AI Navigator case study: from manufacturing use cases to operational prioritisation
SkuzaAI worked with an anonymised manufacturing organisation on AI Navigator use-case prioritisation for manufacturing operations. The public draft keeps the client anonymised and uses only claims approved for publication. The product context is AI Navigator use-case prioritisation for manufacturing operations: a focused AI workflow designed around an operational constraint, named ownership and measurable review.
Challenge Statement
Workflow objective
Connect the AI workflow to a real operating process.
Evidence boundary
Keep human review explicit before any client-facing output.
Decision clarity
Measure the workflow before presenting the result as a public claim.
Prioritisation constraint
The useful leadership question for an anonymised manufacturing organisation was where AI Navigator use-case prioritisation for manufacturing operations could change a real workflow.
The Co-Creation Roadmap
Step 1: Lay the Groundwork for Future Automation
We understood there was a clear need for a digital product, which would offer an extended digital service. The company sold the product to big, medium, and small companies, which were costly to serve with traditional channels. The digital product would potentially offer the advantage of zero marginal cost of distribution.
We also understood that the reason the company failed internally to deliver a digital product was that “software eats the business model.” New digital value propositions reconfigure the existing business so significantly that they almost create a new business model. The company knew how to execute an existing one, improve it every year, but did not know how to build, test, and find a new one, which is what startups are good at.
Two Important Decisions We Had to Make
We needed to design the change like a startup and the only way to do this was to build this externally, allowing for much-needed flexibility, different budgeting, and project management. This is the first we consider product Co-Creation.
It would also allow us to tap an external lean business model talent pool who knew how to move in sprints, tests, and user-centric design. How to design different go to market strategy, differently.
It is essential to understand the reasons why a startup is not a smaller version of a larger company.
Workflow learning
one workflow, one business owner, one metric set, one review path and one correction rule for wrong or incomplete agent output.
To better understand this, consider the opposite. According to conventional wisdom, the first thing every business does when setting a new venture is creating a business plan—a static document that describes the size of an opportunity, the problem to be solved, and… Business plans rarely survive first contact with customers.
At the same time, we onboarded the company talent into a lean startup framework to align cultures.
Prioritisation Roadmap
- computer vision can support package completeness, labelling and real-time inspection.
- demand forecasting and inventory planning are visible candidates for AI support.
- Technical document translation was selected as a documentation workflow candidate for the AI Navigator roadmap.
- the material points to workflow redesign, not only tool selection.
- Readiness review filters each use case by evidence, owner and delivery readiness before it moves forward.
Reinventing the wheel is a mistake corporations and other startups make quite often. It is slow, silly, and the price is steep. The best way to build a prototype is to use what others made in the field plus a lot of duck tape:
Workflow objective
Connect the AI workflow to a real operating process.
Evidence boundary
Keep human review explicit before any client-facing output.
Decision clarity
Measure the workflow before presenting the result as a public claim.
Prioritisation constraint
The useful leadership question for an anonymised manufacturing organisation was where AI Navigator use-case prioritisation for manufacturing operations could change a real workflow.
At this point, we have a great understanding of the disruptive environment: what and how startups solved them, including their revenue models, namely:
Source consolidation
computer vision can support package completeness, labelling and real-time inspection.
Use-case scoring
demand forecasting and inventory planning are visible candidates for AI support.
Reviewed workflow
Technical document translation was selected as a documentation workflow candidate for the AI Navigator roadmap.
Readiness review
the material points to workflow redesign, not only tool selection.
Reviewed workflow
Readiness review filters each use case by evidence, owner and delivery readiness before it moves forward.
We were ready to start the Co-Creation process at this point.
- Connect the AI workflow to a real operating process.
- Keep human review explicit before any client-facing output.
- Measure the workflow before presenting the result as a public claim.
We now had all the most innovative blocks, and we wanted to piece them back together. We workshopped with the company through iterations of revenue/pricing models relying on the lean canvas and validation boards:
Source consolidation
computer vision can support package completeness, labelling and real-time inspection.
Use-case scoring
demand forecasting and inventory planning are visible candidates for AI support.
Reviewed workflow
Technical document translation was selected as a documentation workflow candidate for the AI Navigator roadmap.
Readiness review
the material points to workflow redesign, not only tool selection.
We know what, how, and the go-to-market. We ran profitability simulations to identify investment, cost, expected cash generation and ROI, and alignment with the company strategy, but we still missed one significant element.
Disruptors focus on “hair on fire” customers — people in dire need of your product. We understood clearly that this was an SME because they had been asking the company for a solution already.
Step 4: Conduct Co-Creation
We understood their need for automatic routing changes reducing driver and dispatch involvement and resources on the SME side. It was a fresh idea, and we needed a quick proof we were solving a real problem before we committed to coding.
Source consolidation
computer vision can support package completeness, labelling and real-time inspection.
Governed workflow
The technical decision was to keep AI Navigator use-case prioritisation for manufacturing operations behind evidence extraction, copywriting, tone, fact-check.
Prioritised roadmap
AI Navigator workshop narrowed more than 20 candidate AI ideas to four implementation-ready use cases: packaging vision checks, demand forecasting, technical document translation.
Evidence boundary
The draft presents these as observed case metrics, not guaranteed results for other companies.
Use-case scoring
demand forecasting and inventory planning are visible candidates for AI support.
Step 5: Built It
We drew heavily from routing tech and push notifications tech available from scouted startups at Step 2.
We built and “shipped” 2 prototypes in under ten weeks:
Reviewed workflow
Technical document translation was selected as a documentation workflow candidate for the AI Navigator roadmap.
Readiness review
the material points to workflow redesign, not only tool selection.
The technical decision was to keep AI Navigator use-case prioritisation for manufacturing operations behind evidence extraction, copywriting, tone, fact-check and human-review gates before publication.
As soon as we had early prototypes, including first user interface prototypes, we went out of the building again and tested, analyzed, and co-created the product with the company customers, taking in another new batch of user stories. Customers provided us a few feedback lines, but these two were the most promising:
Workflow learning
one workflow, one business owner, one metric set, one review path and one correction rule for wrong or incomplete agent output.
Next decision
AI Navigator workshop narrowed more than 20 candidate AI ideas to four implementation-ready use cases: packaging vision checks, demand forecasting, technical document translation.
The useful leadership question for an anonymised manufacturing organisation was where AI Navigator use-case prioritisation for manufacturing operations could change a real workflow instead of adding another disconnected tool. Baseline evidence covered AI Navigator workshop narrowed more than 20 candidate AI ideas to four implementation-ready use cases: packaging vision checks, demand forecasting, technical document translation and order-email processing, so the work could be judged against operational metrics.
The implementation pattern behind AI Navigator use-case prioritisation for manufacturing operations is narrow: one workflow, one business owner, one metric set, one review path and one correction rule for wrong or incomplete agent output. Map the next AI use case against real workflow evidence for an anonymised manufacturing organisation, confirm the workflow owner and keep the public story tied to metrics the team can stand behind.
- Connect the AI workflow to a real operating process.
- Keep human review explicit before any client-facing output.
- Measure the workflow before presenting the result as a public claim.
Business Impact
The early and continuous assessment was extremely positive among the SMEs in the local market. The next decision is to scale this up to all European and Asian markets.
The most important metrics we have designed, tracked, and used for further improvements via co-creation was as follows:
It was an essential development for the company. We had proven to the stakeholders, inc. BOD in the US, Turkey, the UK, that the entrepreneurial approach is better at finding new ideas and cheaper at building them, cutting time to market significantly.
Business Impact
SkuzaAI used AI Navigator to turn workshop evidence into a prioritised manufacturing AI backlog. The source material points to computer vision for package completeness, labelling and real-time inspection; demand forecasting and inventory planning; and raw-material or component purchasing as repeatable decision workflows. The case draft keeps these as implementation candidates that need owner, data-readiness and review decisions before delivery claims are made. The uploaded material points to several operational signals: computer vision can support package completeness, labelling and real-time inspection; demand forecasting and inventory planning are visible candidates for AI support; Technical document translation was selected as a documentation workflow candidate for the AI Navigator roadmap.; and the material points to workflow redesign, not only tool selection.
We have shipped a better product as 40% of features were discovered during the work with customers, increasing “unfair advantage.” That proved that Minimum Viable Products (prototypes) could be built and tested with real drivers in real trucks in days/weeks days, rather than months/quarters. The new service improved the acquisition & activation of new customers by 35%.
There are 196,000 SME trucks across the country, which we run experiments and launched as the first territory. Potential subscription revenues at $25/month per vehicle and 4% adoption represent a $2.2M opportunity per year. An achievable incremental 5% market share increase is estimated at $103m per year in fuel revenue (in the first country market alone) or 46x the subscription potential.
Lessons
Why had the company not been able to find the new product offering internally?
The answer to this is that “software eats the business model”—digital value destroys old business models. The challenge is how to find the new one. In this case study, we proved that co-creation is the perfect method.
Our customer was good at executing an existing model.
Startups are good at building new ones.
- computer vision can support package completeness, labelling and real-time inspection.
- demand forecasting and inventory planning are visible candidates for AI support.
- Technical document translation was selected as a documentation workflow candidate for the AI Navigator roadmap.
- the material points to workflow redesign, not only tool selection.
- Readiness review filters each use case by evidence, owner and delivery readiness before it moves forward.