There is a familiar stall point in deal work. You know what you need, but it does not fit any standard model or report. You end up hacking together a one off in Excel or writing a long email to an analyst that still misses. This task cuts that friction. You state the objective, list the inputs, and show what a good output looks like. Vic returns a deliverable that follows your spec, and you can reuse the same recipe next time. Research ~10 min to run Build a Custom Task Vic prompt Use Vic to build a custom task for analyzing lease rollover risk on a 200,000 square foot office building, with output as a table of expirations by year and tenant. Purpose Delivers results in about 10 minutes instead of the 60 minutes a manual build typically requires. Inputs Objective Optional Inputs Optional Output Example Optional Outputs A bespoke deliverable that follows your stated objective and output requirements exactly. Time saved Turns roughly an hour of manual work into about ten minutes. How it works You define three things. First, the objective. Be specific about the decision or question. Second, the inputs. These can be files, pasted data, or a note on what you will provide each time. Third, the output example. This is where most of the value sits. If you want a table, name the columns and their order. If you want a memo, spell out sections and tone. Vic builds to that shape. The run line is simple: "Use Vic to build a custom task for analyzing lease rollover risk on a 200,000 square foot office building, with output as a table of expirations by year and tenant." Swap in your own objective and output. The clearer your spec, the closer the first pass will be to what you would have built by hand. What comes back is a custom deliverable that follows your requirements. If you asked for a table of expirations by year and tenant, you get that table. If you asked for a short memo with defined sections, you get that memo. There is no translation from a generic output to your house format because you set the format up front. The second payoff is reuse. Save the task once it meets your standard. On the next deal, run the same task with new inputs and get the same structure back. That consistency is hard to keep when each analysis starts from a blank sheet or a copied model that has drifted. There is a practical discipline here. If your output example is vague, the result will be vague. If your columns are explicit, the headers will match. If you care about grouping or ordering, say so. Treat the output example like a template you would hand to a junior analyst and expect them to follow. This works best when the question is common for you but not common enough to justify a product task. Portfolio specific rollups, custom lender packages, internal IC summaries with a fixed format, or niche asset types all fit. You avoid bending a standard workflow into something it was not built to do. Time is the obvious gain. The task runs in about 10 minutes versus about an hour to build and clean up a manual version. The quieter gain is fewer back and forth cycles. When the output is defined at the start, there is less rework to get it into the shape your team expects. One dry aside. This will not fix a fuzzy question. If you are unsure what you want, you will get something that mirrors that uncertainty. Spend a minute tightening the objective and the output example. The rest follows.