Multi-agent AI-support facilitating highly customized order completion
The Vstorm team built a multi-agent product advisor using the PydanticAI Python-centered framework, with FastAPI for inter-application processes and a powerful RAG vector store for matching requests with products based on Mixam's always-up-to-date internal knowledge.
Mixam is a self-publishing company that primarily provides printing and fulfillment services for independent authors, publishers, and creators on a global scale. They specialize in high-quality print production, including books, magazines, and other printed materials.
Mixam's services are designed to make it easier for individuals and small publishers to produce and distribute their works without the need for large-scale traditional publishing houses.
Mixam was established in 2007 in the United Kingdom but operates on a global scale, expanding its services to meet the needs of the global market. One of the key aspects of the expansion is the usage of AI in accordance with the user-friendliness of their self-publishing platform.
Visit Mixam →- Industry
- Printing
- Headquarters
- United Kingdom
- Company size
- 51–200 worldwide employees
- Founded
- 2007
Vstorm's impact, the TL;DR
- 10,000+ users now use Mixam's custom tailored AI agent each day, processing 100k custom orders per month
- Improved customer conversion to final sale from ~20% to ~40% overall
- Within 1 day of launching the assistant in Australia, Agent achieved a 11.76% increase in orders created
- Of all the quotes provided by the AI Agent — 62.11% end up being paid and confirmed
- Implemented specially tailored three-agent system to guide 70% of new customers
- Agents access 15 distinct tools to act as fully informed Mixam product consultants
- Validation processes and constrained generation eliminate system hallucinations
- Agents maintain the flexibility of natural language interactions while ensuring outputs remain 100% accurate to Mixam's offer
The challenges of making AI a printing expert
It is typical for people to get overwhelmed when options are too abundant. The same applies to artificial intelligence, which may struggle if it has to pull options from an excessive variety of components to choose from. Such are the challenges of applying LLMs to specific business needs.
The AI agent for Mixam had to be engineered to create order specifications that would be validated when taking orders in. With challenges like this, engineering expertise is key to blending the indeterministic nature of language models with the rule-based backbone of the final delivered solution. This goes against the common perception that the deployment of AI agents is straightforward — it almost never is in a business context, and it requires an art of engineering and expert knowledge to make it work as a reliable part of any business solution.
This also applies to securing the solution in such a way that it's not thrown off and begins talking garbage, protecting it from the reduction of its reliability and from increasing hallucinations. Setting up what are called 'guardrails' forbids the agent from picking up topics that are not directly related to printing orders. As a result, the agent refuses to give restaurant recommendations or offer cupcake recipes. When it acts in the role of a self-publishing advisor, it focuses solely on helping the end-user make the right selection of Mixam options to have the final print meet the customer's needs.
From complex printing options to seamless customer experience
Vstorm designed and implemented an AI agent to help Mixam's customers navigate the company's complex printing offers, smoothing the customer experience in navigating complex publication processes.
Cooperation with Vstorm began when Mixam had already begun using AI elements in various operations. However, the company's ambitious goals required reaping the full potential of AI in increasingly demanding and complex processes.
The initial Vstorm project was centered around creating a satisfying experience for new users who were just starting their self-publishing journey. From book format to paper thickness and structure, it's easy for any non-publishing professional to get lost in the variety of choices that need to be made before their first publication materializes in the desired form.
A helping hand in placing your first book order
Mixam realized that 70% of new users need help in choosing their way through a plethora of options that newcomers might not know anything about. The gutter, the bleed, the paperweight, and its coating are not things that book authors generally think about. Nor should they.
Now, with the help of Agentic AI, they can let the system work out the details by simply stating the intended purpose of the publication and answering a few prompting questions that the chatbot asks for clarity.
The difference between a plain-vanilla chatbot and Mixam's Agent is in the knowledge available for agentic AI solutions to suggest options. While generic bots draw replies from internet sources, Mixam's AI Agent works by using exact product specifications from a catalogue to create orders that can be fulfilled by Mixam in the user's location — taking into consideration any differences in units and product offerings across Europe, the Americas, and other locations.
From chit-chat to expert advice
The challenge of creating a true agentic AI printing expert required overcoming typical issues related to large language models, such as hallucinations. Specifically, to prevent the bot from playing the guessing game and instead offer what's actually printable requires the narrowing down of its options to concrete product specifications. Vstorm achieved this by having the AI agent use Mixam product options, pulling them from Mixam's publishing systems with API calls.
This grounded the solution in choosing existing combinations of cover, page, and binder options. However, having a simple AI bot use the components of Mixam products and build an order from them was not enough — the solution needed to know how to recommend the right combinations of those components to potential customers. This was accomplished by supplying the agent with a Mixam knowledge base that combines both best practices and the most common choices that users have made and proven happy with.
The Vstorm team built the solution using the PydanticAI Python-centered framework, with FastAPI for inter-application processes and a powerful RAG vector store for matching requests with products based on Mixam's always-up-to-date internal knowledge of their product features.
The choice of this environment was made as it is built with safety as a priority and focuses on strict data validation. As a framework relied upon in the healthcare and finance industries, it was deemed adequate for self-publishing.
Results that speak for themselves
Measurable outcomes delivered from day one of the Mixam AI Agent going live.
Systematic evolution through metrics and monitoring
Mixam's approach exemplifies the industry shift from "AI projects" to "AI products." The infrastructure includes sophisticated monitoring, centralized prompt management, and performance optimization.
This operational maturity — treating AI as production infrastructure requiring monitoring, version control, and staged deployment — distinguishes systems built for long-term business value from experimental prototypes. The repository structure itself tells the story of incremental evolution: test logs, performance benchmarks, shop-specific validation scripts, and careful Git management all point to a team iterating based on data, not assumptions.
Meet directly with our founders and PhD AI engineers. We will demonstrate real implementations from 30+ agentic projects.
Related articles
Build RAG that retrieves the right thing
Most RAG pipelines fail at retrieval, not generation. Learn how to prepare your data, choose your stack, and evaluate what actually matters in production.
Read articleLet AI agents delegate what they can't do alone
Subagentic delegation is a design pattern where a supervisor agent decomposes a complex goal and routes each subtask to a specialized subagent — running in parallel, at scale.
Read articleText-to-workflow cuts engineers' tedious task time to seconds with Agentic AI platform
Synera's AI agent platform now generates complex engineering workflows in under 3 minutes — down from 2 hours — with zero hallucinations through multi-step validation.
Read articleReady to see how Agentic AI transforms your business?
Book a free 45-minute session with Vstorm founders. We'll map one real process worth automating.