Preparing a defensible quote
A language model can generate a plausible quote quickly. The harder problem was giving it enough context to produce one that reflected how an electrical business actually prices its work.
I built the quoting harness around three sources of evidence: the electrician’s supplier price book, related historical job actuals, and relevant Australian electrical standards. These gave the model a basis for estimating materials, labour and required work rather than relying on the job description alone.
But better inputs did not automatically make the output trustworthy.
The electrician was still responsible for the quote they sent, so I designed each line item to include a justification explaining where the number came from and why it was included. This made the basis for each line visible at the point where the electrician needed to review it.
Choose the right interaction for the task
I initially imagined Cortir as a conversational product, where electricians could make changes through chat or dictation. That worked well for broad intent, but precise quote edits were often faster by hand.
My first attempt expanded a quote line in place. It exposed the controls, but pushed the surrounding quote away and made the interaction feel cramped.
I replaced it with a bottom sheet designed around a shift in focus. The electrician can edit one line in detail and swipe horizontally through adjacent items. When they need the full quote again, they swipe the sheet down.
Conversation remained available throughout Cortir; direct manipulation gave electricians another way to work when precision and speed mattered.
Talk to it like your office manager
The electricians I spoke to saw the value in software, but often found it confusing and difficult to use.
So I gave every interface a dictation option. Instead of learning where everything lived, the electrician could simply say what they wanted to do, much like they would to an office manager.
The metaphor had limits. Cortir could only act within a defined set of tasks, but it gave electricians a familiar way to interact with the product.
The objective changes with the job
The electrician’s objective changes as an opportunity moves towards becoming booked work.
At enquiry, the goal is to understand and scope the job well enough to decide what happens next. In draft, it is to build a quote that is both profitable and competitive. Once sent, the focus shifts to moving the customer towards a decision and turning the opportunity into booked work.
I designed the workbench to change with the job status, bringing forward the information and actions that matter most to the electrician’s current objective.
A quote the customer can understand
Once the quote is sent, Cortir has a second user with a different objective. The customer needs to understand what they are paying for and feel confident enough to make a decision.
I designed the hosted quote around that objective. It leads with the job, total price and a clear path to acceptance.
Cortir translates the technical scope into a plain language description generated from the quote contents. The full line item breakdown, pricing and business details remain available underneath for anyone who wants to inspect the detail.
A clear Accept Quote action gives the customer a direct way to move forward.
Turn silence into signal
Once a quote is sent, electricians have little visibility into how the customer is engaging with it.
I designed Cortir to make some of that activity visible. View count, recency and device activity sit alongside the quote, giving the electrician a clearer picture of how the customer is engaging with it.
The aim was to give the electrician a better basis for deciding who to follow up with and when.

