When Joe Manico speaks about inventing and innovation, people listen.
During his career that began with the Eastman Kodak Co. in 1975, Manico began collaborating with other inventors and filing patents on new inventions in the realms of digital image processing, film and digital cameras, innovative digital displays, printers and print finishing systems. He has contributed to 251 granted U.S. patents, with assignees that have included Eastman Kodak, Kodak Alaris, Apple and Carestream Health.
Manico was identified as one of the primary contributors to the Kodak Patent Portfolio, which was sold for $525 million. He later joined Kodak Alaris as a research scientist, where he helps protect corporate research investments with intellectual property. He has been using IP.com solutions for 13 years, making him one of the company’s earliest customers, and utilizes IP.com’s workflow-based AI technology for better results, greater efficiency and reduced risk.
He says that fundamentally, inventing has not changed: It is still predicated on solving a problem that needs a better solution. “The inventor of the plow wasn’t happy using a manual digging stick and the idleness of his donkey,” he said.
We’re listening.
What feels most different about inventing today?
The biggest difference is the speed at which an inventor can move from a rough thought to an informed conversation.
Earlier in my career, the first stretch of invention work was slow, fragmented, isolated. You would rely on memory, a few trusted experts, manual searches and whatever documents you could find.
Today, AI can very quickly surface terminology, adjacent approaches, prior art patterns and technical context that might have taken days or weeks to assemble.
That does not make the idea good by itself, but it changes the starting line. Inventors can begin with a broader view of the landscape, and that forces better questions much earlier.
For example, tools like IP.com’s Semantic Gist® search capabilities enable an engineer to see relevance-ranked prior art from plain language descriptions early in the development process. No need for complex Boolean search strings or waiting weeks for external search results.
When you think back to earlier R&D environments, what did the invention process look like before AI?
In the pre-internet days, if you wanted to read a patent you needed a patent number and a legal assistant would order a printed copy from the USPTO—and you’d have it in a few weeks. A prior art search would be required prior to public disclosure, but that would be very late in the commercialization process, would usually be sent out to individuals without intimate knowledge of the invention, and would take significant time and resources to complete.
The process was much more sequential: Someone had an idea, talked to a few colleagues, sketched it out, maybe searched for some patents or papers, and then at some point the IP team entered the process. A lot depended on who happened to be in the room and what they personally remembered.
Search was manual or sent out to patent searchers, terminology was a real barrier, and visibility was often narrow. Good inventors still did excellent work, but the process had more blind spots. You could spend a lot of energy building confidence in an idea and only later discover that the space was crowded, the vocabulary was different, or a similar solution already existed in another field. The worst-case scenario is that you would learn about relevant prior art after filing from a patent examiner.
Has AI changed the act of inventing itself, or has it changed the workflow around inventing?
Invention is a human mindset. I see AI changing the workflow around inventing more than it changes the core act of inventing. The spark still comes from a human recognizing a problem, caring about the outcome, and making a judgment about what might work.
AI is very powerful around that spark. It can help search, organize, compare, summarize, stress-test and draft. But it does not own the problem. It does not carry the responsibility for the decision.
So, the workflow is being transformed, but the inventor’s curiosity, technical judgment and accountability remain central. AI can enhance efficiency by automating workflows, but a human needs to make the decisions and choose the options.
What has not changed about invention, even with all of today’s AI tools?
Invention still starts with curiosity and dissatisfaction with the way something works today. It still requires persistence, because most ideas are incomplete at the beginning. It still requires technical judgment, because not every plausible answer is a workable answer.
And it still requires documentation. You have to capture what was conceived, what problem it solves, how it works, and why it is different. AI can help with many of those tasks, but it cannot replace the discipline behind them.
AI, in my view, is an eager, world-class, collaborative research assistant. At the end of the day, a human still needs to make the judgement about things like the utility of an invention as usefulness is still a human standard—not a machine or artificial intelligence standard.
What does an AI-assisted invention workflow look like from the moment someone has an idea?
I would start by capturing the idea in plain language: the problem, the proposed solution, the context, why it matters. Not only does this provide important documentation, it allows you to communicate the invention to a broader audience.
I’ve had an advantage using IP.com’s AI-powered platform as an early adopter years ago and became accustomed to describing my ideas in plain language, removing the manual convert concept to Boolean step. Now I can use AI to expand the vocabulary around the idea, identify related technologies, suggest alternate embodiments and use Gist to surface curated prior art or analogous solutions.
From there, the inventor should compare the concept against what is already known, refine the point of novelty and document the most important variations. The best workflow is not “ask AI for an invention”; It is “use AI to interrogate, improve and document a human insight.”
What can AI reveal early that inventors often used to discover too late?
A plain-language Semantic Gist® search can reveal relevance-ranked prior art, and AI can reveal crowded technical areas, different terminology, adjacent solutions and competing approaches very early.
It can also show that the value may not be where the inventor first thought it was. Sometimes the original idea is not new, but a particular implementation, workflow, data structure, or use case is more interesting. That kind of early signal is valuable. It helps inventors avoid wasting time, and it helps them pivot toward the part of the idea that may actually be differentiated.
How does early AI-assisted searching change the decision to keep going, pivot or stop?
When you don’t know the prior art, everything seems patentable. A thorough prior art search makes the decision more evidence based.
Inventors are naturally optimistic, which is a strength, but optimism needs discipline. Early AI-assisted searching gives you a faster read on whether the space is open, crowded, or simply described using different language.
If the results look promising, you keep going with more confidence. If the space is crowded, you pivot toward a narrower or more technically distinct contribution.
If the idea is clearly already known, stopping early is not failure. It is good innovation management and saves on expensive legal and filing fees.
What is the biggest mistake inventors make when using AI during ideation?
The biggest mistake is confusing plausible with true. A general-purpose AI can sound confident even when it is incomplete or wrong. That is dangerous during invention, because a polished explanation can make an idea feel more mature than it really is.
Inventors should use AI to generate hypotheses, not conclusions. Validate the output. Check the sources. Search the patents and literature. Talk to technical people.
The human inventor has to remain skeptical, especially when the answer sounds too easy.
Where is AI genuinely helpful in the invention process—and where should humans remain firmly in control?
AI is genuinely helpful in search, summarization, comparison, classification, terminology expansion, drafting support and organizing invention disclosures. It can help an inventor see the landscape and express the idea more clearly.
Humans should remain firmly in control of inventorship, technical conclusions, filing decisions, legal strategy and final review. Those are judgment zones.
AI can assist the workflow, but it should not be the decision-maker for what was invented, who invented it, whether to file, or what legal position to take. Those decisions should be made by humans.
Looking ahead five years, what will separate successful inventors from those who struggle to adapt?
Five years? It’s hard to imagine because with the radical improvements I’ve seen over the last six months, five years now seems far in the future.
Successful inventors will learn how to combine imagination with evidence. They will use AI early, but they will not outsource their judgment to it. They will search broadly, document carefully, validate aggressively, and stay curious.
The inventors who struggle will either ignore AI entirely or trust it too much. The advantage will belong to people who can ask better questions, evaluate answers critically, and keep the human act of invention at the center of an AI-assisted workflow.