by Gary Mintchell | Oct 17, 2025 | Generative AI, News
Praveen Rao, Head of Manufacturing for Google Cloud, spoke at this week’s Connected Worker Conference about a survey the company conducted by National Research Corp. of manufacturing executives regarding their investments in AI. Agents were of specific interest.
We’ll skip the suspense. 78% of the executives surveyed believe they are seeing a return on their investment.
I asked Rao of agents were real or still in the vaporware phase. He assured me that there are many agentic AI use cases. A refresher—agents work with LLMs to “see, hear, think—they can design and simulate.”
There are supply chain use cases, but Rao assured me there are many other use cases. He pointed to engineering and product design workflow improvements when he pointed to 75% of surveyed executives see agents used to improve productivity. Other key areas of interest include improving customer experience (support, etc. 64%), business growth (60%), marketing (creating brochures and the like, 58%), and security (53%).
I’m interested in how these technologies can work with unstructured data. Eventually that will be a massive win.
Back to the survey from the press release:
According to the survey of more than 500 manufacturing execs, 56% reported their organizations are actively using AI agents, with 37% reporting they have launched ten or more. These agents range from gen AI-powered chatbots and single-task agents for specific functions like scheduling production jobs — all the way to sophisticated, multi-agent systems that can take actions on behalf of users, under their supervision.
Additional findings include:
AI agents are being used for core business processes such including:
54% use them for quality control
48% use them for production planning
47% use them for supply chain and logistics
Over half of manufacturing executives (55%) stated their organizations plan to allocate 50% or more of their future AI budget to AI agents,
The primary consideration for executives when evaluating LLM providers is data privacy and security (37%), followed by system integration and scalability and performance.
Google Cloud marketing includes this interesting caveat: In other words, AI success hinges on deep cross-functional collaboration, but it’s top-level support that will truly drive results, aligning AI adoption with business goals and guiding crucial decisions about its fundamental use within organizations.
Yes, technology enables and provides tools, but in the end it’s people and systems and organization that creates a win.
by Gary Mintchell | Oct 15, 2025 | Commentary, Generative AI, Podcast
I’ve published a podcast both on my podcast app (available in Apple, Overcast, or wherever you download them) and on YouTube. You can subscribe on any. Or click the links for podcast or YouTube on the right sidebar.
Why pursue AI? As a tool to help entrepreneurs add value to their companies. The appropriate roll out entails organizing small “pirate ships” empowered to experiment and implement with a budget and air cover. Many concerns about AI’s impact on employment and organization are over blown. History shows that new technology winds up creating more jobs than it destroys. This podcast is sponsored by Inductive Automation.
by Gary Mintchell | Oct 14, 2025 | Generative AI, Manufacturing IT, Operations Management, Software
I am passing this on from the Peter Diamandis newsletter. I don’t think I can link, but click the link on his name to go to his website and sign up. Diamandis sometimes climbs over-the-top optimistic. But that’s a great counter to the usual cynicism and negativity and dysfunctional thinking prevalent in today’s society.
Understanding artificial intelligence (called by Om Malik “augmented intelligence” and by others as neither artificial or intelligent) today requires a healthy dose of realistic thinking and perspective. I offer these thoughts as a counter to your usual AI hype.
Traditional companies are failing to implement AI effectively. Here are five principles to make the technology actually work for you…
1/ AI problems are rarely AI problems – they’re strategy problems disguised as technology problems. Most organizations fail at AI implementation not because they chose the wrong models or hired the wrong engineers, but because they never clearly defined what business problem they’re solving. They see competitors “using AI” and panic-buy solutions for problems they can’t articulate.
2/ Budget size is inversely correlated with AI success. The companies throwing millions at AI initiatives are systematically outperformed by teams running on shoestring budgets with clear mandates.
3/ The 10x rule is the only rule that matters for AI adoption. Anything less than a 10x improvement in speed, cost, or quality is organizational noise. Most AI projects deliver 20-30% improvements that get lost in measurement error and change management overhead.
4/ Competitive intelligence is your fastest path to AI advantage. While you’re debating whether to build or buy, your smartest competitors are already shipping AI-powered solutions.
5/ Pirates beat committees every time. The worst way to implement AI is through enterprise-wide initiatives with steering committees and governance frameworks. Instead, empower your teams from the ground up. Recent studies indicate some alarming news:
- 42% of executives say the process of adopting generative AI is tearing their company apart
- 41% of Millennial and Gen Z employees admit they’re sabotaging their company’s AI strategy
- What’s needed is to enable small teams, “pirate ships,” to move at startup speed (within enterprise contexts). Small teams are optimized to experiment and learn rather than aim for consensus. Give them a problem, a budget, and air cover, then get out of their way.
Here’s the key implementation insight: AI amplifies existing organizational capabilities (and dysfunctions).
by Gary Mintchell | Oct 10, 2025 | Generative AI, Robots, Technology
I’ve been trying to guide AI discussion toward useful applications rather than overly hyped general visions. Let the people dealing in billions of dollars promote themselves. For those who have real work to do, look into the details of AI news to discern real benefits. Perhaps this news from ABB fits that model. This news also follows the trend of larger companies investing in specialist companies in order to drive additional benefits for their products and solutions.
In this case, ABB (robotics) has invested in LandingAI in order to improve the company’s robotic applications for customers. Oh, and we get a new TLA (three-letter acronym). Note the specific examples.
In brief:
- Strategic investment secures ABB’s use of LandingAI’s vision AI capabilities, such as LandingLens, for robot AI vision applications
- Pre-trained models, smart data workflows and no-code tools reduce training time by 80% and accelerate deployment in fast-moving sectors including logistics, healthcare, and food & beverage
- First of its kind collaboration marks a major step towards ABB Robotics’ vision for Autonomous Versatile Robotics – AVR
This first of its kind collaboration will integrate LandingAI’s vision AI capabilities, like LandingLens, into ABB Robotics’ own software suite, marking another milestone in ABB’s journey towards truly autonomous and versatile robots.
“This announcement is the latest in our decade-long journey to innovate and commercialize AI, benefitting our customers by enhancing robot versatility and autonomy to expand the use of robots beyond traditional manufacturing,” said Sami Atiya, President of ABB Robotics & Discrete Automation. “The demand for AI in robotics is driven by the need for greater flexibility, faster commissioning cycles and a shortage of the specialist skills needed to program and operate robots. Our collaboration with LandingAI will mean installation and deployment time is done in hours instead of weeks, allowing more businesses to automate smarter, faster and more efficiently.”
As part of the collaboration ABB has made a venture capital investment through ABB Robotics Ventures, the strategic venture capital unit of ABB Robotics, driving collaboration and investment in innovative early-stage companies that are shaping the future of robotics and automation. Financial details of the investment were not disclosed.
LandingAI’s LandingLens is a vision AI platform that enables the rapid training of vision AI systems to recognize and respond to objects, patterns or defects with no complex programming or AI expertise required.
Through this collaboration, ABB Robotics will reduce robot vision AI training & deployment time by up to 80 percent. Once deployed, system integrators and end users can retrain the AI for new scenarios on their own, unlocking a new level of versatility. This is a critical step in scaling robot adoption in dynamic environments, beyond traditional manufacturing, especially in fast-moving sectors such as logistics, healthcare and food and beverage. ABB is already piloting LandingAI’s technology and actively working to integrate it into existing vision AI applications, including item-picking, sorting, depalletizing and quality inspection.
More information about RobotStudio with generative AI assistant:
- RobotStudio Al Assistant provides real-time, step-by-step guidance for robot programming
- More intelligent and easy-to-use generative Al interface creates faster, easier commissioning and boosts productivity
- Another step in enhancing robot accessibility and versatility beyond traditional manufacturing
Powered by a Large Language Model (LLM) that understands and interprets human language, RobotStudio AI Assistant draws from ABB’s comprehensive library of manuals and documentation to deliver high-quality, context-rich responses to questions, enabling users to set up faster and find rapid answers to questions and technical challenges.
Click on the Follow button at the bottom of the page to subscribe to a weekly email update of posts. Click on the mail icon to subscribe to additional email thoughts.
by Gary Mintchell | Oct 8, 2025 | Generative AI, News, Technology
I wrote recently about my mixed feelings about AI-generated conversations for such situations as customer support and service. This news brings artificial human voice conversations to the receptionist. This is probably better than the “press 1 if…” interfaces we’ve experienced for the past 40 years. It’s remains impersonal, but a touch more friendly. This is another news item from Deepgram, about whom I’ve written many times in 2025.
Deepgram announced that Abby Connect, a premier virtual receptionist service, has successfully launched its new AI Receptionist product line built on Deepgram’s real-time speech-to-text technology.
Note the wording of “human first impression.”
For more than 20 years, Abby Connect has built its reputation on creating a warm, human first impression for every call. But scaling that personal service 24/7 – while managing rising client demand and costs – presented a major challenge. Abby Connect turned to Deepgram to help strike the right balance between efficiency and empathy.
After evaluating Google Cloud Speech-to-Text, AWS Transcribe, AssemblyAI, and Whisper, Abby Connect found Deepgram’s performance to be unmatched:
- Accuracy in the Real World – Deepgram outperformed competitors on noisy calls, including from HVAC job sites.
- Low Latency for Natural Conversations – Sub-300ms streaming latency enabled real-time, two-way AI dialogue without delays.
- Ease of Integration – Developer-friendly APIs and transparent pricing simplified rollout.
- Domain Customization – Tuned for industry-specific terminology, from legal to medical.
Abby Connect is now exploring how to extend Deepgram-powered transcription into even more advanced conversational AI, including large language models trained on call data to detect intent, measure sentiment, and enable smarter escalations.
Click on the Follow button at the bottom of the page to subscribe to a weekly email update of posts. Click on the mail icon to subscribe to additional email thoughts.
by Gary Mintchell | Oct 3, 2025 | Automation, Commentary, Generative AI
Q: Why are we building out this new technology?
A: Because we can.
Better Q: What good will this new technology bring to society?
Voice AI and Voice assistant (also assistance) press releases keep coming my way. They extol how realistic the conversations with (nonhuman) voice AI are becoming. Companies will be able to use these more extensively for customer support.
I understand how challenging finding employees for this sort of work has become. Especially that they would like to be paid. The operations expenditure for a voice assistant is quite low after a modest capital expenditure. Makes it attractive to MBAs.
Just today I was trying to find the source of a problem between my WordPress site (with WP Engine) and Cloudflare. Three weeks ago a DNS update at WPEngine (I think) shut down my site. I searched around looking for an answer. Updated some DNS settings on Cloudflare, and my site was back up.
But there were still little problems. Now my site is showing no statistics. Went to WPEngine. It had forgotten who I was. (Another story). Went to Cloudflare. Checked DNS settings versus what I saw on WPEngine. Updated for the second time in a month.
But support? All I got were generic tree-structure questions and perhaps an AI chat bot. No real help.
I appreciate the technology behind Deepgram, about whom I’ve written five times this year (here, here, here, here, and here). But as a consumer, I’m not all that sanguine about the use of the technology. Just because we can do it, will that make customer service and support more human and helpful?
