Your team runs AI assistants that produce a lot of work fast, and each new session starts cold. Spor gives every assistant and every teammate the context the team already worked out the moment they pick up a piece of work, and the decisions, standing rules, and conversations behind it stay attached as the work moves. Because it all stays connected, you can follow any finished piece of work back to the decision that shaped it and the question that first raised it.
Watch one piece of work move through Spor. Siloed work gets wired into a single connected record; the calls only a person should make compile into a short queue; someone answers one, and that unlocks a task an assistant can pick up and run — with the result flowing right back into the record.
Spor runs alongside the AI assistants your team already uses. Three things happen across a working session while people and assistants get on with the job.
Open a piece of work and the context is in front of you instead of a blank chat. When someone starts the task to build the guest checkout path, the question “Can shoppers pay without an account?” and the decision to keep one cart and attach the buyer later are right there. No one has to go find them first.
While the work happens, the records that bear on it come up on their own. The standing rule that every refund keeps an audit trail surfaces before it is missed, along with the earlier decision to roll guest checkout out to staff stores first and the approaches the team already ruled out. You see what applies without stopping to ask.
When the work is done, what was decided and built is written back into the record. The next person or the next assistant picks up where Maya left off rather than from nothing. The result links to the task it finished and the decision that shaped it, so the trail stays intact.
Start here:
The clear pick is task-guest-path-build (7.38) — finish the guest checkout path. It’s well ahead of the rest, a lot of related work is moving right now, and you touched it two days ago, so pick that up first.
After that there’s a cluster around 5.8–6.4: the card-retry window, the cart tax preview, and holding inventory during checkout. All in progress, all moving.
task-confirmation-email-edit is newer (1d) but climbing — keep half an eye on it.Spor ships as an app for your assistant. Ask for your queue or any view and it renders inline, as boards, the trail behind a piece of work, and decision queues, with buttons that approve, start, or close an item right there in the conversation. It’s built on the open MCP Apps standard, so it runs first-class inside Claude and ChatGPT Apps and falls back to clean text everywhere else.
Your assistants prepare and connect work and keep it current: they draft a task, link it to the decision behind it, and update it when something it depends on moves. People make the calls that carry responsibility, like turning guest checkout on for every store. Both people and assistants work in one shared record, and every action is recorded under the person it belongs to. When Maya and her assistant open the same task, each sees it the way they read it best.
This lets shoppers buy without creating an account, on every store at once. Your agent built the path and ran it end to end against test orders; flipping it on is the human call. Account checkout keeps working exactly as it does now, and turning it back off is one switch.
decision · staged-rollout question · checkout-without-accountAssistants now produce far more work than a team could before, so the scarce thing is human judgment rather than the work itself. They can draft, connect, and prepare work all day; Spor keeps the human queue short and loads each call with the context you need to decide it. Maya opens to the same place everyone does: the calls only a person should make, in priority order, each one ready to decide. Her assistant built the guest checkout path and ran it end to end against test orders, so when the call reaches her she can turn it on for every store, and turning it back off is one switch. A short list is the goal here, and it is a sign of a team that stays on top of its work.
When a decision is replaced, the shared record registers the change, so the retired answer stops surfacing as current guidance and the work that relied on it is flagged to revisit.
Your team first decided to turn on guest checkout for every store with a single switch, then replaced that call with a narrower one: roll it out to staff stores first. Spor knows the second decision replaced the first, so anyone briefed now gets the staff-stores-first answer and the all-stores version no longer shows up as live guidance. Any work that was started against the retired decision gets flagged for someone to revisit or close, instead of quietly proceeding on a call the team has already moved past.
Spor is a connected record of the decisions, tasks, open questions, and standing rules behind a team’s work, with each item linked to what it came from and what depends on it. It runs behind the AI coding tools your team already uses, so it stays current as the work happens rather than waiting for someone to update it.
A ticket tracker holds the what and the status; a wiki holds prose that someone has to maintain by hand. Spor keeps the reasoning attached to the work it explains, so the context reaches whoever is working without anyone searching a doc for it, and a correction you record carries forward to everyone who comes after you. When a decision is replaced, the old one is retired instead of resurfacing, and any call that genuinely needs human judgment goes to a person rather than getting answered automatically.
Spor is opening to a handful of teams at a time. Tell us where you work and we’ll bring you in as the next cohort opens.
sporhq.io