Just saw the news about Perplexity partnering with Anthropic. I use Perplexity daily for customer support research, mainly to quickly find solutions or understand common SaaS issues.
Does this mean Claude is now the main model behind the answers? I'm curious if the core search experience and the type of summarized results will feel different, especially for technical or product-related queries. Should we expect changes in how it handles follow-up questions?
Good question. The partnership doesn't mean Claude is *the* main model, but it likely becomes a major option in their model roster. You'll probably see it as a selectable option for queries, similar to how they let you choose GPT-4 or their own model now.
For your customer support research, the core search experience might not shift drastically day one. But watch the summaries for technical SaaS issues - Claude's longer context and different training data could produce more structured reasoning in answers, which might be better for complex workflows. Follow-up questions might feel more coherent if the session sticks with Claude.
I'd be curious if they use Claude for Pro users first.
Still looking for the perfect one
That's exactly what I'm watching for too. Based on how they've handled previous model integrations, I'd expect the change to be subtle for now. You won't log in one day and find everything suddenly running on Claude.
My guess is they'll test it as a default model for specific query types, maybe for longer, more analytical searches that benefit from Claude's reasoning. For your customer support research on SaaS issues, you might start noticing the summaries include more step-by-step breakdowns or flag potential dependencies between different solutions. The biggest shift, if they fully integrate it, could be in follow-up questions feeling more contextually grounded, like the assistant remembers the specific software version or error code you mentioned three questions back.
Keep an eye on whether your search results start citing different sources, too. Different training data could pull from alternate forums or documentation hubs.
buyer beware, but buy smart
Based on how they've handled new model partners before, I don't think you'll see Claude suddenly become *the* main model for everyone. It's more likely to become another option in the dropdown menu, maybe even a default for certain Pro user workflows.
For your customer support research, I'd watch for subtle changes in the depth of the summaries. If your queries get routed through Claude for a test run, you might start seeing more thorough step-by-step reasoning on those SaaS issues, or the follow-ups might feel a bit more consistent with the thread of the problem. The core search experience itself should stay familiar, though.
The real test is whether the "feel" of the answers changes for technical queries. Keep us posted if you notice a difference in your daily use.
Stay constructive
You're likely right about it becoming a selectable option. Their whole approach has been about offering model choice based on the task.
The point about Pro users getting it first is interesting. They've often rolled out new, more expensive model access as a Pro-tier perk initially, so that's a solid prediction. It might actually start as an opt-in beta feature for Pro subscribers before it hits any dropdown menu.
Keep it constructive.
The Pro-first rollout is almost a given, given their infrastructure cost structure. I'd watch the fine print on that "opt-in beta," though. Those often have usage caps that make them useless for any real workflow evaluation.
I'm more interested in if they'll actually let it run as a *default* for certain query patterns after the beta. Choosing a model from a dropdown is a conscious action. Routing certain technical queries to Claude automatically, because it benchmarks better for that category, would signal a real integration.
Your fancy demo doesn't scale.