Workshop
Requirements and options in one room
B.A. Gonczarek as Solution Architect. Bartosz Rogulski on ML and agentic AI.
Case Study
Mapping out architecture for Machine Learning-based software
Spectrally's Raman systems were already in industrial plants. The software around the models had to grow with the most demanding customers — QA, usage metrics, analytics — without guessing the next three years in a backlog. Vstorm ran a workshop and wrote the path down.
The outcome
A map before more ML code
This engagement is consulting, not a shipped spectrometer. The published artifacts are a focused workshop, a written report of current state and options, and a set of follow-on activities to execute. We do not publish a Raman accuracy figure or a latency SLA for Spectrally.
Quality-assurance support, customer-facing usage metrics, and analytics were the expansion Spectrally named. Those are the opportunities in the report — not features we claim as live.
Report
Current state and a path to execute
Written, then presented to Spectrally's executive team.
Roadmap
Initiatives after the slide deck
Follow-on work to turn the options into results — the workshop was not the product.
Client voice
Working with the Vstorm team was a really positive experience for us. They quickly understood what we do, what challenges we face, and where we want to go. The workshop was super focused, full of good questions and fresh ideas – it helped us organize our thoughts and plan the next steps in a way that actually makes sense. I appreciated how pragmatic and honest the whole process was. I'm excited to see how the ideas we developed together will shape the future of our software and products.
Robert Stachurski
CEO and Founder at Spectrally
About the client
Spectrally is a deep-tech startup in Poland. It does real-time chemical analysis with Raman spectroscopy: laser on the line, molecular composition without pulling a sample to a lab. Industrial processes see the chemistry as it happens.
Commercial traction meant the ML-fueled software had to stop being a side effect of the hardware and become a product in its own right.
Vstorm's impact
Vstorm's impact, the TL;DR
- Architecture workshop sized to Spectrally's actual Raman and ML stack
- Requirements collected for QA support, usage metrics, and analytics
- Written report: current state, opportunities, a path to execute
- Presented to the Spectrally executive team
- Follow-on activities to turn initiatives into results
The challenge
The hardware was in plants. The software had to grow up.
Success in industrial Raman meant more ambitious customers, and those customers wanted more from the software: quality-assurance support, customer-facing usage metrics, analytics. Machine learning was already in the business. The architecture had not been drawn for that scope.
Spectrally asked Vstorm for foresight, not for another model. The question was how the software should evolve — then a document the executive team could act on.
How we delivered
Workshop, report, then the work after the room
Proof of Value here is a path the client can execute, not a demo notebook on Raman spectra.
One room, real constraints
A consulting workshop to collect requirements and name opportunities. Led by B.A. Gonczarek, Vstorm co-founder, as Solution Architect, with Bartosz Rogulski, senior engineer, on ML and agentic AI.
- Requirement capture
- Opportunity list
- Architect + ML in the room
Proof of Value as a written report
Current state and future options, on paper, with a path to execute. Presented to Spectrally's executive team so the next steps were not a hallway memory.
- Current-state write-up
- Opportunity map
- Executive readout
Initiatives after the readout
A set of activities aimed at turning the report into results. The published engagement ends at that handoff — not at a claimed production feature list.
- Follow-on plan
- Execution path
- Shared next steps
How it works
From the workshop table to a document the board can hold
Spectrally already knew Raman. Vstorm's job was software evolution: what to add, in what order, without pretending a slide is an architecture.
Consulting workshop Requirements and opportunities with a Solution Architect and senior ML engineer
↓
Strategic report Current state and a path to execute, presented to executives
↓
Roadmap execution Follow-on work to turn initiatives into results
Spectrally × Vstorm — from workshop to roadmap
Results
Thoughts organized. Next steps that make sense.
Robert Stachurski's published line is the result that matters here: a focused room, honest questions, a plan the company can actually run. We do not list shipped QA dashboards as if they were in this contract. The published artifacts are the workshop, the report, and the path after it.
Work with us
Ready to see how agentic AI reshapes ML software architecture?
Meet directly with our founders and PhD AI engineers. We will walk through real implementations from 30+ agentic projects and the practical steps to integrate them into your workflows.