Case studies

From AI curiosity to production-ready value: extending digital platforms with AI on Upsun

5-minute read

Challenge

For established digital platforms, the primary barrier to AI adoption is rarely a lack of ideas, but fear of the unknown.

Lemberg Solutions found that nearly 80% of their clients shared the same technical anxieties: model accuracy ("what if it hallucinates?"), reputation risk ("what if a user posts a screenshot of a biased answer?"), and data privacy.

These fears often lead to "analysis paralysis," where companies hesitate to overhaul their infrastructure or risk their stable production environments for an unproven AI pilot.

Lemberg needed a way to move clients from "What if?" to "It works" without requiring a massive upfront investment or compromising existing systems.

Solution

Lemberg utilized Upsun to create a "discovery-to-delivery" pipeline that prioritizes business value over technical hype.

By leveraging Upsun’s Git-driven, standardized environments, Lemberg could build a functional Proof of Concept (POC) in weeks instead of months.

The solution utilized a modular "black box" architecture. Lemberg hosted the core Drupal CMS alongside dedicated AI services built on Python’s FastAPI.

Upsun’s ability to combine these stacks into a single, unified project meant that AI logic could be tested and iterated independently.

Crucially, Lemberg used Upsun’s preview environments as "playgrounds" where stakeholders could safely check (and even break) the AI logic against real data before any code touched the live site.

Results

Accelerated deal cycles: MVPs moved from kickoff to live in weeks instead of months, allowing clients to generate data-backed ideas quickly.
Drastic cost reduction: Enterprise clients reported paying half of their previous hosting costs while gaining more features and faster development cycles.
Bypassed legal bottlenecks: Upsun’s compliance certifications (e.g., ISO 27001) helped navigate complex IT and legal checklists for enterprise clients.
Eliminated "Staging Drift": Replicated production-identical sandboxes reduced DevOps time from days to roughly 30 minutes.

Bridging the gap between AI experimentation and business value

Lemberg Solutions acts as a strategic advisor, helping digital businesses ensure they have the necessary data foundations in place before implementation begins to ensure the AI actually creates value

"The time from project kickoff to development is smooth and fast because we can quickly spin up development instances on the Upsun platform using templates." — Roman Paska, Head of Drupal at Lemberg Solutions

Why Upsun was the right fit

For Lemberg, AI integration is an application evolution, not an infrastructure chore.

Upsun provided three critical advantages that made this transition seamless: multi-stack flexibility, safe testing through human-in-the-loop validation, and the safety net of instant backups.

One of the primary benefits was the platform’s inherent multi-stack flexibility.

While AI ecosystems typically live in Python, many digital platforms are built on PHP. Upsun allowed Lemberg to host FastAPI and Drupal side-by-side using a single YAML configuration.

This unified approach helped the team avoid the "operational sprawl" of managing multiple cloud vendors, keeping all services in a centralized and predictable environment.

To address critical accuracy concerns during development, Lemberg implemented "confidence levels" for AI output. Upsun’s isolated preview environments proved invaluable for this process, enabling specialists to verify low-confidence responses in a safe sandbox before they were ever exposed to end-users.

This human-in-the-loop strategy allowed stakeholders to test and iterate on AI features without risking the stability or reputation of the live production site.

Finally, the platform provided a critical safety net through its instant backup and restore capabilities.

In one instance, a client accidentally removed a critical page from their production site. Lemberg was able to quickly roll out a production backup to a new environment and manually restore the lost content, a capability that protected the client's reputation and saved hours of recovery work.

Removing the fear of production risk

One of the most significant barriers to AI adoption is the fear of "breaking the site." Generative AI introduces new variables that traditional digital platforms aren't always equipped to handle.

Lemberg addressed this fear by using Upsun's isolated preview environments as "playgrounds" for their clients.

Stakeholders could check, break, and iterate on AI features in a production-identical environment.

This capability was instrumental in securing internal buy-in, as it proved the AI logic worked within the context of the client’s actual data and infrastructure before a single line of code was merged into production.

Multi-stack flexibility for modern AI needs

AI workloads almost always require Python, while many of Lemberg’s clients run on Drupal or other PHP-based stacks.

Upsun’s multi-language support allowed Lemberg to build a FastAPI service as an additional "black box" service that interacts with the main application.

This modular approach ensures that the AI logic is scalable and secure, while the primary platform remains stable.

By hosting everything on a single CAP (Cloud Application Platform), Lemberg avoided the operational sprawl of managing multiple vendors, ensuring that the total cost of ownership remained low for the client as they moved from Phase 1 to full-scale production.

Beyond the MVP: reliability and compliance

The value of the platform extended into the non-technical hurdles of enterprise growth. Upsun helped Lemberg’s clients navigate complex legal and compliance requirements by providing a secure, auditable infrastructure foundation.

In one instance, a client was able to reduce hosting costs by two times while gaining more testing environments and faster development cycles.

The ability to quickly roll out production backups to new environments also served as a critical safety net, allowing the team to restore inadvertently removed content and resolve data issues without downtime.

Advice for digital leaders

The Lemberg team encourages leaders to move away from "fear-driven" development. By using a platform that handles the plumbing (automated updates, scaling, and compliance), teams can focus on data science and business logic.

"When managed services are available, we can start projects and go live within a few hours. Upsun is a proven platform that we have worked with for six or seven years. It helps us build things faster and manage projects more easily."— Roman Paska, Head of Drupal at Lemberg Solutions

About Lemberg Solutions

Lemberg Solutions is a technology consulting and engineering agency that helps digital businesses implement AI, IoT, and cloud-based solutions. They specialize in discovery, MVP development, and scaling digital products for SaaS, eCommerce, and enterprise clients.