Giving workforce planners
an AI-agent based warning system

Giving workforce planners
an AI-agent based warning system

Giving workforce planners
an AI-agent based warning system

154%

Company

Salesforce

Role

Product Designer

Year

2025

Overview

What if we leveraged AI to turn reactive crisis management into proactive workforce control

What if we leveraged AI to turn reactive crisis management into proactive workforce control

I led the product design for the Capacity Planning effort, which gives field service Planners peace of mind when planning months ahead for a large workforce.

the problem

Field Service Dispatchers get an alert when a job is missed. Sam the planner gets nothing, until it's too late.

Field Service Dispatchers get an alert when a job is missed. Sam the planner gets nothing, until it's too late.

Salesforce Field Service powers enterprise-scale operations; utility companies, telecom crews, HVAC fleets. Within those organizations, the Dispatcher assigns today's jobs. The Field Technician shows up and does the work. But the Planner; Sam; is responsible for making sure there are enough people to show up, weeks and months from now. When the Dispatcher's horizon is a few weeks at most, Sam is the only one looking further ahead. And until now, she's been doing it blind.

Meet Sam, the Planner

“As an Operations Planner, I need to accurately assess current capacity and workload distribution to ensure that we can meet future demand.”

“As an Operations Planner, I need to accurately assess current capacity and workload distribution to ensure that we can meet future demand.”

Competitive Audit

Planners are using tools… that aren't meant for planning.

Planners are using tools… that aren't meant for planning.

Before designing anything, we needed to understand the landscape Sam was navigating; not just what competitors offered, but how the entire enterprise capacity planning space was failing her. The audit made clear what signals were missing and where the real gap between software and reality lived.

connecting with users

We conducted research interviews with many Planners. What they told us changed everything.

We conducted research interviews with many Planners. What they told us changed everything.

The Planner isn't a cut and dry role. They can be a utility operations lead in Texas during wildfire season or a telecom planner managing mutual aid agreements after a hurricane. We mapped their workflows, their pain points, and the moments when the job became impossible. What emerged wasn't a list of feature requests; it was a picture of a role that had been set up without the right tool to ensure success.

Affinity mapping workshop that I co-led and included participants from several teams including Product, Design and Research.

what research revealed

Dozens of planners, one shared nightmare.
What Planners told us:

We used Claude and NotebookLM to synthesize hundreds of data points; customer VOCs, Solution Engineer interviews, UX research documents and meeting records into clear, actionable themes. Here is what the data kept saying:

What keeps Sam up at night

Power outages during storms

Mutual aid agreements between utilities share resources but spike demand.

Humanitarian aid during crises

Companies respond to worldwide SOS crises, such as climate emergencies.

Annual, advertised events

Marketing campaigns and holiday shopping cause service demand spikes.

How Sam is expected to fix it

Autonomous negotiation

Enable AI agents to interact as planners that negotiate between organizations.

Upskill management

Identify skill gaps and training needs; provide pathways for upskilling.

Allocation based on urgency

Measure and prioritize according to urgency of work and service needs

The cost of flying blind

Inefficient scheduling isn't a minor inconvenience for Sam. It compounds. Overtime costs mount, repeat visits erode customer trust, and reactive crisis management becomes the default mode of operation. These numbers aren't hypothetical. They're what the industry accepts as normal.

Reactive scheduling

15%

of total overtime is attributed to reactive adjustments from unplanned work and under-optimized routes

Reactive scheduling

15%

of total overtime is attributed to reactive adjustments from unplanned work and under-optimized routes

Repeat visits

28%

of scheduled appointments require a second visit, likely resulting from poor triage or under-skilling.

Repeat visits

28%

of scheduled appointments require a second visit, likely resulting from poor triage or under-skilling.

Outdated tools

25%

of companies still used spreadsheets for job scheduling, leading to increased error rates and information silos.

Outdated tools

25%

of companies still used spreadsheets for job scheduling, leading to increased error rates and information silos.

Disrupted work

30%

of total work hours are spent handling unplanned emergencies, disrupting planned work and reducing capacity.

Disrupted work

30%

of total work hours are spent handling unplanned emergencies, disrupting planned work and reducing capacity.

Before we could solve the big problem, we had to prove we could solve a smaller one.

Before we could solve the big problem, we had to prove we could solve a smaller one.

There are two ways Sam can respond to a capacity gap: prevent it from forming, or react when it arrives. Capacity Limits was our first step into Preventative gap resolution; It was also a high-impact, low-scope starting point and it was foundational to solving the larger scope of Capacity Planning.

How can Planners resolve Capacity Gaps?

proof of concept

Giving Sam her first warning system

The Capacity Limits Dashboard wasn't the endgame. It was the proof of concept. A daily and weekly view that let Sam compare consumed hours against limits, at a glance, across work types. For the first time, she had something to look at before the storm arrived, not after.

The Capacity Limits dashboard was Sam's first real signal. It proved the concept before we built the system.

What edge-cases might Sam face?

What edge-cases might Sam face?

Before finalizing anything, we mapped every boundary condition Sam might face: overlapping territories, skill mismatches, demand spikes, partial outages. All to make sure the design held up under real-world pressure.

the final design

From capacity gap to gap resolution in one flow

From capacity gap to gap resolution in one flow

The final Capacity Planning dashboard was designed around a single principle: Sam should be able to land, understand her situation, and take action without friction. The AI Gap Resolution Agent is there to take on the load; with guardrails Sam controls.

From a single feature to an enterprise-grade planning system

From a single feature to an enterprise-grade planning system

Capacity Planning was a large-scale effort spanning several teams within Salesforce Field Service. I’m incredibly proud of this team effort.

  • The product matured from a Capacity Limit focus to a Planner-first operational dashboard that can bridge the gap during emergencies and other large-scale events for our customers

  • Upon launch Capacity Planning saw immediate partnerships with 2 enterprise design partners

  • An agent-based capacity gap resolution flow was presented to the CEO of Salesforce (Marc Benioff) within a few months of launch

  • Improved stakeholder alignment through SE and planner workshops, translating field realities into product-ready requirements

“As an Operations Planner, I need to accurately assess current capacity and workload distribution to ensure that we can meet future demand.”

Meet Sam, the Planner

Dave Orian

Dave Orian

Dave Orian