Systems | Development | Analytics | API | Testing

MCP is a new way to integrate LLMs and AI agents with third-party data sources and APIs. It significantly improves how we build tool integrations by eliminating duplicated code and providing a centralized interface for multiple agents to access shared tools. Today, we’re excited to announce the release of Kong’s MCP Server for the Kong Konnect platform. This empowers customers to integrate AI agents and query LLMs to discover APIs, services, and traffic analytics in real time.

Worldwide, over 150,000 companies rely on Salesforce for business-critical operations such as sales, marketing, or customer service. Given its critical role in business success, ensuring that Salesforce applications function correctly is not just a best practice — it’s a necessity. However, Salesforce environments are highly dynamic. Frequent updates, extensive customizations, and numerous third-party integrations mean that changes can introduce unexpected issues.

Large language models (LLMs) have redefined artificial intelligence (AI), pushing the boundaries of natural language processing (NLP) and enabling machines to understand, generate, and manipulate human-like text. From chatbots and content creation to legal and medical applications, LLMs are transforming industries at an unprecedented pace. But what makes these models so powerful? How do they work? And what challenges do they pose?

Teams often find themselves managing a big number of test cases, many of which are nearly identical except for variations in input data. This redundancy leads to a higher risk of missing critical test coverage. Test parameterization solves this challenge by allowing testers to re-run the same test logic with multiple data iterations - teams define parameters within a single test script and execute it dynamically via a dataset with different values.

Despite fears of job loss, QA professionals are leaning into AI faster than ever, according to Katalon’s newly released 2025 State of Software Quality Report. The report reveals that testers using AI tools are twice as likely to fear being replaced by them, a paradox that underscores the profession’s evolving relationship with automation.

Millions of users rushed to ChatGPT overnight, all craving Studio Ghibli-style art. What started as a fun trend quickly went viral, pushing OpenAI’s servers to their limits. The "Ghibli Trend" wasn't just another online craze — it became a live performance and load-testing scenario for OpenAI. Social media users began sharing Ghibli-inspired AI images, creating a massive buzz.

Pharmaceutical labeling is an ideal use case for AI because it’s a complex process that requires high levels of accuracy. Inaccurate labeling can result in: With recent breakthroughs in AI technology, pharmaceutical companies have rushed to explore its potential. But many have not seen the impact they expected. The problem isn’t the AI. It’s how pharma companies are using AI.

Adopting AI? Risky. Not adopting AI? Riskier… It’s an interesting conundrum illustrating a broader truth: playing it safe is sometimes risky. To succeed, companies need to take smart risks—and those who do often achieve the greatest rewards. But a risk that is smart for one organization may not be as smart for another, so what risk is right for you?

BigQuery data preparation is generally available and integrates with BigQuery pipelines and BigQuery repositories and Git.