Fisent: Interview With Founder & CEO Adrian Murray About Building Versus Buying AI Solutions 

By Amit Chowdhry • Aug 27, 2025

The emergence of generative AI (GenAI) has reignited a classic enterprise debate: build bespoke solutions internally or license third-party platforms? In this Q&A, Adrian Murray—Founder and CEO of Fisent Technologies—breaks down the key considerations organizations must weigh when harnessing GenAI for process automation. The conversation zeroes in on how businesses can best deploy these tools to optimize repetitive, content-heavy workflows that still rely heavily on human input.

The Current Situation

Pulse 2.0 (Amit): What are the traditional arguments favoring building enterprise software solutions in-house, and how do these apply to adopting enterprise GenAI applications?

Fisent (Adrian): Companies have historically decided to build solutions in-house for several key reasons. Some of these considerations are especially relevant for GenAI-enabled applications.

One major draw is the ability to create a solution that is perfectly tailored to specific business needs, workflows, and requirements, ensuring a precise fit. This approach provides companies with full control over the solution’s features, integration strategies, and the direction of future development. For GenAI, this means complete autonomy over model selection, fine-tuning, and data handling. While the potential for long-term cost savings by avoiding licensing fees is a traditional motivator, this can sometimes overlook the significant operational costs associated with running and internally maintaining any application.

Building in-house can also create a competitive differentiation by developing a unique solution that provides a distinct market advantage. Furthermore, it allows for better integration with existing legacy systems, as internal teams have a deep understanding of the company’s IT infrastructure.

Additionally, an in-house build leverages internal knowledge and expertise, and provides flexibility to innovate and experiment with new advancements without being dependent on a vendor.

While these traditional arguments hold some weight, the rapid pace of LLM development, its underlying complexities, and the specialized expertise required, significantly alter the calculus, making the build path far more challenging, unsure, and more costly than with traditional applications.

The Case for Commercial GenAI Solutions

Pulse 2.0 (Amit): What are the compelling arguments for buying a purpose-built enterprise software solution, particularly in the context of GenAI-enabled process automation?

Fisent (Adrian): The argument for buying or licensing software solutions is often rooted in efficiency and risk mitigation. These arguments become even more pronounced in the burgeoning era of enterprise GenAI implementation.

  1. Time to Value: With a well-defined process, design, configuration, and deployment implementation can be completed in just four to 10 weeks, leading to rapid, tangible results.
  2. Ensured Success: A GenAI solution provider’s success depends on its ability to deliver value, ensuring it’s uniquely motivated to see the project through to a positive outcome.
  3. Risk: A licensed software product is delivered with a pre-defined roadmap, ensuring that product is stable, secure, and future-proofed against evolving needs and technologies.
  4. Maintenance: A subscription model includes on-going maintenance, roadmap innovations, and compliance updates, offloading these responsibilities from internal teams.
  5. Support: Dedicated vendor support teams offer expertise, troubleshooting, and guidance, particularly with the application of GenAI expertise that is in short supply.
  6. Predictable costs and licensing models: Clear cost structures help with budgeting and financial planning, contrasting sharply with the often-unpredictable expenses of in-house GenAI development (talent, infrastructure, LLM usage, rework).
  7. Trust & Security: Essential capabilities like data governance and security are included in a subscription model, as the provider is responsible for maintaining these features to serve a broad market.

Pulse 2.0 (Amit): What specific capabilities does a purpose-built GenAI-enabled process automation solution offer to challenge the “build it yourself” mentality?

Fisent (Adrian): When considering a GenAI-enabled process automation solution for streamlining repetitive, human-dependent tasks, look for sophisticated capabilities that do more than in-house builds are likely to do, such as:

  1. Full Model Optionality: Invaluable is the ability to seamlessly switch between multiple LLMs (e.g., GPT, Gemini, Claude, Llama) based on the specific task, performance, and accuracy requirements. Internal builds often lock enterprises into a single model or require significant re-engineering to switch. Look for a solution that offers model measurement and optionality.
  2. Enterprise-Grade Integration & File Type Versatility: GenAI-enabled process automation solutions are designed from the ground up to integrate with existing enterprise systems and BPM layers. Crucially, they boast the ability to process a vast array of non-standard file types (e.g., .XLS, .EML, .ZIP), a capability that is incredibly complex and time-consuming for internal teams to develop and maintain.
  3. Proven Use Cases and Scalability: Commercial solutions come with validated success in real-world scenarios across vertical industries (banking, insurance, healthcare) and business functions (HR, sales, customer service). They derisk implementations with proven ability to scale for tasks like contract review or claims processing.
  4. Continuous Innovation and Updates: Solution providers have dedicated teams and significant investment solely focused on perfecting their GenAI offerings, ensuring they stay current with the rapidly evolving GenAI landscape. An internal team, with competing priorities, simply cannot keep pace.
  5. Built-In Governance and Compliance: Essential for enterprises, these solutions integrate critical features like data privacy, auditability, and industry-standard controls (SOC 2 Type II) from inception, saving immense development and compliance effort for internal teams.
  6. True Cost Efficiency: While internal development seems to eliminate licensing fees, it incurs high upfront and ongoing costs for talent acquisition, infrastructure, maintenance, and the direct payment of LLM providers. A commercial solution spreads these costs across many customers and transactions, leading to significant savings and a faster ROI. Factoring in frequent reworks and errors, internal GenAI projects can quickly escalate in cost and risk being abandoned altogether.
  7. Expert Support and Application Focus: Beyond providing the technology, solution providers offer deep domain expertise and support to apply GenAI to solve real business problems, often delivering tangible results in weeks. This contrasts with internal builds that can take months for development, followed by months of implementation and testing. A recent MIT report found that 95% of GenAI pilots at companies are failing, further underscoring this point. However, when companies partner with specialized vendors offering licensed GenAI solutions, their projects succeed about 67% of the time, while internal builds succeed only one-third as often.
  8. Future-Proof Architecture: Designed to adapt to new models and technologies as they emerge, these solutions mitigate the risk of internally built solutions becoming quickly obsolete or requiring costly overhauls.

Navigating Emerging Applied GenAI Process Automation Tools

Pulse 2.0 (Amit): How do the claims of AI agent building tools further complicate the “build versus buy” decision for GenAI automation?

Fisent (Adrian): Tools for quickly creating AI agents have entered the “build versus buy” debate. While these tools promise to democratize AI development and speed up prototyping, they often fall short of the mark for robust, enterprise-grade process automation. This is due to a fundamental trade-off: these tools prioritize breadth over depth. They aim to support a wide range of applications but frequently lack the deep, purpose-built functionalities, specialized pre-trained models, and specific integrations needed for complex enterprise environments.

The challenges go well beyond the initial creation of an AI agent. Many of these tools fail to address the critical last-mile issues of enterprise deployment, such as handling diverse content types, ensuring consistent output quality, managing edge cases, robust error handling, security, and seamless integration with existing business process management (BPM) layers.

Furthermore, they often have significant gaps in governance and compliance, lacking the necessary auditability, data privacy, and security features that are non-negotiable for regulated industries and sensitive data. Enterprises also benefit from having a single, consistent solution for automating repetitive tasks across their operations, rather than managing a collection of independently developed agents that can lead to fragmentation. While these tools are great for building a proof-of-concept, scaling to handle thousands or millions of transactions reliably and efficiently is a different challenge entirely, one that commercial solutions are designed to solve from the start.

Finally, even with “low-code/no-code” tools, the need for specialized GenAI expertise doesn’t disappear. You still need talent to handle tasks like orchestration, prompt engineering, fine-tuning models, managing model drift, and troubleshooting issues, meaning the core problem of finding specialized talent remains.

In essence, while AI agent building tools can be valuable for exploration and rapid prototyping, they often highlight the critical need for a more robust, specialized, and enterprise-ready solution when it comes to implementing GenAI for core business process automation.

Pulse 2.0 (Amit): What does the Fisent BizAI solution offer?

Fisent (Adrian): Solutions like Fisent BizAI address many of these build considerations by offering high levels of customization and control within a robust, managed platform.

Fisent BizAI is an agentic software solution that harnesses GenAI to quickly automate the repetitive, human-dependent tasks common throughout all types of organizations. It functions by natively consuming unstructured content from diverse sources like documents, emails, audio, and images. BizAI then applies GenAI-enabled “actions” to classify, split, extract, verify, and analyze essential information, digitally replicating human interpretation. This processed information is then integrated with core enterprise systems, enabling straight-through process automation and significantly increasing productivity and improving business outcomes. Fisent BizAI is model-agnostic, allowing Fisent to identify the optimal LLM for specific use cases, balancing quality, speed, consistency, and other factors, enabling actionless upgrades.

The solution demonstrates proven impact across various verticals, consistently achieving 98%+ accuracy in processing complex unstructured content in live production environments. For instance, Orco Group, a bank holding company, successfully leveraged Fisent BizAI for its member entities, resulting in a 90% decrease in errors, a significant increase in processing speed, and over 70% reduction in document processing time compared to previous manual processes.

Fisent BizAI provides the dedicated focus and capabilities necessary to move beyond experimentation to real, scalable problem-solving in a matter of just weeks.