword targeting, link volume, position tracking. Discover is built on the parts that transfer.
Can you guarantee we will be recommended?
No, and anyone who does is either lying or planning something that will put you in front of your bar's advertising committee. Models are non-deterministic and their retrieval changes without notice.
What we commit to is the input side: your evidence will be complete, consistent and citable across every source legal AI answers are demonstrably built from, and you will see the resulting share of answer measured weekly. If it does not move in ninety days, you leave and keep the assets.
How many prompts do you actually track?
As many as your market and your plan call for. We build the set from your practice areas, your jurisdictions and the case types you want more of, then weight it by how often each question is really asked. A single-office consumer practice might sit in the low hundreds; a multi-office mass tort programme runs into the thousands, per jurisdiction.
The number is not the point. Coverage of the questions that produce signed cases is the point, and we would rather sweep a tight set every week than a vanity list once a quarter.
How do you attribute a case to something a client typed into ChatGPT?
Three signals, triangulated. First, referrer and assisted-visit detection at the website layer, since AI platforms increasingly pass identifiable referrers, and we log the landing page and the answer that was live at that moment. Second, a rewritten intake question your staff actually ask, wired into your forms and call scripts, with AI platforms as explicit named options. Third, a read-back from your case management system on status and fee outcome.
It is not perfect and we do not present it as perfect. It is materially better than the “direct / none” bucket where this channel currently disappears, and it is honest about the confidence level on every attributed matter.
Our practice is defence-side, corporate or immigration, not plaintiff PI. Does this apply?
Yes, though the mechanics shift. Consumer-facing practices such as injury, family, criminal, immigration, employment and bankruptcy move fastest, because those are the questions people bring to a chat model at 11pm. For corporate and defence work the prompts come from in-house counsel and procurement (“who handles ERISA class defence in the Fifth Circuit”), volumes are lower, and citation weight sits with Chambers and Legal 500 rather than Avvo. Same engine, different prompt universe, different directory emphasis.
Why should a firm trust Specter with this?
Specter builds litigation infrastructure for plaintiff firms: docket intelligence, case scoring and record review across forty-plus jurisdictions, built by litigators. Discover is that team pointed at the top of the funnel. We understand what a board certification is worth, why past-results language is dangerous in some states, and what a managing partner needs to see before reallocating budget. That context is the product.
Find out what the models are saying about your firm this week.
Six days, no engagement required. We run your practice areas through all five engines and show you exactly where you stand, and who is being named instead.