ImpactGround.ai
DATA & TRUSTBring the question.
Bring the question.
Keep control of the evidence.
ImpactGround is designed to help teams reason with sensitive organisational context without blurring sources, product suggestions and accountable human decisions.
See how information becomes a decision record.
Every stage preserves the distinction between what was supplied, what was inferred and what was approved.
Organisation profile, strategies, portfolio data and the live question.
Classification, residency, access and permitted use are recorded before processing.
Sources, assumptions, options, forecasts and gaps remain visibly qualified.
An authorised person records direction, rationale, conditions and dissent.
Versions, approvals, exports, retention and deletion follow the agreed deployment policy.
No vague “we use your data” statement.
This is the operating boundary of the current release. Production commitments are agreed in the client deployment schedule.
DATAWHY IT EXISTSSAVED?AI USEACCOUNTABILITY
DATAPublic website activity
WHY IT EXISTSOperate and improve the public experience
SAVED?Standard hosting logs may apply
AI USENot used to assess an organisation
ACCOUNTABILITYPublic-site operator
DATAOrganisation profile and ask
WHY IT EXISTSFrame the portfolio or initiative question
SAVED?Yes, in the authenticated workspace
AI USEMay structure fields or suggest prompts when a connector is approved
ACCOUNTABILITYNamed workspace users
DATASource documents
WHY IT EXISTSGround claims in client-provided evidence
SAVED?Pilot: checksum, label and file record
AI USEFile contents are not currently parsed, summarised or sent to a model
ACCOUNTABILITYUploader and authorised reviewers
DATACanvases, options and reports
WHY IT EXISTSCreate a reviewable decision and learning record
SAVED?Yes, with versions and attributable activity
AI USEDrafting support never constitutes approval
ACCOUNTABILITYDecision owner and approving body
Trust should be visible in the workflow.
The product should make safe behaviour easier than working around it.
Each organisation has its own users, workspace, storage policy and regional requirement.
Owners, managers, reviewers, approvers, contributors and viewers receive different capabilities.
Publisher, period, reference, method, limitation and review state stay attached to a claim.
Saved changes, report snapshots and approval responses remain attributable.
Forecasts, proxies and product suggestions are never presented as observed impact.
Saudi-restricted uploads and processing remain blocked until the regional deployment is verified.
Intelligence helps the team think. It does not inherit authority.
↗ Structure supplied context
↗ Surface missing evidence and follow-up questions
↗ Retrieve candidate public context through an approved connector
↗ Compare options and draft working materials
× Invent sources or beneficiary results
× Turn a forecast into evidence
× Approve a decision or external claim
× Silently transfer a restricted document
Dammam is the target data plane. It is not live yet.
CNTXT onboarding has been submitted. The public website remains global and separate; a Saudi client workspace will move application compute, database, files, backups and approved processing into a controlled me-central2 deployment.
Dammam application, PostgreSQL, object storage, regional endpoints and tenant gating are defined.
Greenbox has submitted the onboarding request for the CNTXT-managed Google Cloud account.
Provision, scan, test, document the data flow, complete legal/security review and approve the production tenant.
Saudi hosting will be shown as live only after the environment, data flow, contracts and controls have been verified.
Evidence a client can ask for.
Production buyers should not have to rely on web copy alone.
Services, regions, trust boundaries, transfers and subprocessors.
Role, record, action, approval and joiner/mover/leaver boundaries.
Retention, backup, export, deletion and verified disposal.
Configuration review, vulnerability results, penetration test and remediation record.
Model or connector, fields sent, purpose, region, logging and human review.
Detection, escalation, notification, recovery objectives and tested response.
A category block is not a malware verdict.
ImpactGround uses HTTPS on a first-party business domain. If an organisation blocks the wider Artificial Intelligence category, its security administrator can allow this exact service without changing the wider policy.
impactground.ai and www.impactground.ai · TCP 443
B2B impact portfolio planning, comparison, governance and evidence management.
Public pages and the current pilot use same-origin scripts, styles, images and application calls.
Use a custom URL/domain Allow indicator for immediate access; dispute the category separately if the tenant believes it is incorrect.
Local hosting is one control—not the whole answer.
Qualified legal, security and sector specialists must confirm the obligations and evidence required for each client deployment.
PDPL
Personal-data purpose, minimisation, retention and permitted transfer conditions.
Official source ↗NCA CLOUD CONTROLSClassification, cloud risk, protection, localisation and operational controls.
Official source ↗SAMA REQUIREMENTSAdditional cloud, outsourcing, risk and approval requirements for regulated financial institutions.
Official source ↗Review the boundary with us.
Bring your data classes, hosting requirement, identity model, retention policy and intended AI uses. We will map what is permitted, what is blocked and what must be evidenced before launch.
Discuss deployment requirements →