A real piece of work is never one conversation. A product launch is the kickoff, plus the weekly syncs, plus the vendor calls, plus the hallway decisions, plus the review at the end — a dozen conversations spread across two months that all belong to one thing. The trouble is that they don't arrive grouped. They land in your library one at a time, interleaved with every other conversation you had that week, and the work of seeing them as a single workstream falls on you. Projects does that grouping, and then reasons within the group — so a workstream becomes a thing you can actually hold, without filing anything by hand.
Around the fire, the many gatherings about one long undertaking — a season's hunt, a harvest, a journey — belonged together in the telling, even though they happened on different nights. Projects is Bonfiyah keeping the gatherings that are about one thing together, so the long arc of a piece of work stays legible across all the conversations that made it.
Grouping that doesn't cost you anything.
Projects lets the recordings that belong to one workstream live together as a group rather than scattered through your library. A Project is the container for “all the conversations about this thing” — the launch, the deal, the renovation, the case — so you can see the whole arc in one place instead of hunting for the pieces.
The point is that organizing it shouldn't be a tax. The whole reason conversations stay disorganized is that filing them is tedious, so nobody does it consistently, so the structure never quite exists. Projects is built so the grouping happens without you maintaining a filing system — the related conversations come together into a workstream without you treating it as a chore on the side. Organization you don't have to perform is organization that actually holds.
Project Context: reasoning within a workstream.
Grouping conversations is useful on its own, but the real value is what becomes possible once they're grouped. The AI layer can reason within a single project — across just the conversations that belong to it — instead of only within one recording or across your entire undisciplined library.
That's a meaningfully different question to ask. “What's the state of this project?” draws on every conversation in the workstream and nothing outside it: the decisions made across all the syncs, the commitments still open within this effort, how the thinking evolved from kickoff to now. Project Context is the AI reading the whole arc of one piece of work as a connected story, which is exactly the view you want when someone asks “where are we on the launch?” and the honest answer lives spread across eight separate conversations.
It's the same cross-conversation reasoning that makes Bonfiyah different, scoped — pointed at one workstream so the answer is about that work and not diluted by everything else you talked about. And project actions carry that into the concrete: the open items and next steps that belong to the project, gathered within its boundary rather than lost in a library-wide pile.
Why scoping the reasoning matters.
It's worth being clear about why a scoped view beats a global one for this job. Your whole library is a lot of unrelated conversations — work, personal, a dozen different efforts all mixed together. Ask a question across all of it and the signal for any one project gets buried in noise from everything else. Scope the same question to one project and suddenly the answer is sharp: only the relevant conversations contribute, so “what's still open here?” returns the open threads of this work, not a muddle drawn from your entire month.
That's the structural reason Projects matters beyond tidiness. It's not just that grouped conversations are easier to find — it's that grouping them gives the AI a clean boundary to reason inside, and a clean boundary is what makes the reasoning trustworthy.
Where it earns its place.
Running a project across weeks. Anything that spans more than a couple of conversations — a launch, an implementation, a campaign — benefits from a single place that holds the whole arc and can tell you where it stands without you reassembling it from memory.
Client and matter work. Professionals who organize by client or by matter — every conversation about one account, one case, one engagement — get a workstream view that keeps a relationship's whole history legible and reasoned-over as one thing.
Handing a workstream to someone. When a project changes hands, “here's the project” is far more useful than “here are nineteen recordings, good luck.” The grouped, reasoned-over workstream is something a new person can actually pick up.
What it isn't.
Projects isn't a project-management tool, and it isn't competing with the software your team plans and tracks work in. There are no Gantt charts, no sprint boards, no assignment-and-deadline machinery — it won't run the project. What it does is narrower and complementary: it groups the conversations that belong to a piece of work and lets the AI reason across that group. It's about your recordings of a workstream, not the workstream's tasks and timelines. It also only ever organizes what you've recorded, behind the same consent tooling that ships in every tier. Think of it as the memory layer for a project's conversations — sitting alongside whatever you plan the actual work in, not trying to replace it.
It keeps the conversations about one thing together, and reasons within them. That's the scope, and it's a genuinely useful one.
Try it on a real piece of work.
The way to feel Projects is to take something you're actually working on that spans several conversations, group those recordings into a Project, and let the Pro AI layer run. Then ask it where the work stands. The answer — drawn from the whole workstream, scoped to just that effort — is the thing you'd otherwise be reconstructing by hand from a scattered handful of recordings. Pro AI is free to try for seven days, which is enough time to point it at one real project and see the difference scoping makes.
The many nights that told one long story always belonged together. Projects keeps the conversations about one piece of work together too — so the whole arc stays legible, without you filing a thing.
— Richard
