Umbraco CMS is a .NET content management platform used by individuals and large enterprise for everything from personal blogs to global SaaS products. It's powerful, extensible, easy to use, and open source. And it's evolving to embrace AI in ways that genuinely improve developer and content author experiences.
The new features and capabilities are foundational, based on real world needs, taking advantage of the technical opportunities large language models (LLMs) provide, targeting discreet areas in the platform. There are official Umbraco blog posts that talk about philosophy, areas of focus, primers on the individual AI technologies, and more.
The goal of this post is to provide a high-level summary of the changes and who they affect.
Umbraco is constantly evolving, most notably releasing a new major version each year just after Microsoft likewise releases a new major version of .NET. This is a blessing and a curse. On one hand, Umbraco developers get to take advantage of new .NET performance, security, and other features. But supporting a sprawling ecosystem of Umbraco versions is a challenge, especially when the breaking changes and development patterns change so frequently and vary so much between major releases.
To address this, the Umbraco CMS team has added integrations to your AI agent platform of choice. These integrations provide additional technical information about Umbraco CMS, including best practices, as well as provide controlled access to your schema and data through the official Umbraco CMS APIs.
The MCP standard allows the hosting of very specific tools that can be used by AI agents to get things done in a more simple, direct way. Umbraco supports this by providing a separate MCP host application that developers can run to give their agents deep platform and data access through the Umbraco Management API.
There are a myriad of actions that your agent can perform using these MCP extensions, including categories from site structure and user permissions, to monitoring, analytics, localization, and advanced workflows.
They allow the agent to do things with the API that would normally be done using the back office. These are actions you perform during development, like creating document types and setting permissions.
Originally a Claude Code-only feature, but now available in other AI agent platforms like OpenAI Codex, agent skills are a way to tell the AI coding environment about a given technology, which speeds development and ensures that the correct development patterns (best practices) are followed.
Umbraco is providing agent skills that will allow you to better use your AI agent of choice to build on the latest versions of Umbraco (version 17+). It's important to note that as magical as they can appear to be, there are caveats:
Not a replacement for documentation. Skills point to the documentation and teach the AI how to apply it. You should still read the Umbraco documentation.
Not scaffolding tools. They don't generate boilerplate from templates. They teach the AI to write code that fits your specific requirements.
Not perfect. Skills improve results but cannot guarantee correctness. Always review generated code.
They allow you to work with your agent to write code for things like a back office property editor for a YouTube video picker, or an notification that adds metadata to uploaded media, etc.
You can also leverage AI tools in the Umbraco platform itself, through the platform agnostic AI integration in the back office. You simply add the official Umbraco AI package, and configure the appropriate AI provider. Supported providers include:
Anthropic
AWS Bedrock
Microsoft AI Foundry
OpenAI
You finish by adding a new profile (chat or embedding), which controls the type, usage, and behavior of the AI model instance. For example, you can set the model temperature to help with determinism. Note: the profile is the vehicle through which you use the AI instance in your own code.
You can use your configured AI tool in the back office through a chat interface, to help manage content and structure, clean up data, add media metadata, and more. You can also use the configured AI instance in your application code to provide AI-driven features with pipelines and guardrails, all using the Umbraco APIs (which use the Microsoft Extensions AI package under the hood).
The official Umbraco CMS documentation on AI features can be found at the links below. Though not exhaustive, these are key resources you should check out.
Umbraco in AI — Enhancements to make building with AI smarter.
Backoffice extension agent skills — install and use the Umbraco backoffice agent skills repository with Claude Code (and other AI coding agents)
AI in Umbraco — Add AI tools and capabilities to Umbraco CMS
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
In April of 2026 Anthropic released a new AI model named Claude Mythos. This isn't the typical mixture of experts (MoE) model you'd use to find a recipe or answer a health question. It's trained specifically to find software security vulnerabilities at scale.
Even when working with the largest enterprise codebases, Mythos can...
Identify vulnerabilities at scale across major operating systems, browsers, and critical infrastructure
Generate working exploits for the vulnerabilities it finds
Operate at a speed and volume that has no human equivalent in offensive security research
Find vulnerabilities that have been missed for decades
Operate largely autonomously, without human steering
In one documented example, the model identified a 27-year-old vulnerability in OpenBSD, which Anthropic describes as “one of the most security-hardened operating systems in the world.” OpenBSD is not a consumer operating system. It runs firewalls, VPN gateways, and critical network infrastructure across financial institutions and government networks globally. Even the MacBook Pro I'm using to write this post is running an operating system (macOS) that is based on BSD Unix.
The conventional approach to managing vulnerabilities has been a race between patch cycles and attacker awareness. Mythos-class AI accelerates the creation of exploits in ways that strain the ability to code against them.
Needless to say, exploiting a single vulnerability like the one found in OpenBSD could wreak havoc globally across industries overnight. So Anthropic decided to create Project Glasswing, a controlled-access program providing Mythos Preview to twelve major technology organizations, including AWS, Microsoft, Google, Cisco, CrowdStrike, Nvidia, and Palo Alto Networks, backed by $100 million in usage credits. Usage credits are key because the current cost to scan these kinds of codebases can be in the tens of thousands of dollars.
This controlled access gives the good guys a head start on plugging the most critical security holes before this technology gets into the wrong hands. Attackers, according to Anthropic's offensive cyber research lead Logan Graham, will get equivalent capability from other labs within six to eighteen months. OpenAI is reportedly finalizing a model with comparable offensive cybersecurity capability. And while Anthropic has committed to investing in improved guardrails for future models, guardrails can be broken.
We're in the early days of this new paradigm. The industry is handling it well, for the moment, albeit the head start won't last forever. Like the fabled tortoise in The Tortoise and the Hare, eventually the threat actors will catch up.
We're on the verge of seeing an increase in software security exploits. So it's important to focus on the basics, like upgrading your web platforms to their latest versions, and updating the OS and firmware of your Internet-connected computers and devices.
Start the process now. Your window of opportunity will eventually close.
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
While AI "hallucination" is typically viewed as a flaw (often providing comic relief), a recent whitepaper by OpenAI suggests it's actually a natural, emergent property of the way language models are currently trained and benchmarked.
Hallucinations are a logical consequence of optimization strategies that prioritize performance on standard benchmarks. These benchmarks are used to compare AI models, so the providers have every incentive to game the system so their models perform better than their competitors on these benchmarks.
Though that strategy does improve performance, it doesn't hide the inherent flaws in how models are trained.
For example, according to the whitepaper, on "Humanity's Last Exam" (HLE), a benchmark designed to be "google-proof" across dozens of fields, all reported state-of-the-art model scores were below 30% accuracy. This indicates a failure rate of over 70% on expert-level questions.
AI models are often inaccurate in both their answers and their self-assessment of correctness. Most current models on the HLE benchmark show calibration error rates above 70%. Though pretrained base models can be well-calibrated (with errors as low as 0.7%), post-training processes like reinforcement learning (PPO) can increase this flawed calibration to 7.4% or higher, making models more overconfident in their incorrect guesses
Also according to the report, if a certain percentage of facts (like birthdays) appear only once in a training data set, the model is expected to hallucinate on at least that same percentage of those facts. For example, if 20% of birthday facts are singletons, the model’s hallucination rate will be at least 20% for those prompts.
Because models are trained to recognize patterns in large data sets, statistical pressures encourage them to calibrate their performance to the training data. This is because they are trained to minimize cross-entropy loss. For a model to be well-calibrated, it must assign probabilities that reflect the true likelihood of a statement being correct. In these cases, the statistical objective of the model forces it to produce a "best guess" rather than admitting it doesn't know.
Benchmarks typically penalize honesty. Most influential benchmarks (like GPQA, MMLU-Pro, and HLE) use binary (0-1) scoring. In a binary scoring system, a correct answer earns 1 point, while an incorrect answer and an "I don't know" (IDK) response both earn 0 points.
Because there is no penalty for being wrong compared to being silent, the mathematically optimal strategy for a model is to always guess when in doubt. A model that always guesses will statistically outperform a "more honest" model that admits uncertainty on current leaderboards because guesses can be right some of the time.
Users expect modern AI models to "know everything". Even though they are trained on incredibly diverse data sets, this expectation is mostly unreasonable. Given this, any model that attempts to generalize beyond its training data must inherently risk hallucination. Otherwise, it would suffer from mode collapse, failing to produce the full range of valid human responses.
Even advanced techniques like Retrieval-Augmented Generation (RAG) or chain-of-thought reasoning do not eliminate this pressure because the underlying grading system still rewards guessing when these tools fail to find a definitive answer.
Hallucinations are a rewarded behavior. AI models are optimized to be "good test-takers," and in the current world of binary evaluations, hallucination-like guessing is the most successful survival strategy for a model aiming for the top of a leaderboard.
Service confidence is also a concern. If a model responds "I don't know" or "I'm not confident in this answer", how many users would abandon the service? Since an informed guess is correct some of the time, this is seen as a good compromise to ensure service confidence.
There are two obvious answers to this problem. One is to push the industry away from benchmarks and toward confidence scores. In this way the number of incorrect responses matters less than how often a model definitively provides a helpful response.
The second answer is to be dubious about the responses we get, knowing that a significant percentage of them are partially or completely incorrect. Be skeptical! As software developers our focus is to both use AI tools to provide faster, more varied services, and also to know when to check the recommendations and/or work performed by an AI model using the decades of experience we have in our field to do so. This mix of skepticism and "trust but verify" is what we bring to every client project and it has served us well.
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
AI data poisoning is a process where an attacker deliberately alters an AI model’s training data to influence its behavior, causing it to generate biased, misleading, or harmful output. This threat is now recognized as a major vulnerability by security organizations like OWASP.
According to Carnegie Mellon University Security and Privacy Institute: "Modern AI systems that are trained to understand language are trained on giant crawls of the internet," said Daphne Ippolito, assistant professor at the Language Technologies Institute. "If an adversary can modify 0.1 percent of the Internet, and then the Internet is used to train the next generation of AI, what sort of bad behaviors could the adversary introduce into the new generation?"
The Pravda network is a large group of fake news websites created by Russia in 2014. These sites target audiences in more than 80 countries and are designed to spread stories that support Kremlin disinformation. They work by repeating and amplifying messages from Russian media and pro-government Telegram channels. In 2024, the network expanded its efforts by launching sites focused on NATO and prominent political leaders such as Donald Trump and France’s President Emmanuel Macron.
To get around international restrictions on Russian state media, this network has shifted its tactics. Instead of relying only on traditional propaganda channels, it now tries to appear as a trustworthy source so that some of its content is used in resources like Wikipedia.
As a result, AI tools may unknowingly absorb and repeat these biased or false narratives. This can expose users to messaging that favors the Kremlin and criticizes Ukraine or Western governments when they interact with AI chatbots. It can influence elections. And it can drive people to make decisions that go against their self interests.
But fear not! There are ways to combat this problem. Some target the training process, and others put users in control. Here are a couple examples:
A blockchain is a shared digital ledger for logging transactions and tracking assets. You've undoubtedly heard this term when discussed in the context of cryptocurrency. Blockchains provide secure and transparent records of how updates to data are shared and verified due to the fact that existing information cannot be changed; only new items can be added.
In the context of AI training, if you need to change a fact, the original is never touched. So any new items that claim to revise the original stand out like a sore thumb. So by using consensus mechanisms, AI systems with blockchain-protected training can validate additions more reliably and help identify the kinds of anomalies that can indicate data poisoning before it spreads.
It's imperative that users corroborate information they find on the Internet with reputable sources, and this includes AI output. In a past article I used the example of Judge Julien Xavier Neals of the District of New Jersey, who had to withdraw his entire opinion after a lawyer politely pointed out that it was riddled with fabricated quotes, nonexistent case citations, and completely backwards case outcomes.
The old adage "don't believe everything you read" is never more true than when it refers to Internet-derived content. Be skeptical and check your sources!
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
Vector search has been around for a long time. For example, Google has been using it since the late 1990s. This powerful, almost magical technology serves as a core component of most web and app services today, including modern AI-powered search using retrieval augmented generation (RAG). It provides a low-power way to leverage AI's natural language processing to find data with semantic context.
A vector database stores data as numerical embeddings that capture the semantic meaning of text, images, or other content, and it retrieves results by finding the closest vectors using similarity search rather than exact matches. Sounds like an AI large language model (LLM) right? This makes it great for web and app content because users can search by meaning and intent, not just keywords. So synonyms and loosely related concepts still match. It also scales efficiently which is one reason large organizations like Google have used it.
A happy side effect of how these platforms work is that they're also good at handling misspellings, to a point. To really get robust handling of spelling variations, however, two strategies tend to be common:
Spell correct the actual search text before using it
Include character n-grams in your vector database entries
Character n-grams are vector embeddings (like commonly used semantic embeddings) that break text into overlapping sequences of characters, allowing vector search systems to better match terms despite typos, inflections, or spelling variations. Without these n-grams, a misspelled query like "saracha sauce" would likely return a higher score for "hot sauce" entries. But including character n-grams, a combined (fused) search would more consistently return a higher score for items with the correct spelling "sriracha sauce".
Using these n-grams can better handle searches with:
typos
missing letters
swapped letters
phonetic-ish variants
common misspellings
How does this work? At a high level, it adds a character match capability to the standard semantic search used by most vector database implementations. Here's a quick example of what happens under the hood. Take the first word in our previous example:
sriracha
3-grams: sri, rir, ira, rac, ach, cha
4-grams: srir, rira, irac, rach, acha
saracha
3-grams: sar, ara, rac, ach, cha
4-grams: sara, arac, rach, acha
Shared grams:
shared 3-grams: rac, ach, cha
shared 4-grams: rach, acha
So even though the beginning is wrong (sri vs sa), the ending chunks that carry a lot of the distinctive shape of "sriracha" survive (racha, acha, cha). And since the second word is the same, they have even more matching grams.
When these matches are fused with semantic matches, it adds weight to the correctly spelled "sriracha sauce" entry, yielding a better match set.
When it comes to including character n-grams, there are only a couple changes you need to make to a standard semantic vector database implementation:
When you generate embeddings, you also need to generate character n-gram embeddings; this is true both when you store data in the database, and when you search.
When searching, you need to execute a search both on the semantic vectors and the n-gram vectors, then fuse the results using Reciprocal Rank Fusion (RRF), which is a great way to merge disparate result sets and combine the scores.
The following samples will fill those gaps. They are written with C# for .NET, which is part of a common stack we use to build cross-platform, secure, high-performance web and mobile apps and services for our clients. We also tend to prefer the vector database Qdrant for its performance, maintainability, and open source model. So that is also referenced in the samples.
References to AiService.GenerateEmbeddingsAsync() are not covered here. Essentially it's a method to generate standard semantic embeddings. Replace that with your own (likely existing) method. And references to QdrantService.Client are merely references to a standard Qdrant client provided by the Qdrant Nuget package.
Note: Some of the code was generated by AI, but was reviewed and refactored by an actual human developer (me!).
First, you need a way to create n-grams. The CharNGramEmbedding class below will fill that gap. It allows you to generate character n-grams for a given string, and it also provides a method for fusing the semantic and n-gram search results into a single, weighted result set.
Now that you have the character n-gram generation and fusion handled, following is an example of performing a Qdrant upsert of a sample food object, including both sets of vectors.
Lastly, the following example shows how you can search the Qdrant data using both sets of vectors. Embeddings (semantic and character n-grams) for the prompt are generated and used in the search.
For the best fused results each search (semantic, n-grams) needs to return 3-5 times the number of the final result set. This is because you're trying to recover a good final top-K from two imperfect retrievers. If each retriever only returns exactly K (or close to it), you often don't have enough overlap + near misses to let fusion do its job, especially when the two methods return different items, and rank positions aren't directly comparable.
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
AI prompts that reveal insight, bias, blind spots, or non-obvious reasoning are typically called “high-leverage prompts”. These types of prompts have always intrigued me more than any other, primarily because they focus on questions that were difficult or impossible to answer before we had large language models. I'm going to cover a few to get your creative juices flowing. This post isn't a tutorial about prompt engineering (syntax, structure, etc.) it's just an exploration in some ways to prompt AI that you may not have considered.
This one originally came to me from a friend who owns the digital marketing agency Arc Intermedia. I've made my own flavor of it, but it's still focused on the same goal: since potential customers will undoubtedly look you up in an AI tool, what will the tool tell them?
If someone decided not to hire {company name}, what are the most likely rational reasons they’d give, and which of those can be fixed? Focus specifically on {company name} as a company, its owners, its services, customer feedback, former employee reviews, and litigation history. Think harder on this.I would also recommend using a similar prompt to research your company's executives to get a complete picture. For example:
My name is {full name} and I am {job title} at {company}. Analyze how my public profiles (LinkedIn, Github, social networks, portfolio, posts, etc.) make me appear to an outside observer. What story do they tell, intentionally or not?This prompt is really helpful when you need to decide whether or not to respond to a prospective client's request for proposal (RFP). These responses are time consuming (and costly) to do right. And when a prospect is required to use the RFP process but already has a vendor chosen, it's an RFP you want to avoid.
What are the signs a {company name} RFP is quietly written for a pre-selected service partner? Include sources like reviews, posts, and known history of this behavior in your evaluation. Think harder on this but keep the answer brief.People looking for work run into a few roadblocks. One is a ghost job posted only to make the company appear like it's growing or otherwise thriving. Another is a posting for a job that is really for an internal candidate. Compliance may require the posting, but it's not worth your time.
What are the signs a company’s job posting is quietly written for an internal candidate?Another interesting angle a job-seeker can explore are signs that a company is moving into a new vertical or working on a new product or service. In those cases it's helpful to tailor your resume to fit their future plans.
Analyze open job listings, GitHub commits, blog posts, conference talks, recent patents, and press hints to infer what {company name} is secretly building. How should that change my resume below?
{resume text}You'll see all kinds of wild scientific/medical/technical claims on the Internet, usually with very little nuance or citation. A great way to begin verifying a claim is by using a simple prompt like the one below.
Stress-test the claim ‘{Claim}’. Pull meta-analyses, preprints, replications, and authoritative critiques. Separate mechanism-level evidence from population outcomes. Where do credible experts disagree and why?Even if you're a seasoned professional, it's easy to get lost in jargon as new terms are coined for emerging technologies, services, medical conditions, laws, policies, and more. Below is a simple prompt to help you keep up on the latest terms and acronyms in a particular industry.
Which terms of art or acronyms have emerged in the last 12 months around {technology/practice}? Build a glossary with first-sighting dates and primary sources.There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
Consider this... is a recurring feature where we pose a provocative question and share our thoughts on the subject. We may not have answers, or even suggestions, but we will have a point of view, and hopefully make you think about something you haven't considered.
As more people use AI to create content, and AI platforms are trained on that content, how will that impact the quality of digital information over time?
It looks like we're kicking off this recurring feature with a mind bending exercise in recursion, thus the title reference to Ouroboros, the snake eating its own tail. Let's start with the most common sources of information that AI platforms use for training.
Books, articles, research papers, encyclopedias, documentation, and public forums
High-quality, licensed content that isn’t freely available to the public
Domain-specific content (e.g. programming languages, medical texts)
These represent the most common (and likely the largest) corpora that will contain AI generated or influenced information. And they're the most likely to increase in breadth and scope over time.
Training on these sources is a double edged sword. Good training content will be reinforced over time, but likewise, junk and erroneous content will be too. Complicating things, as the training set increases in size, it becomes exponentially more difficult to validate. But hey, we can use AI to do that. Can't we?
Here's another thing to think about: bad actors (e.g., geopolitical adversaries) are already poisoning training data through massive disinformation campaigns. According to Carnegie Mellon University Security and Privacy Institute: “Modern AI systems that are trained to understand language are trained on giant crawls of the internet,” said Daphne Ippolito, assistant professor at the Language Technologies Institute. “If an adversary can modify 0.1 percent of the Internet, and then the Internet is used to train the next generation of AI, what sort of bad behaviors could the adversary introduce into the new generation?”
We're scratching the surface here. This topic will certainly become more prominent in years to come. And tackling these issues is already a priority for AI companies. As Nature and others have determined, "AI models collapse when trained on recursively generated data." We dealt with similar issues when the Internet boom first enabled wide scale plagiarism and an easy path to bad information. AI has just amplified the issue through convenience and the assumption of correctness. As I wrote in a previous AI post, in spite of how helpful AI tools can be, the memes of AI fails may yet save us by educating the public on just how often AI is wrong, and that it doesn't actually think in the first place.
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
The AI train is currently barreling through Hypeville, and it's easy to be dubious of anything branded with "AI". My previous post, Simulated intelligence definitely factors into this topic. And as I wrote at the time, AI is not what people think it is. But even with its flaws it is a transformative technology and it's here to stay. And one AI technology you're likely hearing/reading about lately is AI agents, and it's one to pay attention to.
AI agents are not (always) covert operatives. They are AI powered services that perform tasks, not just answer questions. They can work as an assistant, helping you as you work, or independently perform tasks on your behalf. Agents are specialists, and can be trained to perform tasks that would otherwise be performed by a person.
AI agents are already being used in your favorite web services, from social media platforms to accounting software. In those cases they're typically used behind the scenes to provide features you may not have thought were possible. For example, your accounting platform could auto-categorize or reconcile transactions before you even sign in for the day. And you may have already seen your favorite AI chat platform scour the web on your behalf to give you more up-to-date answers.
Co-working is another (more visible) way you can experience them. An agent trained on your company information (think bios, product information, marketing materials) can work with you to build your next presentation or update sales materials. It could be used to analyze comments or feedback based on context and sentiment, flagging items for follow up. It could find documents based on heuristics, like phrasing inconsistencies in your brand identity. All the odd edge cases you where you had to manually dig and process information could be delegated to an AI agent.
Here's one that everyone will love. Imagine being able to ask your computer to not only find that system setting you can never find, but even ask it to just "do the thing". For example, if there are numerous settings that control performance mode on your laptop, the agent knows which ones to change for you before you run that important presentation.
If all this sounds interesting, there are ways you can play with AI agents on your own and work them into your daily life in meaningful ways. As a software developer I've been using AI agents to enhance my workflow. One I've been using is Github Copilot. It can help perform refactoring and create unit tests, saving me typing and cognitive load so I can focus on planning, strategy, and creative tasks.
You can also try ChatGPT agent. ChatGPT can now do work for you using its own computer, handling complex tasks from start to finish. According to OpenAI:
You can now ask ChatGPT to handle requests like “look at my calendar and brief me on upcoming client meetings based on recent news,” “plan and buy ingredients to make Japanese breakfast for four,” and “analyze three competitors and create a slide deck.” ChatGPT will intelligently navigate websites, filter results, prompt you to log in securely when needed, run code, conduct analysis, and even deliver editable slideshows and spreadsheets that summarize its findings.
There is also a new standard that allows AI platforms to communicate with web services to more reliably and securely perform tasks. It's called Model Context Protocol (MCP). As this tech works its way through various software and services, we'll see more agent-driven features that make a real difference in our lives.
As I wrote at the outset, it's very likely that you're already using AI agents on your favorite social and productivity platforms but weren't aware. They'll be powering more of our digital lives over time and, personally, I welcome our new simulated intelligence overlords (ha!).
It's safe to say that AI agents are the real deal. So we should all strap in and hold on tight. This is going to be exciting!
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
The term AI is more of misleading brand name than an accurate description of technology. That distinction is causing real problems for people in many industries who increasingly rely on AI tools. One example of this is the case of Judge Julien Xavier Neals of the District of New Jersey, who had to withdraw his entire opinion after a lawyer politely pointed out that it was riddled with fabricated quotes, nonexistent case citations, and completely backwards case outcomes. You'd think that a judge would be more careful, but then again, if they're not tech-savvy you can see how they could be misled by the promise of AI.
In the 1983 movie War Games, a teenage computer whiz accidentally hacks into a U.S. military supercomputer (named WOPR for "War Operation Plan Response") while searching for video games, unknowingly triggering a potential nuclear crisis. As the system begins running a simulation it mistakes for a real attack, he must race against time to convince the AI to stop a global thermonuclear war.
So yeah, WOPR is what people today consider AI; artificial general intelligence (AGI) to be specific.
Released in 2008, the movie Iron Man features a billionaire inventor Tony Stark who was captured by terrorists, after which he builds a powerful armored suit to escape his captivity and later refines it to fight evil using a digital personal assistant named JARVIS (Just a Rather Very Intelligent System) to coordinate all his technology through voice commands.
JARVIS is also AGI.
OpenAI recently released ChatGPT 5 to mixed reviews. One such review was the blueberry test by Kieran Healy. He asked ChatGPT "How many times does the letter b appear in blueberry" to which ChatGPT responded "The word blueberry has the letter b three times". No matter how hard he tries to convince the AI that there are only 2 letter Bs in the word blueberry, ChatGPT is absolutely positive there are 3.
People expect and believe that AI has human-level or higher intelligence and is able to understand, learn, and apply knowledge in any domain, adapt to new problems, and reason abstractly. That would include knowing how to spell the word blueberry.
What we have with AI today is really a marketing issue. It is not a mechanical turk. It is a transformational technology and it's here to stay. It will improve over time, and it has the potential to make our lives better in many ways. But we need to understand what it is, and more importantly, what it is not.
Then what is AI?
Modern large language models (LLMs) like ChatGPT are trained on vast datasets covering a wide range of human-created content—from websites and books to transcripts, code, and other media. Instead of simply storing this data, the model uses neural networks to learn patterns in language, encoding knowledge as mathematical relationships. When generating responses, the LLM doesn’t look up answers in a database; it predicts the most likely sequence of words based on the context, drawing on statistical patterns it learned during training. LLMs operate through probabilistic prediction rather than direct retrieval, and they lack true understanding or reasoning in the human sense. Without ongoing training on the latest human-generated content, LLMs will become increasingly less useful.
So we're dealing with a simulated intelligence, not an artificial one. It's like the difference between precision and accuracy. You can be very precise, but completely wrong. So it does matter. There is no real intelligence at play here. Which is why the word blueberry has three Bs, the judge's opinion has non-existent citations, and glue was recommended by Google as the solution for making cheese stick better to pizza.
Once people really see that it's a simulation, albeit a very powerful and helpful one, responsible use of the technology will be far less of a problem.
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
Tools like Google NotebookLM and custom generative AI services are fundamentally changing how users interact with information. We're seeing a transition from static reports and interfaces to dynamic chat-based tools that give users exactly what they need, and even things they didn't know they needed.
If you're not familiar with NotebookLM, it's a tool that allows you to provide your own documents (like PDF, text files, audio), and then chat with the data. You can even listen to an AI-generated podcast that explains all the information. For example, I had loaded a project with PDF documents containing the rule book, technical rules, and officials briefing information for USA Swimming, and was then able to get answers to questions like "how is a breaststroke turn judged?"
It was kinda magical.
We've been working with clients on permutations of this scenario for some time. For example, we partnered with a client in the life sciences space to build a chat-based tool that connects various third party API services with disparate information, providing account managers with a single source for helping their customers recommend products and services to ensure better health outcomes.
This is no small feat when the goal is a best-of-breed user experience (UX) like ChatGPT. It can involve multiple service providers like Microsoft Azure and Amazon Web Services, as well as various tools like cloud-based large language models (LLM), vector search, speech services, cloud storage, charting tools, location services, AI telemetry, and more. But when it's done right, the result is amazing. You can ask questions that span disciplines and contexts and see results you may not have ever seen before.
Most organizations can really benefit from exploring how generative AI can positively impact their offerings and give them a competitive advantage. Like we always say, it's not about the organizations that use AI, it's about the ones that don't.
There's usually more to the story so if you have questions or comments about this post let us know!
Do you need a new software development partner for an upcoming project? We would love to work with you! From websites and mobile apps to cloud services and custom software, we can help!
