Work / Designation

Myi: real-time remote diagnostics for the factory floor

Myi is a real-time remote diagnostics platform for industrial manufacturing. It let operators on the factory floor bring remote OEM engineers into a live, multi-party video session, annotate the failing machine together, and resolve faults that once required an on-site visit costing up to $6,000 a day. Designation designed and engineered it across web, iOS, and Android in a one-year, $250,000 engagement, delivered on time and under budget.

Role
Founder and Principal Product Designer, Designation
Length
One year, on time and under budget
Platforms
Web, iOS, and Android
Outcome
Repairs: days → under 45 minutes

The ask

Take Myi from concept to a production-ready, cross-platform product within twelve months: one that would let frontline workers at industrial food manufacturers such as Bimbo Bakeries and Sara Lee diagnose and resolve equipment failures without waiting for an engineer to travel.

My role

I led discovery research, product strategy, interaction design, and the design system; directed our engineering team; and owned the client relationship end to end. The client was a three-person founding team with deep roots in industrial automation and supply chain. They brought us the concept and needed a partner to turn it into a shippable product.

The challenge

  • Scarce, expensive expertise. Only a small pool of OEM-certified engineers could service this equipment, and each site visit could cost up to $6,000 a day, before counting lost production.
  • Avoidable downtime. Many failures were minor and could have been resolved remotely if described precisely. With no record of past faults and fixes, the same problems recurred, and each one restarted the clock.
  • Tribal knowledge. Troubleshooting expertise lived with a few experienced workers. When they were off shift, no one else knew what to do.
  • A hostile operating environment. Site visits across food manufacturing, warehousing, trucking and loading, and inventory and storage surfaced signal dead zones, mandatory safety gloves, fragile or hard-to-reach devices, and a workforce wary of new technology.

What is Myi?

The first thing I drew, in December 2017, was a one-page answer to the founders' own question: what is Myi? It began as a bundle of video and text, made of snaps, calls, messages and email, filed by someone on the floor and received by a team that could act on it. If the person who could fix the problem wasn't on Myi yet, the request still reached them by email.

Concept map titled What is Myi: video and text branching into snap, call, message and email; tiers of dedicated storage; and a process where a filer and their team send a Myi to a provider, who may or may not have the app
The first concept map: what Myi bundles together, where it's stored, and how it travels from the floor to whoever can fix it.

The packet

In discovery, technicians consistently described a repair as a bundle: photos, marked-up drawings, technical documentation, sometimes a DVD. I made that bundle the foundational unit of the product and its atomic design system. Each issue became a single packet of video, stills, annotations and documents that could be assembled, routed and archived as one object, the ball that people pass back and forth until the machine runs again.

Before any visual design, I wrote down the rules in plain language for the founders: who can do what, the only three states a packet can be in, and where it's stored. Keeping the states to three meant a plant's history stayed honest and searchable.

The rules, as written for the client · March 2018

  1. Super adminsThe Myi partners. Create companies and their admins.
  2. Company adminsCreate teams, team leaders, users and packets.
  3. Team leadersManage who's in their team, and create packets.
  4. UsersEveryone starts here, and can always create a packet.
Open
Still being worked on. Anyone on the packet can close it.
Closed and resolved
The fix worked. Only a company admin can reopen a resolved packet.
Closed and unresolved
It didn't work, and the record says so. No other statuses.

Every company got its own private storage, which not even the Myi partners could see. A packet shared between two companies lived in both, and only its creator could mark it resolved.

The Packet Machine

The founders' real business was the relationships around each machine: a plant, the vendors who sold its equipment, and the engineers behind them. I mapped how packets would move through that network, from a user building a packet to the right vendor for that machine, then to resolved and archived, or kept open.

Service map called The Packet Machine: a company connected to vendors and their machines, a user building a packet that is sent to a vendor, resolved packets archived and unresolved ones kept open, plus notes on the expacket, monetization and an emergency module marked not MVP
The Packet Machine: companies, vendors and machines, with the expacket loop and the ideas we held back.

The map also found the product's growth loop. A vendor who wasn't on Myi could still receive an expacket: enough to see the problem, with limits on downloads and comments, and a free trial to respond properly. Every expacket was an invitation, and Myi's partners heard whenever a vendor joined on a customer's recommendation. Just as important was what the map kept out of the first release: RFID tags to identify machines, and an emergency module.

Reaching vendors outside the walls

Audience options with expacket expanded, showing 200 expackets left and that receivers join with a 15-day free trial
Expackets are limited, and each one comes with a free trial.
Form to enter an expacket receiver's first name, last name, company and email
All a worker needs is the vendor's name and email.
Confirmation that the packet was sent as an expacket to a named vendor contact
Sent: the vendor sees the problem before they sign up.

Designed for the floor

Building a packet takes four steps, each one a single large target that works with safety gloves on: add media, capture, check the staging area, and choose who receives it. Visual annotation replaced voice where the floor was too loud, and the capture flow supported specialist imaging hardware, including industrial endoscopes and mounted hands-free cameras.

New packet screen with options to capture an image or video, add from the gallery, add a document or add a message
1 · Add media
Camera view of a production machine with a large record button
2 · Capture the machine
Staging area listing an attached video, an image and a second video, each uploading, with a packet name and description
3 · Stage everything as one packet
Pick your audience screen with internal, vendor and expacket options
4 · Choose your team, a vendor, or an outsider

Group video without a ceiling

We engineered multi-party video calls with no cap on participants, so an operator, a plant manager and several remote specialists could examine the same machine at once. We shipped this before Apple brought Group FaceTime to the iPhone. When a call ended, its recording was archived into the packet, so the conversation became part of the record.

Choose call recipients screen with three of four people selected
Pick who joins
Four-way call grid with each participant being called
Everyone on the same machine
Call ended screen confirming archiving is complete and the call will be added to the packet
The call is archived into the packet
Packet view with its creator and members, a photo of the machine, a description and a respond button
The packet carries the whole history

The turning point

The first field deployment, at a nuclear power plant in Chemnitz, Germany. Lead-lined rooms blocked every signal, and a product that assumed connectivity simply stopped working. We re-architected Myi to work offline first: workers could capture and assemble a complete packet on the device, assign it to an OEM engineer while disconnected, and have it sync and dispatch automatically the moment Wi-Fi or cellular coverage returned. Designing for the harshest environment made the product more resilient everywhere else.

The final design

By July 2018 the whole product was redesigned for a single React Native codebase: a calmer feed sorted by priority and recency, packets with an explicit priority level, and one place to see a packet's details, members and responses. The founders' industry network tested the product at each milestone, so every major decision was proven on a real production floor before we built on it.

Myi sign-in screen with company email, password and sign-in with Touch ID
Sign in
Packet feed grouped by day, with photos of machines, priority flags and resolved badges
The feed, by priority
New packet form with a message, name, description and a priority level
A new packet, with priority
Packet details with description, members and a response thread
Details, members and responses

Outcomes

Days → under 45 min
to resolve an equipment failure, as reported by the clients
38,000 users
in more than 120 facilities, including Bimbo Bakeries and Sara Lee
Up to $6,000 a day
saved on each engineer dispatch avoided in field testing

Shipped in twelve months across web, iOS, and Android, on schedule and under budget. Onboarding accelerated, and each plant gained a shared, searchable body of solutions that no longer depended on any one person's memory.

What I learned

Choose the stack for the product, not the plan. We started with native iOS and Android apps and a Rails web app, and moved to React Native partway through. Those early framework choices slowed development, and real-time video took considerably longer to engineer than we had scoped.

Hardware is a partner you don't control. We prototyped smart-glasses capture for fully hands-free video. One platform was locked against the customization we needed, and Google Glass proved too costly and complex to integrate at the time.

Protect the team's capacity. The client supplied a broader device matrix than one person could test in the time available, and when our full-stack engineer left partway through, I absorbed his work and renegotiated the timeline and budget. Shipping on time and under budget is not the full measure of success. Now I define scope and working boundaries with clients from the outset, so a team can sustain its best work rather than spend it.