TL;DR
A new piece from Lexicon Legal Content distinguishes performed authorship (a byline on agency-written content) from verifiable authorship trail (signals AI engines can actually follow). Most law firm content claims tier-4 authorship and delivers tier-1. Meanwhile Perplexity opened its Personal Computer agent to all Mac users yesterday, widening the surface where citation visibility matters. If AI cited a legal answer in your practice area today, whose name would appear?
Game changers
Perplexity opened Personal Computer to all Mac users yesterday. Previously locked behind the $200-per-month Max tier, the macOS app now runs for any Pro or Max subscriber. The agent executes tasks across local files, apps, the web, and Perplexity's own servers, pulling Gemini for deep research, Nano Banana for image processing, and ChatGPT for long-context recall.
Why this matters for law firms: more queries are about to flow through Perplexity's citation-heavy architecture. Perplexity already averages nearly 22 citations per response, the highest of any major AI platform. As more clients delegate research to agents instead of typing into search boxes, the surface area where your firm can earn a citation just expanded again.
Search Engine Land also published "8 GEO metrics to track in 2026" this week. The headline number for skeptical partners: when an AI summary appears, users click traditional search results just 8 percent of the time.
Sources: Dataconomy, Search Engine Land
AI search news
A 2026 study of 34,234 AI responses found a 46-times difference in brand citation rates between platforms. ChatGPT cited brands 0.59 percent of the time. Perplexity sat at 13.05 percent. Grok came in highest at 27 percent.
Translation for your firm: if you are willing to test Grok visibility, you have a 46-times higher baseline citation rate than ChatGPT to start from. Most firms ignore Grok entirely. There is a first-mover opportunity hiding in that data point.
Recency is also doing more work than most firms realize. Perplexity cited content published within the last 30 days at an 82 percent rate in one 2026 analysis. A six-month-old blog post consistently loses to a fresh piece on the same topic. Visible year signals (including "2026" in titles and headings) improve citation rates by roughly 30 percent.
Source: Leapd
Trends
Vertical-specific GEO playbooks are showing up faster than general best-practice guides now. Yesterday Demand & Convert dropped a 2026 Guide to AI Search for Plumbers with vertical-specific tactics: geo-tagged EXIF data on photos, hyper-local pricing data to feed knowledge graphs, monitoring of local forums like Nextdoor.
Why this matters for legal: GEO has officially fragmented from general best practices into vertical specialization. Healthcare, real estate, financial services, and legal each have different citation surfaces and different content signals that work. Generalist GEO shops will lose to vertical specialists who actually understand the practice area.
Source: Best Lawyers
Casey’s Take
David Arato at Lexicon Legal Content published a piece this week that was perfect.
His core insight: AI search doesn't rank pages anymore. It cites them. The firm that gets named is the one AI decided to trust. And that trust is built on signals AI can actually follow, not on a name attached to the bottom of a blog post.
He breaks attorney content into a four-tier spectrum. Tier one is AI-drafted with an attorney's name slapped on. Tier four is attorney-drafted core analysis with a writer structuring and polishing. Most firms paying for "attorney-written content" think they are buying tier four. The trail behind the byline says tier one or two.
From my experience, in any honest discovery call, you can tell within five minutes whether the prospect actually understands their own business or whether they have memorized the slide deck. AI is doing the same thing with content. It is reading the trail behind the words. The bar admissions, the linked bios, the cases cited, the speaking engagements, the publications quoting the attorney by name. Those are signals a model can follow. A byline alone is not.
The good news is this is fixable, and the firms that fix it now will be cited for years. AI authority compounds. The trail you build this quarter is the trail that earns citations 18 months from now.
If your firm is paying an agency for content, ask one question this week. If a client asked ChatGPT who wrote this and why they should trust it, what would the answer be?
Wild cards
The 46-times citation gap between Grok and ChatGPT is the most ignored stat in GEO right now. Almost no law firm is optimizing for Grok, and the citation rate sits at 27 percent. That is structurally cheaper visibility than any other AI platform. Worth running a single test query through Grok this week and seeing what comes back.
Perplexity Personal Computer also signals the longer-term shift worth tracking. As more queries move from typed searches to delegated agent tasks, the agent's underlying retrieval mechanism becomes the new search algorithm. The same GEO playbook applies. The surface is just changing.
Tip you can use today
Pick one practice area page on your firm's site. The page you would most want a prospective client to read. Then run this audit:
Is the attorney's name in the byline (not "our team")? Is their bar number visible? Is there a link to their bio with credentials and jurisdictions? Is there a last-updated date? Is the body text written from actual case experience, or could any agency have written it?
Then open Perplexity, ask the question the page answers, and see whether the firm is cited and the attorney is named. The gap between the page you have and the citation you want is the work to be done.
That is one page. One afternoon. The trail you build there compounds across every related question your future clients will ask AI before they pick up the phone.
If this helped, stay subscribed. New brief drops Monday through Friday, with a recap on Saturday.
If you want me looking at your firm's AI visibility directly, grab a free AI visibility audit at the button below.