How I Actually Use AI: From Friction to Workflows
The more I listen to people talk about AI agents, context files, skills, and automated workflows, the more I realize that what many are just starting to name is what I have been building and teaching for a while.
I use AI every day across ChatGPT, Claude, Google AI Studio, OpenClaw, Notion, Slack, and other tools, but I do not start with the tool. I also do not start with a prompt.
I start with friction.
That is the first thing I teach because that is where the useful work begins. I look at the places where work keeps getting stuck, repeated, delayed, miscommunicated, or rebuilt from scratch. I look for the annoying manual task. The repeated decision. The messy handoff. The document pile. The follow-up that should have happened faster. The planning change that creates ten new problems. The meeting notes that never become action. The good idea that stays trapped in someone’s head because nobody has time to turn it into a process.
That is where AI becomes useful.
Most people are still using AI as a chatbot. They open ChatGPT or Claude, type a prompt, get an answer, edit the answer, and then come back the next day to do the same thing again. That can help, but it is still one-off work. It does not change how the work moves.
I am more interested in turning repeated work into workflows.
Here is how I actually use AI step by step.
Step 1: I start with the friction point
When I build with AI, I ask: where is the drag?
In event production, the drag might be a client change request that affects the schedule, the room flow, the sponsor deliverables, the AV plan, the rehearsal window, and the staffing plan. One change rarely stays in one lane. It moves through the whole show.
In business development, the drag might be what happens after a good conversation at an event. Someone says something valuable, there is a clear opportunity, and then the follow-up gets watered down because the context is lost by the time you sit at your laptop.
In speaking, the drag might be the whole pipeline: tracking booked talks, finding new calls for speakers, matching opportunities to my topics, drafting pitches, keeping the database updated, and following up before the window closes.
In judging or proposal review, the drag might be the first read-through. You have a rubric, submissions, categories, scoring criteria, and a need for consistency, but the process still eats hours.
I start there. I do not ask AI, “What can you do?” I ask myself, “What keeps wasting time or weakening the work?”
That question gives the workflow a purpose.
Step 2: I ask AI to interview me
This is one of the most useful habits I have developed.
Instead of trying to write the perfect prompt, I ask AI to interview me. I want it to pull the knowledge out of me before I ask it to produce anything.
If I am building a workflow for show operations, I ask Claude to interview me like a senior event production lead. I want it to ask what documents are involved, what the event goals are, what the client changed, what cannot move, what risks matter most, what the team needs to see, and what kind of final output would be useful.
If I am building a judging workflow, I ask it to interview me about the rubric, the scoring scale, the category definitions, the difference between a strong submission and a weak one, and what should require human review.
If I am building a speaker pipeline workflow, I ask it to interview me about my speaking topics, my strongest session angles, the audiences I serve, the types of events I want, the ones I do not want, and how I want opportunities ranked.
That interview step matters because a lot of what makes a workflow useful is sitting in your head. AI cannot apply your standards if you have never taken the time to name them.
The interview turns instinct into structure.
Step 3: I build the context before I expect good output
This is where most people rush and then blame the tool.
They upload a few documents, ask for a strong answer, and expect the AI to understand how the work should be done. That is not how this works.
AI does not automatically know that a signed contract carries more weight than an early proposal. It does not know that the final agenda beats an old planning doc. It does not know that venue rules beat a creative idea. It does not know that a client-approved description should not be casually rewritten because the tool found a smoother sentence.
So I build context before I expect quality.
This includes primers, source hierarchy, examples, output rules, voice rules, standards, and red flags. I teach the tool what role it is playing, what it should protect, what it should question, what it should flag, and where it should stop.
For me, the primer is not the whole story, but it is the foundation. It gives the workflow a brain before I ask it to produce work.
Without that foundation, AI gives you answers that sound fine but do not understand the stakes.
And in event production, “sounds fine” is not good enough. A clean summary will not save you when the rehearsal window is gone, the sponsor promise is impossible, the room turn is too tight, and nobody noticed until show day.
Step 4: I turn repeated tasks into skills
Once the context is built, I look for the work that repeats.
That is where skills come in.
A skill is the reusable layer that lets AI perform a specific type of work without me rebuilding the instructions every time. It can carry the rules, examples, formats, and decision boundaries for that task.
One of my examples is a Show Ops workflow. I feed Claude the current schedule, venue logistics, client change request, and other planning documents, and the workflow helps rebuild the run of show, flag cascade risks, preserve key time blocks, and draft team communication.
That saves time, but the more important part is that it catches what gets missed when everyone is moving fast.
A room change might affect AV.
A sponsor promise might affect timing.
A meal window might affect the whole afternoon.
A speaker delay might affect graphics, walk-on music, camera cues, mic assignments, and rehearsal.
AI can help scan across those dependencies and give the production team a clearer starting point.
Another example is my AI-assisted judging workflow. I built it to review submissions against a scoring rubric, apply the criteria consistently, explain the reasoning, and support the first layer of evaluation. It does not replace the final decision. It gives the reviewer a structured read before the human judgment comes in.
I also built a speaker pipeline agent that reads my speaking database, understands my topics, identifies new calls for speakers, drafts pitches based on past sessions, and sends updates through Slack or Discord when something is worth reviewing. That workflow supports revenue visibility, not just task completion.
I have used this same thinking for lead follow-up after events. Before the event, I can define the ideal buyer, the offer, the conversation goals, and the follow-up criteria. After a meaningful conversation, I can record a quick voice note with who I met, what they cared about, what opportunity came up, and what should happen next. AI can clean it up, organize the context, prioritize the lead, and draft the next step while the conversation is still fresh.
That is how I use AI. I look for repeatable work and build support around it.
Step 5: I connect the right source material
A workflow is only as strong as the material it can work from.
For event work, that might include the BEO, contract, run of show, production schedule, venue logistics, sponsor deliverables, AV quote, floor plan, client notes, and change requests.
For speaking, it might include booked talks, past session descriptions, audience types, speaker bios, previous pitches, event themes, and calls for speakers.
For judging, it might include the submission form, scoring rubric, category definitions, eligibility rules, and sample submissions.
For content, it might include podcast transcripts, LinkedIn posts, newsletters, language rules, audience themes, and performance data.
This is where I see a lot of people get sloppy. They think AI is weak because the output is weak, but the input material is scattered, outdated, or missing the actual decision rules.
AI needs the right material, and it needs to know how to treat that material. Otherwise, it will pull from everything as if everything has the same weight.
That is how you get confident nonsense.
Step 6: I define the output before I ask for the work
I do not want a long summary unless a summary is the actual deliverable.
Most of the time, I want something I can use.
If I am reviewing event risk, I want a table with the risk, source, impact, owner, urgency, and next step.
If I am prepping for a client call, I want the top context, what to ask, what to avoid, what decision needs to be made, and where I need to listen carefully.
If I am reviewing award submissions, I want the score, rationale, strengths, weaknesses, and reviewer notes.
If I am following up after a live event conversation, I want the person’s context, the opportunity, the priority level, the suggested next step, and a draft that does not sound like a lifeless sales email.
The output shape matters because the workflow should lead to action.
A pretty answer that creates more work is not a useful answer.
Step 7: I keep human judgment in the workflow
I do not use AI so I can stop thinking. I use AI so I can think at a higher level and stop wasting brainpower on work that can be structured.
AI can draft, compare, summarize, flag, organize, and prepare. It can surface patterns. It can catch missing details. It can help me see what I might have missed.
But it does not own the decision.
In show operations, AI can flag the risk, but the producer owns the call.
In judging, AI can apply the rubric, but the human owns the final score.
In speaking, AI can draft the pitch, but I decide if the opportunity fits my direction.
In client follow-up, AI can prepare the message, but I approve the tone, the timing, and the relationship move.
That is the part leaders need to understand. AI should expand judgment, not replace it.
Step 8: I turn one workflow into the next one
Once a workflow works, I look for the next pattern.
A Show Ops workflow can become a broader event-risk review process.
A judging workflow can become a proposal review process.
A speaker pipeline can become a larger visibility and business development workflow.
A lead follow-up workflow can become a relationship intelligence process.
A transcript workflow can become content, sales enablement, speaker prep, audience insight, and team training material.
That is when AI stops being a random tool and starts becoming part of how the work gets done. This is the difference between basic prompting and building with AI.
Basic prompting depends on remembering what to ask every time. A workflow carries the thinking forward. It remembers the structure. It knows the sources. It follows the format. It respects the boundaries. It gives you something closer to usable work on the first pass.
That is what I teach inside my AI bootcamps and leadership workshops. I help teams stop chasing prompt lists and start looking at the actual work in front of them.
- Where does work repeat?
- Where does context get lost?
- Where do decisions slow down?
- Where does quality drop?
- Where do handoffs break?
- Where is the team still manually rebuilding something AI could help structure?
That is the real starting point.
I use AI by starting with friction, letting the tool interview me, building the context, creating primers and skills, connecting the right source material, defining the output, keeping human judgment in the loop, and turning repeated work into usable workflows.
That is how I use AI in my own business, in event production, in speaking, in content, in judging, in lead follow-up, and in the bootcamps I teach.
The future of AI at work is not about who has the cleverest prompt.
It belongs to the people who know how their work actually works.
Event Intelligence Built From the Work You Already Do
This is where AI becomes a business advantage for event teams, agencies, associations, and AV production companies.
Most organizations are sitting on years of valuable information: proposals, contracts, RFPs, run-of-show files, production schedules, BEOs, post-event reports, survey data, sponsor deliverables, sales notes, client emails, change requests, vendor quotes, budget revisions, and team debriefs.
The problem is that this information is rarely easy to access. It lives in different folders, platforms, inboxes, spreadsheets, and people’s memory. Leaders know there is insight in the work, but they often cannot access it fast enough to make better decisions.
That is the opportunity.
AI can help teams turn that scattered knowledge into usable business intelligence.
For an event agency, that might mean building a proposal intelligence workflow that reviews past proposals, RFPs, client feedback, scope changes, and won or lost business to identify what is working, what is underpriced, what gets questioned, and where scope tends to expand.
For an AV production company, that might mean building a margin intelligence workflow that compares estimated labor, actual labor, gear packages, travel, change orders, vendor costs, and last-minute requests so leadership can see where profit is being protected and where it is leaking.
For an internal event team, that might mean building a post-event intelligence workflow that pulls together stakeholder feedback, attendee surveys, sponsor fulfillment, session data, production notes, and team debriefs into a clear report that informs the next planning cycle.
For a sales or account team, that might mean building a lead intelligence workflow that captures conversations from events, organizes follow-up priorities, identifies buyer intent, and helps the team respond while the context is still fresh.
For leadership, it means asking better questions of the business:
- Which event formats are worth repeating?
- Which clients require more support than the scope allows?
- Which services should be packaged differently?
- Which project types create the most avoidable pressure?
- Which opportunities deserve more focus?
- Which decisions are still being made from instinct when the evidence is already sitting inside the business?
That is the layer I help teams build.
I work with event teams, agencies, AV production companies, associations, and business leaders to design AI-supported workflows around real business decisions. This is practical work built around the documents, conversations, handoffs, and decision points that already exist inside the organization.
The goal is simple: help teams see their work more clearly, reduce repeated manual effort, protect quality, and make stronger decisions from the information they already have.
This can begin with a keynote, leadership session, AI bootcamp, or private team workshop, but the deeper value is in the workflow strategy itself.
The most common starting points are:
- Proposal intelligence: understanding what wins, what stalls, what expands scope, and what needs clearer positioning.
- Post-event intelligence: turning feedback, performance data, sponsor notes, and production lessons into usable decisions.
- Client intelligence: tracking recurring needs, communication patterns, stakeholder preferences, and account opportunities.
- Margin intelligence: comparing estimated work against actual effort, change requests, labor, travel, and vendor costs.
- Show-risk intelligence: reviewing schedules, BEOs, contracts, venue rules, sponsor commitments, and change requests before issues become expensive.
- Sales follow-up intelligence: turning conference conversations, lead notes, and buyer signals into timely, relevant next steps.
This is the move from using AI for isolated tasks to using AI as a decision layer inside the business. The teams that invest here will have a clearer view of their work, their clients, their margins, their risks, and their opportunities.
- They will be able to make faster decisions because the information is no longer buried.
- They will stop rebuilding the same knowledge from scratch every time a new proposal, event, client request, or leadership question comes up.
That is where I come in.
I help teams map their repeated work, identify the strongest AI workflow opportunities, design the context and source structure, define the outputs, and build practical workflows that support real business decisions.
If your team is ready to move beyond basic AI usage and build workflows around proposal intelligence, post-event reporting, client insight, margin visibility, sales follow-up, or show-risk review, let’s talk.
I am currently booking private AI workflow strategy engagements, leadership sessions, bootcamps, and keynotes for event teams, agencies, AV production companies, associations, and business leaders.
About the Author
Anca Platon Trifan, CMP, WMEP is an AI strategist, keynote speaker, technical producer, and CEO of Tree-Fan Events Productions LLC. With more than 20 years of experience in event technology, AV production, and high-stakes live event execution, she helps organizations rethink how they work under pressure.
Her work sits at the intersection of AI, operational resilience, systems thinking, and human performance. From leadership decision-making and event operations to AI adoption and workflow design, Anca helps teams build systems that reduce cognitive overload, strengthen execution, and improve performance where it matters most. A competitive bodybuilder and triathlete, Anca brings a unique perspective to the conversation by connecting the disciplines of elite performance with the realities of business and event leadership. She is the host of Events: Demystified and the creator of the #FIT4EVENTS™ framework, which explores how leaders can build the mental, physical, emotional, and operational capacity required to perform consistently in demanding environments. Ready to explore how AI, operational strategy, and high-performance leadership can strengthen your team or event? Book a call with Anca here.
