Four in build, four on deck, and the order is the point

The department holds itself to a limit of four builds in active development at a time. Nineteen initiatives are on the books: four active, four sequenced next, and a backlog that stood at eleven until the August 11 kickoff review added to it. A place on the list is a commitment to order, and dates get committed only when a build goes active.

In active build

1. 01The WorkPro connector. Live as of August 12 for Bigge Claude users, after passing IT cybersecurity review. Install it from the capabilities menu, sign in with your WorkPro credentials, and fleet and service records answer in plain language. The functionality is a starter set that grows as each addition passes review, and it already handles things like standing up a field service order from inside a chat. Most of the roadmap depends on this layer existing. This issue's From the Engineer's Desk covers how it is built.
2. 02Parts manual search and lookup. Every serial number has its own manual and most are not searchable today. The vision, straight from the working sessions with mechanics and foremen: type the equipment number painted on the side of the crane, land in the right serialized manual, see the exploded diagram, and describe the problem in plain language. One lookup serves service, the shop, and the parts buyers. Optical character recognition on the manual set is already under way, starting with the highest volume models.
3. 03One service history per unit. Pick a crane and its story is split: inspections in FastField, work orders in WorkPro. Roughly 200,000 FastField inspections migrate into WorkPro so the full history lives on the unit record. We tried this in 2021 with the automation tools of the era and it broke faster than it could be fixed, because nothing in the pipeline could read a PDF and know what kind of inspection it was. The difference now is an agent that categorizes each record with a confidence score and raises its hand on the ones it is unsure about, instead of failing silently. Internally we call it CraneFax, and yes, that is the pitch.
4. 04Warranty claims and campaigns. Sany carries a three year bumper to bumper warranty, the Cummins powertrain warranty runs five, and Bigge expects more than 300 Sany units in the fleet within a year and roughly 600 by the end of next year. Today the whole recovery process lives on a manual log of about 2,600 lines, with roughly 200 claims filed this year and volume about to multiply. At kickoff, Brian Noga added a requirement the log cannot answer today: visibility into every short paid and denied claim, so the record walks into the next purchase negotiation with us. That gets built in.

On deck, and why this order

Four more sit sequenced behind the actives, and the sequence is a dependency chain rather than a preference list. Each one either feeds the next or waits on data the actives create.

1. 01Parts stocking analysis. Historical analysis of parts spend, quantity, frequency, and lead time, so supply chain holds the parts that put cranes back to work instead of the parts that gather dust. We have been burned by slow moving stock before, which is why this has to be specific rather than broad. It unlocks when the parts index makes serial level part identification reliable, which is exactly what the active parts build produces.
2. 02Quality control on repair descriptions. When a mechanic closes a work order, the description of what was actually done is often too thin to analyze, and every downstream ambition dies right there. This sequences directly behind inspection quality control, which sits at the top of the backlog. That one was an active build until August 11, when we demoted our own project to make room for the WorkPro connector. Priority lists that never cost you anything are decoration.
3. 03ServicePro shop estimates. Today a shop estimate leans on an average across categories too broad to predict anything. An engine line item does not know whether it is a spark plug or a starter. The working idea from kickoff is to build estimates from the parts actually ordered on comparable work orders, starting with mid size and large projects where a full budget exists. Two honest caveats: it needs both quality control layers above it to exist first, and a dedicated estimator hire is under consideration, so this one may leave the AI list entirely. We would rather lose an initiative than force one.
4. 04BELA commitment tracking and rate capture. Track fully executed BELA commitments so the same unit is never promised twice, block commitments on units that are not service ready, enforce price book rates at quote approval, and flag at-risk units before their commitment date. The kickoff supplied the war story: signed agreements that never reach WorkPro because notifying the ASO is a manual button, and cranes held verbally on a side sheet in the meantime. Brian Noga put the standard plainly: if it is not in WorkPro and there is no blue box, it is not true. The first measurement is already requested: how many units were committed more than once in the last twelve months.

What the kickoff added

The board was reviewed with Brian Noga on August 11, and it grew there. He backed the shape of the portfolio and then pushed on the gap in it: almost everything above is operational, and you cannot forget how we make money. Fair. The customer facing pillar stays unfunded as a bar rather than a policy: the moment an idea is good enough, it gets a slot. Finding that idea is now standing homework for this department.

1. 01Facilities and lease visibility. Every leased property carries a contract, terms, exit requirements, and a monthly cost, and nobody owns the full picture today. Wanted: pooled data and alarms when something drifts outside a guideline.
2. 02HR compliance training tracking. Who is required to take what, who has not taken it, and a system that actually comes knocking.
3. 03Safety. Our number one core value, and leadership wants a standing AI presence in it. Scoping starts with the safety team.
4. 04Customer self service. Invoices on demand, off rent requests from an app, an account a customer can log into. Under discussion, and it starts with asking customers and the sales team what they would actually use.
5. 05The crew board. A utilization view for people like the one for cranes: who is booked, where, until when, with alarms when a job runs long. Most of this lands on the WorkPro side, and it is recorded here so the dependency stays visible.

The rules every build answers to

1. 01A named owner in the business, outside this department. Unowned work does not start.
2. 02A measured baseline of what the work costs today. Baselines for the four actives are being captured now, and that is deliberately the first block of work.
3. 03Agreed stop conditions. A pilot that misses gets shut down on schedule instead of quietly maintained forever.
4. 04A person stays in the loop in everything we ship.

The number these answer to

Utilization is the north star metric for the fleet. Each of these builds shortens a wait somewhere between a unit coming off rent and going back on: a record nobody can find, a part nobody can look up, a claim sitting in a pile. That is the connection we will keep reporting against.

Put something on the board

1. 01Run Automation Scout in Claude or ChatGPT and send the brief, or describe the task in the AI General channel.
2. 02Bring the cost with it if you can: runs per week, minutes per run. Submissions with numbers move first.