ИИ-планировщик Skody AI помог двум цехам устранить узкие места и повысить соблюдение сроков

Published Flexible, AI-driven software has helped two job shops solve scheduling bottlenecks before they occur and improve their on-time delivery rates in the process. Software Shop Management Software Senior Associate Editor, Modern Machine Shop Share Plans never survive contact with reality. Three Sigma Manufacturing, a job shop in Kent, Washington; and Yeager Manufacturing, a job shop in Winter Park, Florida, both encountered this problem many times with regards to job scheduling. A setup would take too long, or a shopfloor expert would get sick, or a piece of material wouldn’t arrive on-schedule, and on-time delivery and profitability would take a hit. Scheduling software from Skody AI has enabled both shops to tackle setup with more flexibility, resulting in surging on-time delivery rates and confidence in long-term planning. The Hardware-Software Disconnect Three Sigma tackles a broad spectrum of capabilities to handle difficult jobs in the commercial aerospace, spaceflight, motorsport and precision industries. It uses five-axis mills, three- and four-axis VMCs, an HMC, lathes, twin-turret machines, a multitasking machine, hones, deep hole drills and a CNC grinder, while also manufacturing its own carbide stick tooling to improve iteration of end mills for its work. The parts it makes go from the size of a finger to the size of a person’s torso, with most of its five-axis parts between 10 and 15 inches long, and its larger machines making parts within a stroke of one meter. Frankel says his staff at Three Sigma Manufacturing was quick to adopt Skody AI because it assists in letting them do their job without requiring manual input. “People are good and decent,” he says. “You just need to give them tools in order to enable them to be helpful.” Image courtesy of Skody AI. Tolerances for ground parts can go as low as 0.0005 inch, diameter tolerances on holes can be down to 0,001 inch, and President Ken Frankel says that the true position and profile tolerances on its milled items are typically hard to achieve. What’s more, most of the shop’s work uses materials about 50 HRC, with heat-treated alloys, stainless steels, Inconel, titanium and other hard materials in regular use. The shop runs around the clock from Monday mornings to Friday evenings, with occasional weekend work, and automates through multi-pallet machines with high tool counts, but Frankel says robots are not suitable for most of the shop’s jobs. While Frankel says profits are not as vital to the shop as the love of the challenge, over the past few years he has wanted to account for costs to make jobs more efficient. Hardware expenses made this unfeasible for a first pass at optimization, so when Skody AI CEO Alex Klimov reached out in early 2024 for assistance testing a software-only solution for optimizing scheduling, Frankel offered to collaborate with Skody and let the company test its software with Three Sigma’s high-mix, low-volume job mix. Yeager Manufacturing started in 2018, and has since become a two-time INC 5000 company due to its swift growth. The company has moved five times as it has expanded, with lots of growth due to large, short-term projects in 2023 and 2024. The shop has spent 2025 and 2026 filling the gaps from these projects with durable smaller contracts and using the time to enhance its capabilities and prepare its current building for more work. Founder Zach Yeager focuses on being upfront with delivery timelines, and targets reaching these as much as possible for both the shop’s fabrication and machining work. Yeager says his shop can produce just about anything except castings, as it performs sheet metal and tube work, general fabrication, powder coating, five-axis machining, three-axis machining and turning. As long as the parts fit within a five-foot cube, the shop can work with them. Most of the machining work has 0.005-inch tolerances, with ±0.001-inch hole diameters or tolerances in the tenths for some tight-tolerance turned parts. Most jobs have volumes between 30 to 500 parts, with turning providing some higher order numbers the shop handles through automated lights-out work. The shop uses a robot attached to one of its laser welders, a robot loader for one of its lathes and one of its mills, and two bar feeders. Its workload covers a diverse range of industries including aerospace, defense, agriculture, drone manufacturing, sewer inspection truck manufacturing, and fire truck and ambulance manufacturing. This diversity of work and combination of machining, fabrication and finishing in one business made scheduling a pain point for Yeager. Most software the shop tried either focused on scheduling people (which worked well for fabrication) or scheduling machines (which worked well for the machine shop) but did not support both. When Skody representatives reached out and offered a demo, Yeager tried the software. After finding the team’s engineers were open to adding new features based on the shop’s feedback, and that it would be compatible with both people-focused and machine-focused scheduling, Yeager adopted it in February 2026. Schedule Drift The main problem Skody tackles for both shops is how to handle schedule drift, or how reality on the shop floor differs from the scheduled plan. Zach Yeager says this might occur due to changes in machine capacity, part demand or everyday delays like someone taking sick leave or a machine needing maintenance. The effects of this drift can be drastic, as when Yeager’s on-time delivery rate slipped to 70% during a period with many expedited orders, with 80% being a hard cap for the shop in better times. First in, first out scheduling proved ineffective, and attempts to reconfigure the schedule cost hours only for the shop to see additional deviations within a shift that had knock-on effects for deadlines. To counter this, Skody acts like “Google Maps for production planning,” as Klimov puts it. It forms a schedule optimized around on-time delivery based on data sourced from a shop’s ERP, then adjusts as new complications emerge. Users can move onto the next job if their current one hits roadblocks, and insight into the shop’s full workload reduces situations where a well-intentioned staff member optimizes for one job but unintentionally sets multiple other jobs behind schedule. After pressing the button to generate a schedule, the system defaults to showing the shop’s work cells, which tasks need to get done in each, and which tasks have been assigned to which person. As per Yeager’s needs, it can also show machine-based and worker-based schedules. Klimov says users can click on recommendations to see the raw data behind them and the reasoning chain that led to that recommendation, a feature which Frankel says improved shopfloor acceptance. Skody also includes a bill of materials function that lists what materials a shop needs to manufacture its parts, what tools and materials are already in inventory, and what the lead time will be to acquire the rest. Yeager says his shop also tracks subcomponents through this function, streamlining the full scheduling of a job and its components. The system shows two weeks’ worth of scheduling by default, but supports looking further out toward future jobs, with delivery maps per customer and across customers. Klimov says this enables top-down planning for which revenue will enter the company in a particular month, and bottom-up planning as the software enables users to simulate alternative production scenarios (like buying another machine, running over time or reassigning someone between departments) to reach that revenue target. While anyone can conduct a simulation on the schedule, Klimov also notes that managers can set permissions to be sure only designated people can overwrite the current production plan with a new one. While the software itself cannot foresee complications and prevent schedule drift, Klimov says the schedule predicts which jobs will be on a tight schedule so that shops can negotiate with their clients about possible extensions, with data to back up their request. Frankel says Three Sigma’s on-time performance has improved since his partnership with Skody began and continues to trend positively. He continues that his shop increasingly relies on Skody day to day, as it helps them communicate more effectively about scheduling data through pictures and charts, rather than by sifting through textual reports. Doing so has also helped ensure they don’t focus so heavily on achieving on-time deliveries in the short-term that they lose any chance at it a few months down the road. User Experience Three Sigma had a hand in influencing the software to meet this goal, as Frankel and his team have provided feedback to improve the legibility of the data. This includes additional dashboards, as well as the use of different colors to illustrate if a job is delayed by a bottleneck, and by how much (a job two days late is very different from a job twenty days late, Frankel jokes). Frankel also says his team has grown more apt to using the software as these improvements have continued, as “a system that causes you to do more work to get the value out of it is one that generally is not going to have lots of people flocking toward it.” Yeager has also made its contributions to the software, as its suggestions helped develop a skill matrix the system can take into account when assigning people to jobs. Zach Yeager says this has been useful for determining machine-operator pairings in fabrication, removing some bottlenecks in that department. By the time I spoke to him in May 2026, three months after starting with Skody in February 2026, Zach Yeager said he has already seen results of 85-90% on time delivery, trending toward 95% by summer. What’s more, he says it has enabled his shop to become proactive about future bottlenecks, mitigating or eliminating them weeks before they have a chance to slow production. Related Content Five-Axis Machines Speed NASCAR Engine Production Moving from an aging set of five-axis mills to more advanced machines enabled Hendrick Motorsports to dramatically improve its engine production. Read More Sponsored Building a Foundation for AI Success on the Shop Floor Strong ERP infrastructure and proper planning enable shops to harness the power of AI. Read More Sponsored The Smarter Way to Take Full Control of Your CNC Machine Shop Designed to bridge the gap between CAM programmers and shop floor operators, SolidShop provides a seamless, real-time solution for managing G-code, tracking production and eliminating costly mistakes.   Read More 2 Secondary Coordinate Systems You Should Know Coordinate systems tell a CNC machine where to position the cutting tool during the program’s execution for any purpose that requires the cutting tool to move. Read More Read Next ERP Software Gives Shop Control Over Scheduling After relying on time cards for labor tracking and dealing with hot stickers on almost every job, Watson Grinding switched to Global Shop Solutions’ ERP software for automated tracking and intuitive scheduling.  Read More How Proper Scheduling Leads to a Successful Machine Shop Proper scheduling of jobs is paramount when it comes to running an efficient machine shop. Read More Manufacturing Scheduling System Keeps Shopfloor Priorities Straight Focusing on the now rather than adhering to a plan improves throughput, allows for variation and lays a strong foundation for predictive analysis.   Read More