Google is testing a simple but consequential change to Gemini: context-aware follow-up prompts that nudge you to keep asking questions. Instead of ending a session after a single answer, Gemini will suggest next steps—comparisons, definitions, pros and cons, or “what’s next” actions—right on the overlay and inside the app.
Early sightings suggest the feature is rolling out to a small group of users, with reports tied to the Google app version 16.34.58 and a server-side switch. It’s a classic growth lever for conversational AI: reduce friction, coach the user, and extend the thread.
Why these prompts matter
Most people still treat AI assistants like search boxes—ask once, move on. That habit underuses what large language models do best: build on context. By offering smart follow-ups, Gemini lowers the cognitive load of figuring out “what should I ask next?”
UX research backs the approach. Nielsen Norman Group’s “recognition over recall” principle shows users succeed more when systems present options rather than demanding perfect queries. Google’s own People + AI Guidebook encourages “suggested next steps” to guide novices and reduce dead ends. Follow-up prompts operationalize both ideas in a conversational setting.
How the feature shows up in Gemini
When you ask an open-ended question—say, “How does a car engine work?”—Gemini now surfaces suggestion chips such as “Compare engine types,” “Electric vs. gas,” or “Maintenance checklist.” The prompts also appear after statements like “Planning a weekend trip,” reframing your intent into next-step actions like “Draft an itinerary,” “Estimate costs,” or “Find weather and events.”
These suggestions are context-sensitive and thread-aware, meaning they adapt as the conversation evolves. In practice, that makes it easier to pivot: from explanation to comparison, from research to decision, and from ideas to execution (like generating emails, tasks, or summaries).
A nudge toward deeper, longer sessions
Gemini 1.5 offers unusually large context windows—up to around one million tokens in supported tiers—so it can remember and work across lengthy threads, documents, and media. The challenge has been getting users to actually use that depth. Follow-up suggestions are the on-ramp, encouraging multi-turn exploration instead of one-and-done lookups.
This has clear utility across tasks. In learning, prompts can steer from definitions to worked examples and practice questions. In shopping, they can shift from specs to side-by-side comparisons and total cost of ownership. In planning, they can move from ideas to schedules, checklists, and shareable summaries.
Limited rollout, server-side control
So far, the feature appears in the Gemini overlay and app for a subset of users and queries, suggesting a controlled experiment. That aligns with Google’s typical pattern: ship plumbing in an app update, then gate activation via the server. Availability will likely vary by region, language, and account settings as the company tunes quality and safety.
Expect it to expand gradually if engagement and satisfaction metrics improve. Historically, Google scales conversational features in stages, prioritizing precision and guardrails in sensitive categories like health, finance, and civic information.
How it stacks up against rivals
Competitors already lean on suggestion chips. ChatGPT often proposes follow-on prompts after responses, and Microsoft’s Copilot surfaces “Try asking” options that pivot to comparisons or actionable tasks. Perplexity uses guided questions to push deeper retrieval. Google adopting the same pattern inside Gemini—across both overlay and app—signals a broader shift: assistants are becoming proactive guides, not passive responders.
Benefits—and the guardrails required
Done well, follow-up prompts improve learnability, reduce dropout, and help users reach outcomes faster. They can also expose useful capabilities—code generation, data extraction from files, or multi-step planning—that many people don’t realize Gemini can handle.
But prompts must avoid leading users, overconfident suggestions, or biased framing. Clear labeling, safe defaults, and the option to dismiss suggestions are essential. For sensitive topics, suggestions should emphasize reputable sources, uncertainty, and choices rather than definitive prescriptions—an approach endorsed across AI safety research communities and human-centered AI guidelines.
What to watch next
If the test scales, look for tighter integration with Google services: quick adds to Calendar, Drive, and Keep; citations for web answers; and continuity across devices. The broader goal is unmistakable: retrain users to treat Gemini as a partner in multi-step work, not just a faster search box.
Follow-up prompts are a small UI change with outsized behavioral impact. They’re how Google plans to keep the conversation going—and, more importantly, make those conversations more productive.