Heardlist: trace a task back to the words that created it

Heardlist is a self-initiated concept: a landing page for an imagined tool that turns a client call into assigned tasks, where every task links back to the moment it was said. I designed the product idea, the identity and the page, and built the page in Framer.

Concept project. Heardlist is not a live product; the Pellmore Studio call, its date and the people in it are fictional.

A meeting tool can hand you a list of action items, but a list on its own asks you to trust it. I set myself one question: could a task be checked at a glance, showing who said it, what exactly they said, and when?

Priya Raman says, "I'll send the revised scope by Thursday." That line becomes a task: Send revised scope, owned by Priya Raman, due Thu 10 Sep, with its source at 00:14:32.

The page explains how each part is known, in plain words. The task wording is taken from the line itself. The owner is the speaker who made the commitment, from "I'll". The due date comes from "by Thursday" read against the call date, Wednesday 9 September. The source timestamp never moves or counts: it is proof, not playback.

Colour carries meaning. Copper marks the spoken source and verdigris marks the committed task, on a warm limestone ground. One copper-to-verdigris line runs from the moment to the task, so the layout itself shows the transformation.

Two voices in the type. Schibsted Grotesk speaks for the product; Alegreya Sans, set upright, is kept for the words people actually said.

The link really works. Selecting 00:14:32 moves focus back to the spoken line, marks it as the returned source and announces the return for screen readers. It works from the keyboard and does not depend on colour, hover or motion.

The interactive landing page is built in Framer, with the source-to-task record as a custom code component. A separate case study on my studio site walks through the process. The short film at the top was made for this post: the sculpted relief behind the record is a visual motif and the pointer is a demonstration, not product features.

A complete concept: product idea, identity, a working landing page and a short film, showing one way provenance can be designed into an AI tool's output. There are no users or business results to report; the page and its sample call are demonstrations.