Thumbtack

Business sites dominate Gemini’s local AI citations

A Steady Demand study of 1,487 local AI search queries across 50 U.S. metro areas found that nearly 60% of Gemini’s citations pointed to business websites, outperforming directories, review platforms, and forums. The analysis of 14,472 citations also revealed large variability in AI recommendations: repeating the same Gemini query produced overlapping cited sources only ~40% of the time, and Gemini returned the same top business about 7% of the time. In contrast, Google’s local pack was far more stable, returning the same top listing ~90% of the time. The researchers also ran the queries through ChatGPT and found low overlap with Gemini — platforms shared many of the same domains only 8% of the time and the same top business 4.2% of the time. The report warns that single AI searches are snapshots and recommends monitoring multiple AI engines for AI visibility.

Pixel 11 Emphasizes Gemini AI, Few Hardware Changes

Google’s Pixel 11 family makes modest hardware changes but expands Gemini-powered on-device AI and camera capabilities. All models run the new Tensor G6 chip, which Google says brings power and performance gains — roughly 20% better power efficiency, ~25% faster web browsing, ~15% faster app launches and 50% more TPU compute — and when paired with Gemini Nano can run on-device AI up to 3.5× faster using up to 3.5× less energy. Gemini adds agentic features (ordering groceries, booking rides, calling businesses) with user controls and Rambler clean transcription, plus third‑party integrations such as Otter.ai and Ticketmaster. Camera upgrades include a 48MP main sensor (56% more light sensitivity), Magic Capture, Camera Looks, upgraded telephoto (base up to 30×, Pro up to 120×), Instant Night Sight and 8K video. Devices start at 256GB, from $899, and are available to preorder.

Speed-to-lead: Win the Job in Under 2 Seconds

This technical blog post argues that rapid lead response (a 'first meaningful reply' under two seconds, 24/7, without a human initial touch) is a systems engineering problem for local service businesses. Citing research from MIT and Harvard Business Review, the author explains why naive autoresponders fail and outlines engineering challenges — input normalization across multiple lead channels, intent understanding, follow-up logic, idempotency, and voice callback handling. The author (who runs the product Auto-Respond) describes how their service answers leads across Yelp, Thumbtack, Facebook/Instagram, and Google Local Services Ads, qualifies and books jobs, follows up on scheduled days, callbacks missed calls, and syncs to CRM systems.