
AI Operational Intelligence for B2B Manufacturers
There are two ways to start: the two-week Strategy-First AI Pilot, or the four-week Operational Intelligence Scan.
Why we start with your goals
Over fifteen years of working with Taiwanese B2B companies, we’ve seen the same pattern again and again. A company buys a system or a tool because a competitor did, or because a vendor pitched it well. Eighteen months later the tool is running and the invoice is paid, and the company is no closer to where it wanted to be. The technology worked fine. It was simply pointed at the wrong problem.
AI makes this mistake faster and more expensive, because it’s easy to start and hard to evaluate. Every vendor has a demo and every integrator has a use-case list. What most companies are missing is a way to decide which option actually connects to the outcome they need.
Two ways to start
Strategy-First AI Pilot
about US$2,600
- Scope: 1 workflow
- Your team’s time: one 60-minute executive call plus half a day on site
- What you get: a working AI prototype on your data and a 6 to 8 page brief
- Best for: testing on one workflow whether AI is worth investing in
Operational Intelligence Scan
about US$16,000
- Scope: 3 workflows, 6 interviews
- Your team’s time: Week 1 meetings and interviews on site, then the Week 4 presentation
- What you get: a 25 to 35 page diagnostic report, one Quick Win delivered and running, and an implementation roadmap
- Best for: leadership teams ready for a bigger AI decision who need the full picture
The Pilot fee is credited in full against the Scan if you commission it within 60 days of handover. We also offer introductory terms for the first three companies; just ask.
Strategy-First AI Pilot
Every Pilot starts with the same three questions, answered before anyone talks about tools:
Your answers decide the workflow. We don’t arrive with a list. In practice the answers often point to front-office work, such as qualifying and routing incoming enquiries, preparing quotes and proposals, or handling orders and customer follow-up.
How it runs
Week 0: preparation
The CEO or GM and two to four department heads fill in a 15-minute survey. Then a 60-minute executive call agrees the workflow, and you send us a data sample such as recent quotes or a week of incoming enquiries.
Day 1: half a day on site
With six to eight of your people we map the workflow end to end, rate whether your data and systems are usable (red, yellow or green), and score three to five AI options together. One is chosen for the prototype. The decision-maker joins the first and last 30 minutes.
Days 2 to 10: build
Our technical team builds the prototype on your data sample, and we write the Pilot Brief.
Day 10: handover
In 60 minutes we demonstrate the prototype on your data, walk through the Brief and agree the next step. If there’s no reason to go further, we’ll tell you.
What you receive
- The strategic rationale in writing: your answers to the three questions
- A map of one end-to-end workflow, built with your team
- A red, yellow or green rating of your data and systems readiness
- Three to five AI options, scored and ranked
- A working AI prototype on your data, demonstrated live
- A Do / Don’t / Later recommendation with effort and cost ranges
- A 6 to 8 page Pilot Brief and one costed next step
about US$2,600
The full fee is credited against the Operational Intelligence Scan if you commission it within 60 days of handover. If you don’t continue, everything from the Pilot is still yours, to implement with us, with another partner or with your own team.
Introductory terms are available for the first three companies.
Operational Intelligence Scan
How it runs
Week 1: discovery on site
A 90-minute leadership session, six interviews across the three workflows (45 to 60 minutes each), a data infrastructure assessment, and a first view of the Quick Win.
Weeks 2 and 3: analysis
Remote workflow mapping, a People, Process and Data assessment of each workflow, ROI modelling and opportunity scoring, and development of the Quick Win. You get a status update every week.
Week 4: delivery
A 90-minute executive presentation, handover of the Quick Win with training, and a discussion of the implementation roadmap.
What you receive
- A 25 to 35 page diagnostic report
- Analyses and maps of the three workflows
- An AI opportunity matrix with ROI estimates
- Red, yellow and green data readiness scores
- A Do / Don’t / Later implementation roadmap
- One Quick Win, delivered and running
- The executive presentation and 30 days of support after delivery
about US$16,000
The Pilot fee is credited in full if you commission the Scan within 60 days of the Pilot handover.
What a finding looks like
This illustrative example shows how we reach a recommendation. A precision machinery maker with about 85 employees takes three to five days to send a quote, while its main competitors reply in one to two. After looking at sales quoting, production planning and quality reporting, the recommendations were:
Smart quoting assistant
AI matches each request against past jobs and helps sales prepare the quote. The data is already structured and the team wants it, so it can start right away.
Data integration
Production data sits in three separate systems (ERP, MES and the warehouse system). Until they are connected, no AI on the production side will work.
Full predictive maintenance
Retrofitting sensors on every machine costs more than it saves, and the current preventive maintenance is good enough. A sensorless tool-wear prediction is a better fit for now.
Who you work with

Richard deVries
Canadian, in Taipei for over twenty years, fifteen of them working inside Taiwanese B2B manufacturers and hidden champions on strategy, positioning and how they go to market. Richard leads every engagement personally, from the first call to the handover. He hosts a monthly AI Roundtable for business leaders and has written on AI in Taiwanese industry for CommonWealth Magazine.

Jonathan Rudich
Jonathan has spent more than five years delivering AI, automation and data solutions in manufacturing, defence, healthcare and e-commerce, with a focus on making systems work in practice. His technical team built an AI imaging platform for a US teleradiology firm that saves it more than US$1 million a year, and an IoT and cloud analytics platform for a major process manufacturer in South Africa, supporting real-time traceability and predictive maintenance. On our engagements he judges whether your data and systems are usable, and leads the build.

Ofelia Yang
Has optimized global marketing and channel workflows, managing 80,000+ units a year across international markets. Works on the process side.

Lulu Ding
SAP specialist with 5+ years in insurance, cloud and semiconductors, and a full-stack developer. Works on the technical side of the prototype.

An Kung
Your day-to-day contact for scheduling, the data sample and keeping the project on track.
The team runs the project day to day in Chinese. Richard works out whether something should be built, and where. Jonathan and the team work out whether it can be built, and build it. Either way you get a straight answer, including “not yet” and “don’t”.
What's not included
No technology demo
No software vendor in the room, nothing to license and no platform for us to push. The prototype is built on whatever your systems can support.
No large system rollout
The Pilot prototype proves the approach on a data sample. Live integration into your CRM or ERP comes after, if the results earn it.
No long-term lock-in
Both options have a fixed scope and a fixed timeline, and you decide afterwards whether to continue.
AI Operational Intelligence FAQ
Should we start with the Pilot or the Scan?
If you want to test on one workflow whether AI is worth investing in, start with the two-week Pilot. If you’re ready for a bigger decision and want the full picture across three workflows, go straight to the Scan. If you’re not sure, one call is usually enough to tell.
What size of company is this for?
Mainly B2B companies with 50 to 1,000 employees, especially manufacturers: businesses with real operational complexity but without an enterprise-scale AI team or infrastructure.
Do we need AI projects running already?
No. It works best before any AI project starts. Many companies find out that the projects they had planned were aimed at the wrong workflow.
What data do you need, and is it kept confidential?
For the Pilot, only a small sample from the chosen workflow. We’ll tell you exactly what after the executive call. All workflow analysis, data and findings are covered by a confidentiality agreement.
Will AI replace our people?
We’re honest with your team about what AI will and won’t change for them. The people who run the workflow map it themselves, and they score and choose the AI options with us.
Is this a technology implementation service?
Neither option is a large system rollout. The Pilot delivers a prototype; the Scan delivers a Quick Win and a roadmap. If the results justify it, we can support the implementation, or your own team or another partner can take it on.
What does "Quick Win" mean?
During the four-week Scan, we find and deliver one small, concrete operational improvement, such as an automated step or a fixed data flow, so you see value right away instead of in six months.
Start with a call
Further reading: Where Should a Taiwanese Manufacturer Start with AI?