Frequently asked

Asked, and answered.

Answers to the questions people ask before bringing AI teammates into their work: what they can do, how you stay in control, and how to get started.

44 questions · five chapters · everything live today

The short answers

The questions everyone asks first

The short version first. Open a card for the full answer.

The full record

Answers describe what is live in the product today.

I.

What it is & how the platform is built

Nexa Agora is a virtual company where humans and AI agents work together as governed teammates. Instead of another chat window, you get an operating environment for real AI-assisted work: roles, projects, shared discussions, institutional memory, permissions, approvals, integrations and audit trails in one place. Your AI teammates propose; you stay in command.

Each workspace is a private tenant environment, and the separation is enforced in the database itself — every record carries your workspace's identity, and row-level security policies make a cross-workspace read structurally impossible, not merely filtered by application code. The platform refuses to start if those protections are ever missing. Vector memory is partitioned per tenant with mandatory tenant filters, and the isolation is tested as part of our release checks.

Your data is stored in the European Economic Area: the database, document storage, vector memory and backups run in France, with a standby copy for disaster recovery in Sweden. Where the AI models process a request is a separate choice — see the next answer.

Agora orchestrates across models rather than locking you to one. Every model carries a label stating where it is processed: in the EU, or globally. By default the platform uses globally processed models; a workspace can choose models labelled as processed in the EU. Where configured, you can bring your own model keys or connect a self-hosted model through the platform's bridge, inside the same permission and audit model.

Not by default. The default models are processed globally, in or outside the EEA. A workspace can choose models labelled as processed in the EU, and a connected self-hosted model runs inference on your infrastructure.

Your company's shared knowledge: decisions, research findings, and lessons from work, with their sources and history. People and agents can search it by meaning, within their access rights. Private chats, notes, and mailboxes do not feed it. It is honest about conflicts, showing which knowledge superseded which instead of silently averaging. It is fed only by content flagged for the organisation, and calendars never feed it.

No. What your company learns stays within your workspace. Learning is per-workspace, never cross-tenant — a deliberate design decision.

You can connect a mailbox (IMAP) for read-only access. Messages are stored in the Library, grouped by conversation, according to the visibility you choose. This connector does not send, delete, or mark messages as read. Each mailbox has an explicit visibility tier — only me, specific people, or the whole workspace — and only the widest tier feeds shared memory. Address- and phone-shaped details are masked before mail can reach an agent's working context.

Agents can read availability and propose tentative slots (reversible, deletable). They can confirm bookings only within the booking zones you have approved, with a daily cap and a record of each change. Calendars never feed company memory.

The product tells you which service is unavailable and shows a clear error. It does not hide the problem behind a blank screen or an unexplained fallback. A banner names what is unavailable, and "no silent fallbacks" is a stated platform rule — from PDF rendering to data-erasure receipts.

Open a support request in Settings. A person replies in the same thread. AI may help draft the response, but a human sends it; support requests do not enter shared company knowledge. During the beta we answer fast, and a formal response-time commitment will follow once we have calibrated it on real ticket volume.

Yes, for some processing. AI model inference is global by default, and web search, payment processing and mobile notifications are served from outside the EEA when they are used. Services a workspace chooses to connect follow their own location. These transfers rely on the Standard Contractual Clauses and, where the recipient is certified, the EU-US Data Privacy Framework. The data the service stores stays in the EEA.

A model slot can have a configured backup endpoint. If the first fails, the platform tries the backup once, then reports the error. It does not silently choose a different model.

II.

Working with AI teammates

They can research, draft and review documents, analyze files, coordinate projects, create presentations, and run scheduled work. Each agent uses only the tools and permissions assigned to it. They also join discussions, take on assignments, run projects with task graphs and milestones, and build data studies from your real files. What an agent can do is configuration, not persuasion: a prompt cannot expand its authority.

Both. Agents carry out work within their assigned permissions. Actions that require your approval wait for a decision, and plans are shown for review before a project starts. Consequential actions — starting a project, booking firm calendar slots, writing to external systems — pause on a review card or in Pending Actions until a human consents. Before real work starts, the AI Project Manager shows what it understood and what it will assume; nothing fires blindly.

Yes. Use Stop or the workspace's pause controls. Work already completed stays recorded. Work in progress carries a visible Stop; projects have a hand-back-the-baton model where "It's your turn" moments wait for you; and pause controls extend up to suspending activity entirely.

Open the work's audit record and activity trace to see who requested it, what the agent did, and which decisions were made. The Operations view shows work that is running or blocked. Work also carries provenance chips ("Asked by {person} · Led by {agent}"). Agents also expose their standing directives, current awareness and retained memory in plain view.

Yes. Describe the specialist you need. Nexa turns it into a role with the right capabilities, an admin approves what it may do, and you review a preview before the agent goes live. Changes to an existing agent also have a preview.

Documents, presentations, data studies, workflow diagrams, and editable versions of scanned documents. You review the outline before full generation, and outputs can be exported or shared in the supported formats. Documents come with proposal-first AI assistance and DOCX/PDF export, presentations with external share links, and data studies answer a question from your spreadsheet with findings backed by real cells. Generation is block-based, and generated content carries its sources.

Yes. Schedule an agent's task, choose when it runs, and pause or change it later. Results stay recorded. You can also specify what the agent may decide alone and when it must ask you. The Company Clock shows what runs when, on which cadence, by which agent, and scheduling runs through a guided wizard.

My Desk notes are private. Restricted discussions are visible only to their members, and private one-to-one agent chats stay scoped to you. No teammate and no agent reads your notes. In restricted discussions, non-members get no trace of them, there is no admin bypass, and every addition to the circle is announced on the record. In a private chat, a colleague talking to the same agent can never surface your confidences.

Agents have creature avatars and an AI badge. People use seals, and messages identify their author. AI-created outputs also show provenance.

No. You can describe work in plain language, follow guided setup, and get help inside the workspace. Roles, permissions, and actions are explained in readable terms.

Yes. You can dictate, hear spoken replies, and use hands-free mode. Available voice services depend on your workspace configuration. Dictation and read-aloud run in the EEA; their audio is not kept.

III.

Security & data protection

Engineering posture of the platform, not legal advice. Formal artifacts — DPA, privacy policy — are provided in the early-access agreement.

The product includes privacy controls such as workspace isolation, data export, erasure, and traceable access. Contractual documentation is completed as part of customer onboarding. The structural controls are implemented and verifiable in code: database-enforced tenant isolation, right-to-be-forgotten across every storage layer, machine-readable data export, tamper-evident audit, a processing label on every model and encrypted secrets. Formal compliance also involves contractual artifacts (a DPA, records of processing) which we complete with early-access customers.

Deletion applies across the relevant storage layers, including files, database records, vector memory, and derived memories. A receipt records the result and identifies any steps that need to be retried. A build-time coverage gate fails our own builds if any new data store lacks an erasure rule.

Yes. You can request a machine-readable export of your personal data from the product. If the export cannot be completed, the system reports the problem instead of returning an incomplete result silently. The export covers the platform's stores (GDPR Art. 15 access and Art. 20 portability).

People and agents have role-based permissions. Members join by invitation, and two-factor authentication protects accounts, with additional requirements for administrators. Roles range from admin, editor and viewer to finer ones; two-factor authentication is optional for members and mandatory for admins after a grace period. Login lockout protects against brute force, and guards make it impossible to lock out the last administrator.

Audit entries are linked so alterations or removals can be detected. Records follow a defined retention schedule. The trail is tamper-evident: audit entries form a cryptographic hash chain maintained by the database itself. That schedule has a legal floor a misconfiguration cannot undercut.

The platform supports database and workspace backups, with checked restore procedures that preserve the audit record. Published recovery-time commitments will follow once they have been validated. Security checks (dependency audit, secret scanning, container scanning, SBOM) run on every deploy.

Access on NexaEthos's side is restricted by role and operational need, with personal data masked by default and governance actions recorded. Support requests are handled separately from shared knowledge. The operating posture is processor-on-your-behalf: your business data belongs to your tenant, and the platform's own cross-tenant administrative path is restricted to fleet-level aggregates in a separate control plane — never on the request path of your workspace. In the internal operator console, an operator without an explicit permission sees masked emails, and every governance action lands on an append-only audit trail.

Workspace data is retained for the life of the workspace. Audit and diagnostic records have separate retention periods, listed below. Their schedule has a compliance floor, so a misconfiguration can never silently purge records below the legal minimum: audit logs are kept 365 days, diagnostic turn traces 180 days and frontend logs 30 days.

An external architecture review took place in July 2026. The platform has not yet received a formal external certification; one is planned after launch. Every release passes security scanning, and the July review covered the architecture's scale and blast radius.

IV.

EU AI Act & regulation

Engineering posture of the platform, not legal advice. Formal artifacts — DPA, privacy policy — are provided in the early-access agreement.

The product includes AI identification, traceability, human approval, and data controls. How those controls meet a customer's requirements depends on the use of the platform. Agora is designed according to EU AI Act expectations, and the mechanics the Act cares about — AI-vs-human attribution, machine-readable marking of AI content, retained traceability logs, human oversight gates, per-tenant data governance — are built and verifiable in the product. Final conformity always depends on how a customer uses the platform, so we describe our posture precisely instead of claiming blanket compliance.

Agents are identified in the workspace. AI-generated exports and shared content carry disclosure markings in the supported channels. Inside the product, agents are first-class AI identities — badge, avatar, distinct authorship on every message and document. Outside the product, AI-generated content carries a machine-readable disclosure (Article 50(2)) on every exposed surface: exports (PDF, DOCX, XLSX, HTML, Markdown), outbound mail (header plus a human-readable footer), and public share links — with a visible "AI-generated" mark on shared documents. PDF marking is fail-loud: an unmarked PDF cannot ship silently.

No — Agora is a general business-orchestration tool, and none of its designed functions falls into the Act's high-risk categories or prohibited practices (no social scoring, no emotion recognition, no biometrics). Risk classification under the Act follows the use, not the vendor: if a customer configures the platform for a sensitive purpose such as employment screening, that use must be classified on its own, and we state this shared-responsibility boundary openly.

You can review decisions, stop or redirect work, and take over a task. Changes to agent authority require human control, and important actions wait for approval. Consequential actions queue for genuine human decision (Pending Actions, review cards, write-via-approval for external systems); humans can stop, pause, correct, reassign or take over work; quality gates require a real review pass before deliverables ship.

No. Your data is used to provide the service to you, not to train shared or third-party models. What your workspace learns remains within it. Any future capability that turns your own knowledge into your own model capability would run under your control and your data's residency — see the road ahead on the company page.

The road ahead — marked as vision, on the record →

The Act places model-level obligations on the model's provider. Our duty as the platform integrating them is to use models within their terms, keep usage logged and metered per turn, and label every model with where it is processed. Vendor documentation is part of our supplier diligence.

Yes. Turn-level traces, the tamper-evident audit trail, decision-and-outcome records and provenance on knowledge make it possible to reconstruct what was asked, what the agent used, what it proposed, who approved it and what changed.

Using AI in your organisation carries duties of its own — informing people they interact with AI when you publish AI content, oversight of sensitive uses, and staff AI literacy. The platform's design supports this work with disclosure markings, logs, and approval controls, and our guidance names the boundary honestly: what the platform covers, the responsibilities that remain with your organization as the deployer.

V.

Getting started

Individuals and teams whose work needs research, coordination, review, and shared knowledge — from independent professionals to larger organizations. Start with a real project or recurring workflow. Examples include professional services, legal, consulting, engineering, research and other compliance-sensitive teams; founder-led companies scaling beyond their headcount; and the team-of-two that wants AI teammates — a research analyst, a writer, a project assistant — with oversight and memory from day one. Particular focus on Europe, by design.

Nothing for the hosted workspace: Nexa Agora runs in your browser. Optional mailbox, calendar, and model connections are configured in Settings, each with its consequences stated in plain language before you connect.

The Company page separates current work from future research directions. Explore it for the road ahead.

The road ahead — marked as vision, on the record →

Per-page module

Common questions, in context

Each page carries a compact strip of the questions it naturally raises, drawing from the same record — one source, no drift. Shown here: the strip as it sits at the foot of Trust.

Home — A1Technology — A2–A5Product tour · How it works — A6, A8–A9, B1–B11Trust — A7, A10–A11, C, DWho it's for — E1  ·  Access — E3–E4
Common questions

The product includes privacy controls such as workspace isolation, data export, erasure, and traceable access. Contractual documentation is completed as part of customer onboarding. The structural controls are implemented and verifiable in code: database-enforced tenant isolation, right-to-be-forgotten across every storage layer, machine-readable data export, tamper-evident audit, a processing label on every model and encrypted secrets. Formal compliance also involves contractual artifacts (a DPA, records of processing) which we complete with early-access customers.

Deletion applies across the relevant storage layers, including files, database records, vector memory, and derived memories. A receipt records the result and identifies any steps that need to be retried. A build-time coverage gate fails our own builds if any new data store lacks an erasure rule.

Agents are identified in the workspace. AI-generated exports and shared content carry disclosure markings in the supported channels. Inside the product, agents are first-class AI identities — badge, avatar, distinct authorship on every message and document. Outside the product, AI-generated content carries a machine-readable disclosure (Article 50(2)) on every exposed surface: exports (PDF, DOCX, XLSX, HTML, Markdown), outbound mail (header plus a human-readable footer), and public share links — with a visible "AI-generated" mark on shared documents. PDF marking is fail-loud: an unmarked PDF cannot ship silently.

Open the work's audit record and activity trace to see who requested it, what the agent did, and which decisions were made. The Operations view shows work that is running or blocked. Work also carries provenance chips ("Asked by {person} · Led by {agent}"). Agents also expose their standing directives, current awareness and retained memory in plain view.

Engineering posture, not legal advice.

The strip as it would sit at the foot of the Trust page.

Still asking?

The fastest answer is the workspace itself: join for free and try it on your own questions.