Built with itself
Palimpse is built inside its own project journal.
The best test of a continuity product is building the product inside it. Everything on this page comes from the real Palimpse project journal: the decisions, the rejected alternatives, the tasks still open, the sessions that moved the work forward.
Nothing here was written for marketing first. It was logged as work, then quoted.
Two real decisions, straight from the journal.
In Palimpse, a decision carries its why and the options it ruled out. These two shaped the home page you may have just come from.
Stop making the demo look like a chatbot conversation.
The earlier landing simulated a long chat exchange. It was rejected because the conversation was not the product: what mattered was what survived after the conversation ended.
Adopt “The project should outlive the conversation” as the real home page.
Three alternatives were recorded and ruled out: keep the old structure and only swap the hero, wait for a formal positioning validation before touching the landing, and keep the simulated demo but simplify it further.
When someone asks why the home page looks the way it does, the answer is one read away.
Drafted in ChatGPT. Shipped by Claude Code.
The current home direction crossed two different AIs on its way to production. Here is how the handoff actually happened.
One evening, in ChatGPT
The new home direction was drafted in a ChatGPT session and logged into the project journal, decision included.
The next day, in Claude Code
Claude Code ran catch_up on the same journal, received the direction and its rationale, and implemented the page.
Two different AIs, one project, nothing re-explained.
AI memory helps an agent remember facts. Project continuity helps a project preserve its state, history and rationale so that someone or something else can pick it back up later.
What is still open, right now.
An honest journal includes the unfinished work. These tasks are open in the Palimpse project as this page ships.
Move full-text search to
websearch_to_tsqueryAgents write boolean queries that silently return zero results. Found by an audit session, open until fixed.
Frame the “topic” concept as the unit of resumption
Deliberately postponed until the scope is clear. The postponement itself is on the record.
Reconcile the canonical positioning doc with the new home
The landing moved faster than the doc. The gap is tracked, not forgotten.
These sit on top of every briefing until someone closes them.
How the loop runs, day to day.
Start briefed with catch_up
Every session opens with what changed since the last visit, the open tasks on top, and the project memory of durable facts.
Two real entries from the project memory
- “UI copy and docs: never an em dash.”
- “The two marketing pillars (team + neutrality) merged into one notion: continuity.”
The first one governs the copy on this very page.
Work in whatever tool fits the session
Some sessions happen in claude.ai, some in Claude Code, some in ChatGPT or Cursor. The journal does not care which client did the work.
Close with log_session
The session ends with a record: a summary, the decisions with their rejected alternatives, the docs touched, the tasks opened or closed. The next session starts from that state.
Decisions preserve the why. Open work preserves the present. Documents preserve the knowledge. Sessions preserve the movement.
Nothing here was reconstructed after the fact.
Even this page is a trace of the loop it describes. The SEO strategy was written as a doc in the project, an agent broke it down into tasks, and executing one of those tasks produced the page you are reading. It is the project journal speaking.
Project continuity is the ability to resume a project with its current state, history and rationale intact, regardless of which human, AI, tool or session picks it up next.
Give your project the same journal.
Everything quoted above lives in a working Palimpse project. Start one for your own work: free for solo, early access for teams.