Led a cross-functional team to reimagine Salesforce Mobile with an AI foundation and present a vision to the Product EVP. The result showcased agent-powered workflows, transforming dozens of taps into a single, voice-based interaction to increase productivity.

Turning mobile friction into AI opportunity

This project was a collaboration between myself, the VP of Product Management, and a principal architect to explore how AI could fundamentally transform the Salesforce Mobile experience. Together, we defined three pillars for an AI-first mobile vision: agent-powered task execution to help users complete work on the go, proactive, context-aware assistance to predict meaningful actions, and intelligent, multimedia search to answer complex questions through natural language. Our shared goal was to create a bold yet realistic vision grounded in a human-in-the-loop methodology that reduced input, maximized output, and minimized reliance on Salesforce tribal knowledge.

Through research, we learned that users found getting work done on mobile tedious and search frustratingly limited. Many delayed updates until they returned to their laptops, risking lost data and inefficiency. To make the opportunity tangible, we focused on a B2E user creating follow-ups after a client meeting, a process that took 48 taps and 7 page loads. By reimagining this workflow with AI, we identified opportunities to automate tasks, simplify updates, and help users stay productive anywhere.

A more natural way to get work done

After meeting with a client, users typically create and assign tasks, update opportunities, and log calls with detailed notes. This process is laborious, filled with taps, modals, and delays as data is processed and saved.

The mockup below demonstrates a new entry point for interacting with AI. The goal is to remove the cognitive load of deciding what to type as a starting point. I believe the future of working alongside AI will rely on voice, photos, screenshots, or AI-determined context, rather than manual text entry.

Users can naturally share high-level goals with AI, along with screenshots or photos as supporting documentation. This approach moves beyond database-centric interactions and enables a conversational, human-centered workflow.

The ability to attach handwritten notes, whiteboard photos, or laptop screenshots allows users to update Salesforce more quickly and thoroughly. Updates become richer, more accurate, and less of a chore.

Once data is submitted, AI generates suggested record updates based on the transcription and attachments. A human-in-the-loop flow allows the user to verify, edit, and approve these changes before submission.

The final step offers transparency and trust. Before saving, users can review a confirmation screen that summarizes all actions taken by AI. This gives users confidence in what the system did and builds long-term trust in automation.

Designing a proactive AI assistant

Using AI to summarize inputs and suggest record updates is only the beginning. The next phase is enabling AI to anticipate user intent and recommend actions automatically.

For example, if a user consistently creates tasks after meetings, AI can recognize this pattern and proactively send a push notification with a quick action button to start creating tasks. This reduces friction and encourages users to keep Salesforce data current.

To make the interaction even faster, the initial "What's your goal?" input can be pre-populated based on context AI already understands:

• The user recently had an event with Electric Motors.
• The user often creates follow-up tasks after meetings.
• The next logical action is to record notes or create new tasks.

Now, instead of navigating multiple screens, the user can simply record a short voice memo describing the meeting outcome. After recording, they submit the context and audio directly to AI. Importantly, this workflow keeps users out of a chat-centric interface. They are getting work done with AI, not by chatting with it. The result is a seamless experience where AI supports the user's goals quietly and efficiently.

As Salesforce administrators begin deploying custom agents to assist end users, there will also be a need for a centralized space to track agent-driven activities. The existing Notifications panel provides a natural home for this new class of agent activity, allowing admins and developers to integrate updates directly into an established platform.

This vision builds toward an AI assistant that doesn't just respond to users but works alongside them, anticipating needs and guiding action at the right moment.

Designing intelligent search for mobile

In nearly every research session, users expressed frustration with finding information in Salesforce. Searching is slow, fragmented, and requires navigating multiple pages and related records to locate simple answers. Complex questions are nearly impossible to ask. This is where AI can transform the experience.

In the mockup below, the user asks, "Who did I meet with at Acme last week?" Traditionally, this would require searching for the Acme account, opening related opportunities, and scanning each one for associated contacts. It is tedious and time-consuming.

However, the user's true goal is not to find a single record. They want to call the contact, review the related opportunity, and follow up on next steps. AI can understand the intent behind the question and gather all relevant records — including contacts, opportunities, tasks, and activities — into one contextual view. Because the user mentioned "meet with," the results can even include actions taken after the meeting, such as an opportunity stage update or newly created tasks.

Beyond Salesforce, we explored how this vision could extend across the broader product ecosystem. One concept, now live in Slack, allows users to ask complex, cross-product questions such as, "What were the most popular products sold to Acme Industries last quarter?" Answering this requires combining data from Salesforce, Slack conversations, and documents stored in Google Drive. The interface can also include dynamic Tableau charts to make insights immediately actionable.

Through this exploration, search evolves from a passive query tool into an intelligent assistant — one that understands context, connects systems, and delivers knowledge rather than records.

Presenting the vision to executive leadership

This vision project culminated in a presentation to the Product EVP and Chief Design Officer, showcasing how Salesforce Mobile could evolve into an AI-first platform where agents anticipate user needs, complete workflows, and connect data across the enterprise. The presentation inspired collaboration between Slack, AI Frontiers, and Lightning Platform, leading to a new search experience that uses AI to generate answers spanning multiple enterprise systems.

This project reinforced my belief that design leadership goes beyond solving immediate problems. It is about anticipating what comes next, aligning cross-functional teams around a shared vision, and transforming that vision into a roadmap others can build upon. Leading this AI-first mobile initiative demonstrated how design, engineering, and product collaboration can redefine productivity for millions of users.