Penny is software. She follows published rules, using the answers you give, and she is paid the same whether you build or not. This page is the whole method, including what she is not good at.
Most of this market earns when you buy a build. Penny's verdict comes from published rules and is the same whether you build or not.
There is no payment page on this site and we never ask for card details.
Every job goes through the same checks, in the same order. The first one that fails is the verdict. Penny never assumes a job can be fully automated, because exceptions never go to zero.
If yes, Penny stops. That work needs a person with the right qualifications, not a scope of works.
If yes, no software fixes that yet. Verdict: fix the process first, and the scope of works maps the steps, what to fix by hand, and what to automate after. Not sure never ends the check: the scope starts with a process check instead.
Penny knows what Xero, MYOB, HubSpot, Shopify and Microsoft 365 can do. Often the answer is a setting.
For these categories, buy the established tool instead of building one, unless you genuinely need your own workflow.
Then a build costs more than the problem. Verdict: leave it alone for now.
A paid ChatGPT or Claude subscription, with a person checking the output, is usually enough.
Then it is an automation, not an AI build. n8n, Make or Zapier. Cheaper and more reliable.
Then it is worth building. Judgement or messy input in the middle, real value on the other side.
Every fact in a scope of works is labelled. "Not sure" stays unknown. Penny never quietly promotes a guess to a fact, and a blank answer can never produce a verdict.
Vendor documentation and primary sources first. Every claim in a scope of works carries where it came from and when. Where research is thin, it says so.
She checks the job, writes the scope, and never sells you a build.