HOW CORE WORKS
People contribute. AI helps investigate. Humans decide.
CORE is a virtual Earth for working on real problems. The path has five stages. Each one below says what works today and what is not built yet.
WHY AN EARTH
Why a shared virtual Earth?
A real problem happens in a place. On one shared Earth, the street, the problem, its evidence, what people have seen there and the analysis sit in the same location. You can see how they connect, instead of hunting through separate reports, threads and agencies.
It is also the simplest way to start: go to a place you care about and see which questions are open there. Today this exists only for San Francisco. Other cities are on the Earth without an investigation for now.
1 · DISCOVER
Where do problems come from?
Today
The CORE team records problems from dated public data it is allowed to use. San Francisco has two open missions: Folsom Street signal timing and the affordable-housing pipeline.
CORE does not automatically find the world’s problems. Other cities on the Earth have no investigation yet.
Planned, not built yet
Residents, researchers, institutions and companies bring problems they see, and agents examine them using authorized data.
2 · CONTRIBUTE
What do I actually do?
Today
More than chat. You open a mission, question the agents and challenge what they say, test your own assumptions in a calculation, compare scenarios and write your notes.
You can also save an observation or idea with a place and a source link. It stays in this browser: it does not sync between devices, nobody receives it and nothing is sent to any institution. There is no public sign-in and no billing yet.
Planned, not built yet
Accounts, shared contributions with moderation, photos, and votes on priorities. Participation stays free, and a subscription never buys influence.
3 · INVESTIGATE
What do the AI agents do?
Today
Agents are software roles, not people. They use a mission’s authorized evidence, run permitted calculations, state what is missing and set options side by side. They cannot act in the real world, and they decline if you ask them to.
On this public site, agents answer from recorded work and bounded calculations, with each answer labeled. A seven-agent investigation run on a local model is shown as a recording, not live.
Planned, not built yet
More capable agents that work live on broader authorized data.
4 · EVALUATE
What validates a solution?
Not one thing, and not the AI. Six separate checks build on one another. Today, only the first two exist.
- Evidence checksWorking
Sources, licenses and declared gaps are checked in code before anything is used.
- Exploratory simulationsWorking, uncalibrated
Useful to compare ideas. They are not forecasts and prove nothing about the real street.
- Scientific model validationNot done yet
Models must first be calibrated against real measurements.
- Independent expert reviewNot available yet
No reviewers are connected. A review request stays in your browser and nobody receives it.
- Institutional decisionOutside CORE
Only the responsible institution can decide. CORE controls no government or public infrastructure.
- Measured real-world outcomeNone so far
No CORE proposal has been implemented or measured in the real world.
WHO DECIDES
Who validates the conclusions and who decides?
Neither CORE nor its agents. Independent experts are meant to review the methods and conclusions. None is connected yet, so nothing on this site has been validated by them.
The decision belongs to the institution responsible for the place, under its own authority: for Folsom Street that would be San Francisco’s transport agency (SFMTA), which has no agreement with CORE. A subscription never buys a vote or priority.
5 · FOLLOW IMPACT
What happens to my contribution?
Today, your work stays with you in My CORE. Everything after the first two steps below is the intended path and is shown for illustration only: none of it exists yet.
- Saved in your browserWorks today
- Marked for review, in this browser onlyWorks today
- SubmittedIllustrative
- Under investigationIllustrative
- Evidence assessedIllustrative
- Scenario exploredIllustrative
- Awaiting expert reviewIllustrative
- Considered by an institutionIllustrative
- Approved for implementationIllustrative
- Implementation underwayIllustrative
- Impact measuredIllustrative
Who can act? Only the people and institutions responsible for the real place. CORE can prepare the evidence for them. It does not control their decisions.
TOWARD THE REAL WORLD
How could a solution reach the real world?
CORE applies nothing by itself. The intended path: a report with its sources and uncertainty goes to experts and the responsible agency. If the agency chooses to act, it follows its own process: studies, approvals, budget and works. Measured results then come back to contributors.
Today you can produce a decision brief stamped “not approved”. No later step exists yet, and no agency has agreed to receive one.
COMING BACK
Why would I come back?
Today: your missions, notes and observations wait for you in My CORE, in this browser. You can resume an investigation, test other assumptions and follow a mission.
Planned: with accounts and shared contributions, follow a place, see agents work on it and learn what was decided. This is a hypothesis: no return rate has been measured.
WHO IT IS FOR
Who is CORE for?
Residents and curious people, for free. Explore, question the evidence, test assumptions and save what you have seen.
Researchers and experts. Inspect the sources, methods and limits of each result. Expert review is not connected yet.
Organizations and institutions. Private workspaces, authorized data and defensible reports are proposed, not available. See the plans.
WHERE CORE STANDS
The honest picture, today
- Vision
- A shared virtual Earth where people and AI work on real problems, growing into a global infrastructure for collective intelligence. This is the long-term aspiration, not a present achievement.
- What exists today
- A public prototype: a 3D Earth with six cities, two San Francisco missions you can run in your browser, and local contributions. No accounts yet.
- AI agents
- The registry holds 510 software agent identities, 501 of them for San Francisco. 35 have recorded work. One recorded investigation used seven agents on a language model; on this site agents answer from recorded work and bounded calculations.
- Scientific status
- No model is calibrated, no independent review has taken place and no real-world improvement has been measured.
- Traction
- About 30 early testers have given feedback, as reported by the founder. No revenue, no customers and no signed partnerships.
- What comes next
- Accounts, shared and moderated contributions, independent scientific validation, then more cities and more capable agents.
- CORE World and CORE Business
- CORE World is the free public Earth, where participation stays free and equal. CORE Business and institutional plans are proposed, not available: private workspaces, authorized data and defensible reports for organizations. Their revenue is meant to fund free access, and a subscription never buys influence.