The context you build in Stoa belongs to you. It travels with the work, not with the vendor. When your team makes a decision in a session, that decision doesn't get locked inside a platform you have to remember to open. It's immediately available — in your next session, for any agent running in the sandbox, in a local tool you're building with, wherever the work goes next.
This is not a minor product detail. It's the architecture.
What this actually means
Three things follow directly from this position:
Accessible wherever you work
In the web app, running an agent in the shared sandbox, working locally, connecting a custom tool via API — your team's decisions, open questions, and accumulated context travel with you. There's no “upgrade to export” gate, no integration that requires an enterprise plan, no context that only exists inside the product's own interface.
Portable, not just exportable
“Export” is what incumbent knowledge tools offer when they're trying to seem open. A PDF of your meeting notes. A CSV of your decisions. A lossy dump that strips the relationships between decisions and the conversations that produced them.
Stoa stores your context in a structured, queryable format. When you move it — to a new tool, to a local environment, to an agent — it moves with its meaning intact. The decision carries its rationale. The rationale carries its participants. The participants carry their timestamp. Nothing is flattened.
Yours, not theirs
Your team's institutional memory — the decisions that shaped the product, the reasoning behind the pivots, the customer insights that changed the roadmap — is yours to keep. It doesn't become a training set. It doesn't get analyzed to improve someone else's product. It doesn't disappear if you cancel.
The context layer that Stoa builds from your conversations is the persistent asset. Everything else exists to keep feeding it.
Bring your own files
Every artifact Stoa produces can be a file. Not a proprietary blob locked in a database — a real file, with structure you can read, tools can parse, and agents can reason over.
This matters in two directions:
Files Stoa creates for you
When a session ends, you get more than a recording link. Decisions are extracted into structured records — title, rationale, participants, timestamp, and the session that produced them. Open questions are tracked. A team context layer accumulates across sessions. All of this exists in a format that tools can work with directly, not just humans staring at a screen.
You can pull any of this into a local directory, a shared drive, a Git repo, or a tool you've built. It stays current. It stays connected to the session it came from.
Files you bring to Stoa
You can also bring your own context in. Drop in a spec doc, a previous decision log, a product brief, a customer research summary. The agent sandbox has access to it. Your next session starts with it already loaded. Context flows into Stoa as easily as it flows out.
The practical test
Could you, right now, give an AI agent access to everything your team has decided and discussed in the past six months — in a format the agent could actually reason over? Not a transcript dump. Not a pile of PDFs. Structured decisions with context and participants and dates, connected to the conversations that produced them.
For most teams using incumbent tools: no. That's the gap Stoa is built to close.
Working locally
If you prefer to work locally — or if you're building something that does — your Stoa context is designed for that. The session output isn't a web-only artifact. It's structured data you can sync to your machine, version alongside your codebase, and feed to local models or agents without touching the Stoa interface again.
For engineering teams, this means the decision log from Tuesday's product meeting can live in the same repo as the code it shaped. For founders, it means the early customer conversations that drove the pivot are there when you're prompting a new agent to draft a spec — not buried in a drive somewhere.
For teams
Context isn't personal. When a decision is made in a Stoa session, it's immediately available to everyone on the team — in the shared sandbox, in the web app, via API. The team context layer is shared by default. New hires can be granted access to the full decision history the day they start, in a format that agents can surface intelligently.
Why this matters for founders
The decision log is the most valuable thing a conversational startup produces. It's more important than the roadmap, more durable than the pitch deck, more useful to a new hire than any onboarding document you'll ever write.
And yet most founders store this intelligence in a way that makes it almost impossible to actually use. Decisions in Notion. Context in Slack threads. Call recordings in a drive somewhere. The knowledge exists; accessing it is the problem. None of it is available to an AI agent without manual curation. None of it travels cleanly to a new tool. All of it is effectively imprisoned in the tool that captured it.
Stoa is designed around the assumption that your context should work as hard as you do. When a decision was made in Tuesday's session, it's there Wednesday morning when an engineer asks why you're building it this way — whether they're asking in Slack, in the sandbox, or in a local tool. It's there when you're prompting an agent to draft a spec. It's there when a new hire is trying to understand how the company thinks. Wherever the work happens, the context is already there.
The opposite of lock-in
The competing knowledge products — the ones that promise to be your “second brain” or your “team wiki” — have a business model that depends on friction. The harder it is to get your knowledge out, the less likely you are to leave. Features compound around the data, not to serve you, but to make the data harder to extract.
This creates a particularly cruel situation for founders. The institutional memory you're building right now — the early decisions, the customer conversations, the strategic reasoning — should compound over time. Instead, it gets stratified into whatever tools you happened to use when you were making those decisions. Two years in, the context from the early days is effectively archaeological: technically present, practically inaccessible.
Your knowledge should get more useful over time, not more trapped.
Stoa's position is simple: every session adds to a context layer that's searchable, structured, and available wherever you're working — in the product, outside it, with agents, without them — today and five years from now.
Your best thinking happens in the room. Stoa makes sure it doesn't stay there.