Tradespeople building AI tools

A homeowner used to need a truck roll, a site visit, and often a $3,000 to $5,000 panel-upgrade quote to learn if an electric-vehicle charger would fit. Jason Walls, a master electrician in Kentucky, turned that first step into a $12.99 web check built with Anthropic’s Claude despite having what he says was “zero coding background.” (x.com, youtube.com) The tool is called ChargeRight, and the pitch is simple: upload a photo of the electrical panel, answer a few house-load questions, and get a report based on National Electrical Code Section 220.82. Walls says many charger buyers are being quoted upgrades without anyone running that calculation first. (x.com, youtube.com) Section 220.82 is the “optional method” for dwelling-unit load calculations, which is electrician language for a code-approved shortcut that estimates a home’s real demand instead of adding every appliance at full blast. Electrical Contractor Magazine and Electrician U both describe it as a streamlined alternative to the standard method for one-family homes. (ecmag.com, electricianu.com) That difference can change the answer by dozens of amps. In ChargeRight’s own example, the same 2,400-square-foot house with a 200-amp panel and a 48-amp Tesla charger comes out at 149 amps under Section 220.82 but 204 amps under the standard method, which is the gap between “fits” and “needs an upgrade.” (youtube.com) Walls says that mismatch is common enough that roughly 70% of the homes he saw did not need the upgrade they had been told to buy. He built the software around that narrow job, not as a general chatbot, and said Claude helped him assemble the front end, back end, payment flow, report generator, and deployment over six months. (x.com, forums.hardwarezone.com.sg) That is the part making the story travel beyond electrician circles. Anthropic has spent the past two years pushing Claude from a chat window into a tool for building apps, including hosted “artifacts” that can be shared without the builder standing up their own full software stack. (anthropic.com, anthropic.com) So the new pattern is not “artificial intelligence replaces the electrician.” The new pattern is “the electrician writes down the code rule, the workflow, and the customer pain point, and managed artificial intelligence fills in enough software work to ship a tiny product.” (x.com, anthropic.com) Trades influencers are now pointing to that pattern as something other shop owners can copy. Todd Saunders highlighted the idea on X, and Walls has framed it in blunt terms: trade school knowledge plus a real bottleneck can now become software without a venture-backed engineering team. (x.com, x.com) The products showing up are lightweight on purpose. They are not trying to replace estimating platforms, dispatch systems, or full construction software suites; they are trying to shave one expensive step out of a job, like a pre-quote panel check, a code lookup, or a report generator. (x.com, anthropic.com) That makes the economics different from the last wave of software startups. If a solo electrician can build a niche tool in a kitchen, charge $12.99, and save a homeowner from a $5,000 mistake, the software does not need millions of users to be worth building. (x.com, youtube.com) The catch is that these tools still live or die on the underlying trade knowledge. National Fire Protection Association code is still the authority, and a polished interface does not fix a bad load calculation, which is why the people getting traction here are licensed tradespeople building around code sections they already use on real jobs. (nfpa.org, x.com)