//Door three: bring Matt in to teach
Prompting Training for Investigators and Legal Teams
Matt Aubin teaches people to work with AI the way a professional works with any tool they are accountable for. He teaches a proven pattern that takes a person from zero to expert in prompting, on any large language model, at any level, and the room leaves able to get real output instead of guessing.
By Matt Aubin, Founder, Southern Recon Agency and E3 Legacy Intel. CDFE, FBCI. Investigating since 2009. Updated .
Matt Aubin
Prompting isn't everything, it's the only thing.
Every other AI decision a firm makes sits downstream of whether its people can actually operate the thing. The tools are already in the building. The skill is not.
What does prompting training actually cover?
It covers the questions people actually ask in the room: how to interface and operate with AI, how to get the most out of whichever models, tools and systems your people already have in front of them, how to do it safely, what is legal and what is not, how to protect intellectual property, and how to own intellectual property when AI helped make it. Underneath all of it sits a proven pattern that takes a person from zero to expert in prompting. The pattern is taught in the room. It is not published anywhere, including here, because it is the tradecraft people are paying to learn.
The ownership questions get their own page, written from the primary sources and kept at a teaching level: who owns work that AI helped make. Matt teaches the tricks of the trade. He does not give legal advice.
Who is this for?
Investigation and security firms, legal teams, associations and government cohorts, and it works for a room of mixed ability rather than a room of specialists. The people who get the most out of it are the ones who have to sign their name to the result: the owner, the case manager, the analyst, the lawyer, the supervisor who will be asked how the work was done. No coding background is needed and none is assumed.
How is this different from a free course from Microsoft or Anthropic?
A free self paced course from a model maker is a good way to learn what the buttons do, and anyone who has time for one should take it. This is a different thing. It is live, it is run on the kind of work the room actually handles, and it is taught by a working investigator who is personally accountable for work that AI touched. The questions it is built to answer are the ones that come up after the buttons make sense: how far to trust what came back, what is safe to put in, and how to say afterwards how the work was done.
Which AI models does this work on?
Any large language model. Nothing taught is tied to one company's product, one subscription or one interface, because firms rarely run on a single tool and the tools change every few months anyway. People arrive using different models, different assistants and different in house systems, and they all need to get more out of what they already have. The teaching is built so that when the model changes underneath them, the skill still holds.
How do you teach people to use AI safely?
By treating it as a handling problem rather than a technology problem. Material put into a public tool can leave the firm's control, and federal law at 18 U.S.C. 1839(3) protects information as a trade secret only where the owner has taken reasonable measures to keep it secret. The federal standards body makes the same point in its generative AI profile, NIST AI 600-1, published July 2024, which names intellectual property among the risks specific to generative AI and expressly covers eased exposure of trade secrets. The session therefore covers what those sources say, what a sanctioned place to do the work looks like, and how a person checks output before anyone relies on it. This is teaching, not legal advice, and a lawyer should be asked about any specific matter.
What do you teach about owning the work AI helped make?
At a teaching level, from the primary sources, and only far enough that people stop guessing. The United States Copyright Office said in its registration guidance at 88 Fed. Reg. 16190, published March 16, 2023, that based on the functioning of current generally available technology, when an AI technology determines the expressive elements of its output, the generated material is not the product of human authorship. The District of Columbia Circuit held in Thaler v. Perlmutter on March 18, 2025 that a human being has to be the author, and the Supreme Court denied review on March 2, 2026, which left that decision standing without deciding anything itself. What a person still owns is their own writing, their creative selection and arrangement, and their meaningful edits. Matt teaches what those sources say. He does not give legal advice, and a lawyer should be asked about any specific matter.
The full set of sources, dated and named, is on who owns work that AI helped make.
Tell me the room, the date and the format.
Keynote, half day, full day or a cohort across several sessions. You get a tailored outline and availability back, scoped in writing.
Does it count for CLE or CPE credit?
Training can be structured for continuing education credit where the awarding body allows it. No accreditation is claimed today and none is in place today. Tell me which body you answer to and what its rules require, and the session is built to fit those requirements, including the session length, the learning objectives and the record of attendance the body asks for. If the body will not approve it, you will be told that rather than sold around it.
What about the people who build with AI rather than just use it?
That is a separate track and a separate audience. Some people in a firm are not trying to write a better request to a chatbot, they are trying to build AI programs and tools, or to use AI to help build something, and they want agentic AI and AI agents doing real work in their favor. Those people need to know how to direct an agent, where to put a human approval, and how to check what the agent actually did on a live client file. That work has its own page rather than being crammed into a prompting day.
The builder track is set out on AI agents for investigation and security firms.
What changes after the session?
People stop typing a question and hoping. They get usable output on the first or second attempt instead of the tenth, on whichever model is in front of them, and they can say out loud why the output is or is not good enough to use. They know what they are not allowed to paste into a public tool and why. They know which parts of an AI assisted work are theirs and which are not. And the firm has one shared way of working rather than eleven private habits nobody can audit.
How do you know it worked?
Because the work is done in the room on the kind of material the audience actually handles, and everybody can see what came back. The measure is not a certificate or a quiz score, it is whether a person who arrived guessing leaves able to run a piece of their own work end to end and explain the result to a supervisor. Where the room wants it, the session ends with each person holding output they produced themselves rather than notes about output somebody else produced.
//Proof
Rooms he has taught, with names and dates you can check.
He taught at World Orlando, the State Department sponsored conference where he trained a delegation of Brazilian prosecutors in Orlando. On the training circuit, he delivered "A-I for the Modern Private Investigator" at Intellenet's 2026 conference on April 15, 2026. In 2019 he keynoted the feature stage at The Prism Group's White Label World Expo in Las Vegas, an event with more than 7,500 in attendance.
The reason those rooms work is the same reason this page exists. Investigating since 2009. Founded his first firm in 2010 and built it into Southern Recon Agency in 2014. He built a proprietary AI investigation platform his own firms run on live cases, so the teaching comes from work that had to hold up, not from a curriculum written about work somebody else does.
The two training tracks and the executive briefing are set out on the AI training page, the keynote formats and the booking detail are on the speaking page, and the deeper session for investigators is AI training for private investigators.
Tell me the room, the date and what your people need to be able to do afterwards.
Investigators, a legal team, an association audience or a government cohort. You get a tailored outline, the format that fits, and availability.