AI Browsers Grow Up: Dia Introduces Skill Gallery, Perplexity Adds Task Features to Comet
As AI becomes more and more incorporated into the everyday digital experience, AI-supported web browsers are undoubtedly one of the key battlegrounds. While the utopian dream of browsers doing complex multi-step tasks for us is still a long way off, two of the field’s major players — Dia and Perplexity — are making headway toward greater utility and usefulness. This week, both companies announced new capabilities aimed at making it easier to interact with AI on the web.
- Dia introduced its “Skill Gallery,” a feature enabling users to save and reuse prompt-based tasks.
- Perplexity enhanced its Comet browsing agent to include task chains for streamlined workflows.
These developments represent an incremental but crucial evolution in how AI browsers serve real-world productivity — even if the dream of a wholly autonomous web agent remains distant.
A Shift Toward Workflow Efficiency
The value proposition of AI browsers is clear: reduce the gap between user intent and online action.
Whether summarizing articles, providing data, drafting emails, or booking travel, AI tools are positioned as always-on assistants, ready to do more than simply return search results. But until now, these browsers have been mostly reactive — addressing individual prompts rather than executing continuous, multistep actions.
Both Dia and Perplexity seem to recognize that real-world users demand efficiency, consistency, and low-friction automation, not just novelty.
Dia’s Skill Gallery: Constructing an Action Library
Dia’s new Skill Gallery tackles the problem of repetitive AI use. Users often find themselves retyping similar queries:
- “Summarize this article for me like I’m 5 years old.”
- “Find recent news on clean energy startups.”
- “Generate a social media caption based on this page.”
With the Skill Gallery, users can:
- Save prompts (called “skills”) for easy reuse
- Organize skills in a personal library
- Customize tasks (e.g., with multiple URLs or different inputs)
- Pin common tasks for quick access
- Share skills with others, opening up the possibility of a marketplace for workflows
“Skills are like saved instincts,” the company noted, “letting you bring your best ideas and workflows with you, wherever you browse.”
Target Users
Early data shows Skill Gallery resonates strongly with:
- Knowledge workers
- Marketers
- Students
- Researchers
These are users frequently engaged in repetitive cognitive tasks on the web.
Perplexity’s Comet: Tasks for Persistent Agents
Perplexity, known for combining conversational AI with up-to-date sources, is pushing its Comet browser closer to agent-like behavior.
Previously, Comet responded to user queries in isolation. Now, it allows chaining tasks into cohesive, repeatable workflows.
Example Workflow:
- Collect recent articles about climate policy
- Summarize each article
- Compare perspectives
- Write an opinion paragraph based on the findings
These steps can be saved as a single, repeatable task, getting closer to a virtual assistant than a search engine.
Though not fully automated, Comet is now highly automate-able, with the potential to scale into more autonomous behavior.
Key Feature: Long-Term Memory
Comet integrates with user history and learns from past interactions, enabling it to:
- Improve responses over time
- Adapt to user preferences
- Increase relevance and trust
This persistent memory is crucial in building human trust in AI agents.
Challenges: Context, Control, and Trust
Despite these advances, major challenges remain:
1. Context Awareness
AI browsers still struggle to:
- Interpret ambiguous commands
- Prioritize competing instructions
- Adapt or form opinions on the fly
2. User Control and Transparency
Automation can raise concerns:
- What data is being accessed?
- What assumptions is the AI making?
- Can users intervene or audit the AI’s actions?
Balance is key — users must see and shape what the AI is doing without being overwhelmed by complexity.
The Future: From Suggestions to Autonomy
Today’s innovations — Skill Galleries and Task Chains — are more than feature creep. They’re foundational steps toward agentic reasoning:
- Set a goal
- Chart a path
- Execute autonomously
Envisioned Capabilities:
Imagine asking your browser to:
- Find your next flight
- Monitor fluctuating prices
- Check your schedule
- Book the shuttle when the timing is right
Such use cases require:
- Autonomy
- Memory
- Dynamic adaptation
- A user-friendly interface
We’re not there yet — but we’re inching closer.
A User-Centered Race
The competition is heating up among AI browsers — from Arc and Brave to Microsoft’s AI offerings.
But the key differentiator won’t be raw intelligence. It will be:
- Usability
- Workflow integration
- Consistency with minimal friction
Both Dia and Perplexity understand this. Their recent updates reflect a shared commitment to improving user experience through repeatable, modular AI tools.
Conclusion: Not Quite Yet, But Closer
The vision of a browser that understands your habits, anticipates your needs, and executes digital tasks autonomously remains aspirational.
However, features like Dia’s Skill Gallery and Perplexity’s Comet Task System are clear indicators that:
- We’re moving from reactive AI to semi-autonomous companions
- Trust, transparency, and usability will define the winners
- The next generation of AI browsers will be built around repeatability and modularity
These are early steps — but they point unmistakably toward a future where AI browsers do more than assist — they understand.
