Transparency and oversight No black box.
No black box.
A glass box.
AI rarely fails on technology, mostly on trust. Trust comes from sight. Every step, every model, every metric is traceable and auditable.
Every step traceable.
One dashboard across every agent on the platform: what they deliver, which model they use, how long they take, and what they cost. Steer by agents and tasks, not by people.
12
Agents in production
1,248
Runs this week
100%
logged in the audit trail
CHF 320
Cost this week
Agent usage · last 7 days Anonymized · no employee surveillance
| Agent | Runs | Model | Avg time | Cost |
|---|---|---|---|---|
| KYC review | In-House | 9s | CHF 84 | |
| Tariff calculation | Swiss | 7s | CHF 60 | |
| Policy analysis | In-House | 18s | CHF 62 | |
| Customer correspondence | EU | 12s | CHF 73 | |
| Reporting | Swiss | 22s | CHF 41 |
100% logged No black box.
No black box.
A glass box.
Every agent action is captured: role, model, data, duration, and cost. You steer by agents and tasks, not by people.
Power and control. Finally both.
Until now you had to choose: powerful AI as a black box on US servers, or safe AI that gets nothing done. AIgent is both, no compromise.
Real work, not just chat
Agents write emails, build reports, and review dossiers. Finished results over tested code, not just text.
Your data stays yours
Swiss-hosted or on-premise. You pick the model per agent. No US Cloud Act exposure.
Every step visible
Plan, model, tool, result, and cost. No black box, a glass box. Fully auditable.
Live in 15 minutes
Pre-configured agents, centrally set up. Pay-per-use, no per-seat fees, no lock-in.
Trust Most AI projects fail not on technology but on trust. AIgent makes every agent step visible and auditable, so IT and compliance can sign off.
Common transparency and oversight questions.
- What is recorded in the audit trail?
- The audit trail records every AI interaction with model, tokens consumed, cost, and response time. It also captures the agent’s own run: which plan was made, which tool was called, and which result was produced. Every operation can be fully reconstructed and audited after the fact, completely, instead of being pieced together from scattered logs. Every API call lands in the same audit trail as actions in the UI. KPIs are built over agents and tasks, not individual employees, so no surveillance of people is possible. Conversation history is stored per session, and data export and account deletion are built in. The audit trail is designed for the requirements of revFADP, the EU AI Act, and DORA, giving regulators and auditors verifiable processes.
- What does glass box instead of black box mean?
- A black box only outputs an answer; the path to it stays invisible. A glass box shows the whole path: every agent discloses step by step what it does, from the plan through the chosen model and every tool call to the result. You can inspect the run before you trust it. Every interaction is recorded with model, tokens, cost, and time, auditable, complete, and without reference to individuals. AI rarely fails on technology; it usually fails on trust, and trust comes from visibility. This is exactly the visibility IT and compliance need to approve an agent. The difference from many providers’ black boxes: with AIgent, traceability is built in from the start, not reconstructed afterwards.
- Does AIgent enable employee monitoring?
- No, the architecture rules this out from the start. All KPIs are built over agents and tasks, not over individual people: the dashboard shows what agents accomplish, which model they use, how long they take, and what they cost, not who operated them. Reporting is anonymized, so performance monitoring of individual employees is not possible. This respects employee data protection requirements and simplifies alignment with works councils and data protection officers. At the same time, oversight remains complete: every operation is recorded in the audit trail with model, tokens, cost, and time, and can be reconstructed. At AIgent, transparency applies to the machine, not the human: what is audited is what the agent does, not how much a person works.
- Can I export and delete my data?
- Yes, both are built in. Data export and account deletion are core platform functions, not a special process or a support ticket. You can take your data with you or remove it at any time, keeping control over what stays stored and what gets deleted. This covers data subject rights under revFADP and GDPR, such as the right to data portability and to erasure. What is stored is the conversation history per session plus the audit trail of every interaction with model, tokens, cost, and time. The data resides in Swiss data centers by default or, with on-premise operation, exclusively on your own infrastructure. No lock-in is part of the principle: your data remains yours, even when you leave the platform.
- How do I keep track of all agents?
- Through a central dashboard that shows all agents on the platform together: what they accomplish, which model they use, how long they take, and what they cost. You see the number of runs per agent, average duration, and weekly cost at a glance, one hundred percent of it captured in the audit trail. Steering works through agents and tasks, not people, and reporting is anonymized. IT and business teams see early which agent delivers value, where costs arise, and where to adjust. For auditors, every single operation can be reconstructed down to plan, tool call, and result. Cost transparency and oversight come from the same data, without an extra reporting project.
Sovereign AI. For every discerning company.
On your use-case, in your environment, within a clearly defined scope. See measurable results within a few weeks.
Secure early access. No per-seat fees. No lock-in.