Context · Decisions, glossary and the knowledge map
What the team agreed on, known to every AI.
Context is the hub’s memory: decisions, a glossary and the links between them. Each teammate’s AI checks it before answering, so ChatGPT, Claude and Gemini stop giving one team three different truths. When two decisions clash, the hub says so.
Any agreement is context
A price limit, a typeface, a sample size, a trip budget: if the team settled it, every AI should know. Pick a team to see its context at work.
Private hub “Launch v2”: R Rasul · ChatGPTN Noor · ClaudeB Berık · GrokS Shigeo · Gemini
Context of “Launch v2”
- accepted Discounts 10% off at most; anything more goes through the founders Rasul via ChatGPT
- proposed Launch date Launch on March 15 Berık’s Grok · waiting for a person
- term Pilot: the first 30 days are free
- conflict “Launch on March 15” against “Launch on March 1”
Rasul’s private chat with ChatGPT
The same answer in every teammate’s AI: it comes from the hub, not from one chat’s memory.
Try saying to your AI
- “Save that we launch on March 15 as a team decision” The hub checks it against earlier decisions, finds “Launch on March 1” and tells you about the conflict.
- “What did we decide about discounts?” Your AI answers with the team’s decision and who saved it, not with a guess.
Context of “Riverside rebrand”
- accepted Typeface Headlines are set in a serif; sans only for body text Maya via Claude
- proposed Delivery Final files go out on Friday Ines’s Gemini · waiting for a person
- term Round: one batch of client feedback; three rounds are included
- conflict “Final files go out on Friday” against “Client review takes five working days”
Ines’s private chat with Gemini
The same answer in every teammate’s AI: it comes from the hub, not from one chat’s memory.
Try saying to your AI
- “Add “round” to our glossary: one batch of client feedback, three included” The term is saved. If a similar one already exists, the hub shows it instead of making a duplicate.
- “What have we agreed with the client so far?” The hub’s decisions with who saved each, including the ones still waiting to be accepted.
Context of “Q1 pipeline”
- accepted Pilots A pilot lasts 30 days; extensions go through the head of sales Arman via Grok
- proposed Enterprise discount 20% off for enterprise deals Dana’s ChatGPT · waiting for a person
- term Qualified lead: budget confirmed and the decision-maker met
- conflict “20% off for enterprise deals” against “10% off at most”
Dana’s private chat with ChatGPT
The same answer in every teammate’s AI: it comes from the hub, not from one chat’s memory.
Try saying to your AI
- “Is Northwind a qualified lead by our definition?” Your AI takes the team’s definition from the glossary instead of inventing its own.
- “Show me the open conflicts” The pairs of decisions that can’t both be true, waiting for the team to settle them.
Context of “Sleep study”
- accepted Sample 52 participants, aged 18 to 35 Aiko via Claude
- proposed Exclusions Exclude shift workers Samir’s Gemini · waiting for a person
- term Sleep latency: minutes from lights-out to sleep onset
- conflict “Exclude shift workers” against “Recruit all employed adults”
Lena’s private chat with ChatGPT
The same answer in every teammate’s AI: it comes from the hub, not from one chat’s memory.
Try saying to your AI
- “Save that we exclude shift workers” It contradicts an accepted decision, so it is saved as proposed until a person settles it.
- “Show how the sample decision changed over time” Every version of the decision: who saved it and when.
Context of “DevConf 2027”
- accepted Schedule Doors at 9:30, keynote at 10:00 Zara via ChatGPT
- proposed Lunch Lunch at 13:00 Miguel’s Claude · waiting for a person
- term Green room: room 204, speakers only
- conflict “Lunch at 13:00” against “Workshop B runs until 13:30”
Priya’s private chat with Gemini
The same answer in every teammate’s AI: it comes from the hub, not from one chat’s memory.
Try saying to your AI
- “What changed in the plan while I was away?” New and edited decisions, terms and conflicts since you last asked.
- “Close the lunch conflict: the workshop time stays” The earlier decision stays, the new one is marked rejected, and the history keeps both.
Context of “Lisbon trip”
- accepted Dates The trip is 12 to 16 May Mum via ChatGPT
- proposed Day 2 Beach day in Cascais Sam’s Gemini · waiting for a person
- term Budget: €1,800 for everything except flights
- conflict “Day 2: beach day in Cascais” against “Day 2: Sintra”
Alex’s private chat with Claude
The same answer in every teammate’s AI: it comes from the hub, not from one chat’s memory.
Try saying to your AI
- “Save that the budget is €1,800 without flights” It joins the family’s context, and everyone’s AI plans within it from now on.
- “What have we decided about the trip?” A short list of what is agreed and what is still open.
The life of a decision
An AI can save a decision, but it can’t make the team agree. When a new decision contradicts an accepted one, it waits for a person.
- Proposed
Saved by an AI and waiting for a person. Other AIs see it marked as not agreed yet.
- Accepted
The team’s shared context. Every teammate’s AI treats it as agreed.
- Replaced
A newer decision took its place. It stays in the history.
- Rejected
Turned down or withdrawn. It stays in the history too.
Everything in Context
Each part works two ways: by hand, in the hub’s widget or on the website, and through your AI, in plain words.
Decisions and conflicts
A decision is one thing the team agreed on: a topic, a summary and the reason. Saved once, it is known to every teammate’s AI.
By hand
- Add a decision on the website or in the widget: topic, summary, rationale
- Accept or reject a decision that an AI proposed
- Settle a conflict: keep the new decision, the earlier one, or both
- See who saved each decision, and through which AI
Through your AI
- “Save that we agreed…” stores the decision in your words
- Every new decision is checked against the earlier ones for contradictions
- “What did we decide about…?” answers from the team’s decisions
- “Accept it” or “close the conflict” works right from your chat
Glossary
The words your team uses its own way. With a glossary, “pilot”, “round” or “qualified lead” mean the same thing in every AI.
By hand
- Terms with the definitions the team agreed on, and their synonyms
- A similar term already exists? The hub shows it instead of saving a duplicate
- Edit a definition any time; earlier versions are kept
Through your AI
- “What does … mean in our team?” answers with the team’s own definition
- “Add … to the glossary” saves a term for everyone
- A different definition replaces the old one only after you confirm
Knowledge map and search
As the context grows, tags and links keep it navigable, and search finds the right piece by words and by meaning.
By hand
- Tags cut the context by theme: “Pricing”, “Q1”, “Client X”
- Links between records: relates to, depends on, refines, replaces, contradicts
- Import: paste your notes or a Markdown file and review the drafts before saving
- Export the whole context as one Markdown file
Through your AI
- Asks the hub by topic and gets the matching decisions, not the whole archive
- Linked records come along: what a decision depends on or contradicts
- “Show the knowledge map” lists tags, records and the links between them
- Tags and links records on request (owner and admins)
History and restore
Every version of a decision or term is kept, with who saved it and when. An earlier version can be brought back.
While you were away
Open the hub and see what changed since your last visit, person by person. A stand-up without the meeting.
Review rules
The team chooses when an AI’s decision needs a person: only when it conflicts, always, or never.
Honest attribution
Every change records who made it and whether it came from a person or from their AI.
Updates as it happens
A new decision or conflict shows up in the open hub without a reload.
For coding agents
One section in your repository’s AGENTS.md, and any coding AI in the project works from the hub’s context.
Context is fed by the rest of the hub: decisions are usually born in a chat or on a call in Sync, and Apps can read them to stay within what the team agreed.
Questions about Context
- How is this different from my AI’s own memory?
- An AI’s memory belongs to one person and one app. Context belongs to the team: the same decisions and terms reach ChatGPT, Claude, Gemini and every other AI connected to the hub.
- Can an AI change what the team decided?
- It can save a decision when you ask. If that contradicts an accepted decision, it is saved as proposed and waits for a person. Every version is kept and can be restored.
- How does the hub find contradictions?
- Each new decision is compared with the team’s current decisions. If two of them can’t both be true, the hub opens a conflict and shows both sides.
- Can we take our context out of Beringio?
- Yes. The whole context exports as one Markdown file, and notes or a Markdown file can be imported back.
Private to your team
Everyone brings their own AI, and the hub stays yours.
Invite only
A hub is private from the start: only the people you invite can open it.
Encrypted at rest
Chat messages and files are stored encrypted, with a separate key for every hub.
Your AI acts when you ask
It posts, saves or changes something in the hub only on your direct request.
Hub text is data
What teammates wrote reaches your AI as data with its source, never as instructions.
Context is one third of the hub
The three parts feed each other. See how Context connects to the rest.
- Context + Sync Sync: Chat, calls and a shared board Every AI checks the context before it answers in chat, and an open conflict is a good reason for a quick call. Explore Sync
- Context + Apps Apps: Small apps your AI builds for the team Turn decisions into a kanban, the glossary into a quiz: apps read the context with the team’s permission. Explore Apps
Start a private hub for your team.
Sign in with email, connect the AI you already use, and send your team an invite link.