Sharebird
Sign Up / Sign In
Paresh Vakhariya

AMA: Atlassian Director of Product Management (Confluence), Paresh Vakhariya on Leveraging Agentic AI Trends for Success

December 9, 2025 @ 10:00AM PT

View AMA Answers

  1. How do folks effectively communicate new releases (big or small) so customers are aware that we have potentially solved a problem or frustration?

    Paresh Vakhariya
    Paresh Vakhariya

    Atlassian Director of Product Management (Confluence) | Formerly PayPal, eBay, Intel, Verizon • 6mo

    • Use various communication methods: in-product nudges, tooltips, release notes, email digests, demo videos and short “before/after” demos.
    • Tie every announcement to a specific customer frustration solved, not a feature shipped.
    • Add quick wins like “Try it now” prompts to convert awareness into usage.

    1,673 Views
    1 request

  2. How are you evaluating which parts of your product roadmap could benefit most from agentic AI capabilities, and what criteria are you using to prioritize these opportunities?

    Paresh Vakhariya
    Paresh Vakhariya

    Atlassian Director of Product Management (Confluence) | Formerly PayPal, eBay, Intel, Verizon • 6mo

    • Start with workflows that are repetitive, multi step, high cognitive load, or frequently considered manual by users.
    • Score each opportunity by: customer impact, data/context availability, risk of incorrect autonomy, engineering complexity, and operational cost of AI (LLM cost).
    • Validate concepts with lightweight Figma prototypes before committing fully.

    883 Views
    1 request

  3. What does your competitive landscape look like in terms of agentic AI adoption, and how is this influencing your product strategy for the next 12-18 months?

    Paresh Vakhariya
    Paresh Vakhariya

    Atlassian Director of Product Management (Confluence) | Formerly PayPal, eBay, Intel, Verizon • 6mo

    • Track competitors’ moves in GPT, copilots, agent frameworks, workflow automation, and enterprise AI platforms.
    • Identify gaps in reliability, personalization, trust, and integration depth as differentiation levers.
    • Plan for faster iteration cycles to avoid being boxed into “me too” releases.

    • Test with customers to make sure you are building what matters, not just for competitive parity but also for differentiation.

    693 Views
    2 requests

  4. What metrics are you planning to track to measure whether agentic AI features are actually delivering value to users versus just being technically impressive?

    Paresh Vakhariya
    Paresh Vakhariya

    Atlassian Director of Product Management (Confluence) | Formerly PayPal, eBay, Intel, Verizon • 6mo

    • Time saved on key workflows• Task completion rate• Cost reduction from fewer manual steps, reduced support tickets, and lower operational overhead.• Error reduction • Adoption curves• Frequency of AI agent overrides or user rollbacks, which helps flag trust issues, hallucinations, or unclear decisions.• Customer-reported trust, measured through surveys, NPS deltas, in-product prompts, or qualitative feedback.• Precision and recall on agent actions when the expected result is known, especially ...Read More

    649 Views
    1 request

  5. How are you preparing your product infrastructure to support agentic workflows, particularly around API design, tool integration, and handling multi-step autonomous tasks?

    Paresh Vakhariya
    Paresh Vakhariya

    Atlassian Director of Product Management (Confluence) | Formerly PayPal, eBay, Intel, Verizon • 6mo

    • Standardize your internal APIs so agents can reliably trigger actions, retrieve context, and store outputs.• Introduce task orchestration primitives for multi step actions: retries, fallbacks, and conflict resolution.• Build safe execution sandboxes • Add a central “agent runtime layer” that handles memory, context stitching, error handling, and step-by-step observability.• Set up rate limiting, quota management, and service isolation to ensure agent workloads don’t degrade core product perfor ...Read More

    675 Views
    1 request

  6. How are you thinking about the balance between AI autonomy and user control in your product design, especially as agents become more capable of taking independent actions?

    Paresh Vakhariya
    Paresh Vakhariya

    Atlassian Director of Product Management (Confluence) | Formerly PayPal, eBay, Intel, Verizon • 6mo

    • Default to a supervised mode where the agent proposes actions and the user approves.
    • Add adjustable autonomy levels so advanced users can opt into more automation as they go.
    • Provide transparency: show reasoning, data sources, and planned steps to build trust and transparency with AI.

    for e.g. showing chain of thought as AI processes it's responses.

    650 Views
    1 request