As CSOO at NexPatch, Philip owns strategic business development and the path from first requirement to signed contract: scoping the solution, the effort, the price and the proposal. He translates technical architecture into decisions that management and the business side can back together, and tests every commitment against whether it still holds after go-live.
His conviction: AI projects rarely fail on the technology. They fail on too many interfaces and promises that are too big. That is why he works with defensible numbers, measurable acceptance criteria and clear service packages that settle from the outset who is accountable for running the system.
Philip brings over 14 years of experience in strategy, project leadership and business development, including at APEMENDelights and BlueCacadu.dev.
“A proposal is only good once we can deliver on it in production.”
Specialization
Education
Key Achievements
CURRENT TECH STACK
Articles by Philip Foitzik
13 ArticlesOwn GPUs vs token pricing: what AI on your own hardware really costs
Your own GPUs or pricing per token? We show you at what volume running AI on your own hardware pays off and which operating costs are often left out.
Managed AI operations: what needs to happen after go live
What has to happen after an AI system goes live: monitoring, retraining, updates, access rights. Here is how you organize AI operations as a managed service.
Private AI: definition, distinctions and when it pays off
What is private AI? We define it, compare it with cloud AI and on premises, explain the US CLOUD Act and show when your own data center pays off.
AI vendor exit strategy: how to take your models, data and prompts with you
Your AI vendor exit strategy: which data, models, prompts and logs you take with you when you switch, in what format, and what must be settled before you sign.
How to check AI vendor sovereignty: 10 pieces of evidence you should request
How to check AI vendor sovereignty: 10 pieces of evidence, from hardware location to exit, and how you recognize a good answer. With a process for your IT team.
Model drift explained: why AI systems get worse after a few months
Model drift explained: why AI systems get worse after a few months, the typical causes, how you measure drift and how to plan retraining as a fixed process.
Measure AI project success: the baseline before you start
We show how to measure AI project success: with a baseline before you start, a suitable metric for each use case and clear thresholds to scale up or stop.
SLA for AI systems: how to agree on monitoring, retraining and response
An SLA for AI systems covers more than availability: quality, costs, retraining, updates and severity levels. Here are the components you should agree on.
Fixed price AI project: what a pilot can deliver in 72 hours and what it cannot
A fixed price AI project needs a narrow scope and a measurable goal. We show what a pilot delivers in 72 hours, what it does not, and what comes next.
LLM Hosting and § 203 StGB: What Law Firms Must Verify in AI Providers
What § 203 StGB and § 43e BRAO mean for LLM hosting: three hosting models compared and a verification checklist for law firms selecting AI providers.
Demand forecasting: definition, methods and data needs
Demand forecasting explained: what it predicts, how it differs from sales forecasting and MRP, and which methods and data it needs.
From pilot project into operations
Why AI pilots in mid-sized businesses fail and how to transition into reliable ongoing operations: six action areas from data integration to service level agreements.
How long does an AI implementation take?
From initial consultation to productive ongoing operations: realistic timelines, typical bottlenecks and why 72 hours are enough for a pilot.
Meet The Team
Work directly with the team that ships.
From architecture to rollout, you collaborate with the same core team responsible for delivery quality and measurable business results.