Smart Jobs

What does e-commerce look like when the system already knows what the customer needs? Sibi's users order appliances, paint, and HVAC for thousands of rental homes. The old flow made them browse a catalog and enforce their company's specs by hand, about five minutes per order. Smart Jobs inverts it: enter an address and a need, and get three compliant recommendations, ready to order in 90 seconds.

My roles: Concept collaboration, design lead, full design execution.

Results

Start With an Address and a Need

Property managers start by entering an address. Smart Jobs pulls in what Sibi already knows about the home - square footage, bedrooms, appliance fuel type, year built, occupancy - so nobody has to look it up.

From there they pick the job and the products they need. Each choice reveals the next step, keeping the flow short and focused.

Landing

Property details loaded

Products chosen

---

Appliances: Three Recommendations, One Tap

Smart Jobs starts from what's already in the home. It knows the existing appliances are stainless steel and electric, and combines that with the company's order history and compliance specs to offer three ranked options for each appliance. Each card explains why it's there: it meets company standards, it's in stock nearby, and it's popular in the organization. Selecting one collapses the options into a confirmed line item, with installation, disconnection, and disposal already attached.

Paint: The Right Amount, Automatically

With paint, nobody wants to browse products; they want the right amount of the right paint. Smart Jobs uses the home's square footage to recommend coats for ceilings, walls, and trim, totals the gallons, and fills in the paints the company already uses. Supplies like caulk, joint compound, and tape are suggested right below, one tap from the order.

Multiple Jobs, One Order

After appliances and paint, we extended the pattern to HVAC, each category with its own recommendation logic. Users can add several jobs for the same property and check them out together as one order.

Second job added

Recommendations

Ready to check out

---

Earning Trust From Power Users

The design challenge wasn't the AI; it was earning trust from power users with years of habit. We shipped to one named user behind a feature flag, watched him buy the new way on day one, and designed every iteration against the question "does this still make sense for Marcus?"

Users told us they tried to break the recommendations. They couldn't. They didn't go back.

---

Bringing Smart Jobs Into the Shop

Next, we started bringing the Smart Jobs experience into Sibi's traditional shop. When users open HVAC or Paint, they're shown a recommended order right away, built from the property's details, instead of a catalog to browse. Even when it needs adjusting, they start from something that's nearly right.

To the left of the recommendations is an AI order assistant. Users can tell it what's different about the job, like a smaller system or a flatter sheen on the walls, and it updates the recommended products for them.