Video
Explore, Experiment, Automate: A T-W-O Project Bootcamp
Posted: | Wei Chen | 1:44
See how a real business problem becomes an AI workflow through exploration, step-by-step testing, and automation with human review. This short explainer follows a daily portfolio-brief example.
Key takeaways
- Start with a real problem and explore where AI helps—and where it fails.
- Break the workflow into steps, check each answer, and fix weak steps before connecting them.
- Chain tested skills into an automation with a trigger, a useful output, and a human review checkpoint.
- Use the example to identify which of your team's own workflows is worth testing next.
Start with work, not a tool
A useful AI experiment starts with a problem someone owns. In this short bootcamp explainer, the problem is a retail investor's morning: the portfolio moved overnight, but news, sentiment, and earnings still need to be turned into an understandable brief.
The example is a learning exercise in workflow design, not investment advice. The important questions are what AI can handle, where its answers need checking, and who reviews the result.
Explore what AI can really do
Before designing an automation, try the individual tasks. Ask AI to summarize a headline, read an earnings call, or assess market sentiment. Compare useful answers with weak ones. Exploration reveals both possibilities and limits; a confident answer is not the same as a reliable one.
Experiment one step at a time
Break the morning brief into a sequence: get the portfolio, check relevant news, assess sentiment, review fundamentals, and draft a recommendation for human review. Test each step separately. When a step stumbles, fix it and run it again before relying on its output elsewhere.
This makes the workflow inspectable. The team can see where an answer came from and decide what evidence or judgment is still needed.
Automate with ownership intact
Once the steps work, package them as reusable skills and connect their inputs and outputs. Add a trigger, such as a daily morning run, and deliver the result in a useful format, such as a PDF brief or slide deck.
Keep a human review checkpoint before anyone acts on the recommendation. Automation changes the sequence of work; it does not remove responsibility for the decision.
Bring your team's problem
The portfolio brief is one example. Your starting point might be a recurring report, a handoff, or a daily review. Choose a workflow with a clear owner and a result the team can evaluate.
Explore bootcamp formats or discuss a bootcamp for your team.
Transcript
[00:00]In a T-W-O Project bootcamp, your team doesn't just hear about AI.
[00:04]You build with it in three steps.
[00:06]Explore. Experiment. Automate.
[00:10]Every bootcamp starts with a real problem.
[00:13]Here's one. A retail investor wakes up to a portfolio that moved overnight.
[00:18]News, sentiment, earnings.
[00:19]And no idea what any of it means for them today.
[00:23]Step 1. Explore.
[00:25]Before any plan, the team simply plays.
[00:27]Ask AI to read a headline.
[00:29]Summarize an earnings call.
[00:31]Judge the mood of the market.
[00:32]Even ask it to predict tomorrow's price.
[00:35]Some of it works brilliantly.
[00:36]Some of it doesn't.
[00:37]And that's the point.
[00:39]You discover what AI can really do.
[00:42]Step 2. Experiment.
[00:43]Now we take the investor's morning and break it down.
[00:46]Get the portfolio.
[00:48]Check the news for each holding.
[00:49]Read the sentiment.
[00:51]Look at the fundamentals.
[00:52]Recommend buy, hold, or sell.
[00:54]Then we walk through it, one step at a time.
[00:57]AI does the work, and the team checks every answer.
[01:00]When a step stumbles, we fix it and run it again.
[01:04]Step 3. Automate.
[01:05]Every step that works becomes a skill.
[01:07]And skills can be chained, so the output of one becomes the input of the next.
[01:12]Add a trigger.
[01:13]6 a.m. every morning.
[01:15]Add a checkpoint, so a person still reviews the call.
[01:18]And deliver it where it's needed.
[01:19]A brief, as a PDF or a slide deck, ready before the day begins.
[01:24]Explore what AI can do.
[01:27]Experiment step by step.
[01:28]Automate by chaining skills together.
[01:31]Your team leaves with a working automation.
[01:33]And an eye for which of their own tasks could be next.
[01:36]This was one problem.
[01:37]Bring yours.
[01:39]Book a bootcamp with the T-W-O Project.
Related resources
Video
Why we built the T-W-O Project
AI makes tasks faster, but experts still review in hours what it generates in seconds. This short talk explains why AI pilots stall and why the T-W-O Project starts with workflows and owners.