Support
Quick answers to the most common questions about Aqta
AqtaCore sits between AI and high-stakes actions. Before the action executes, it verifies the trigger and leaves a cryptographic attestation that it did. The wire clears, the patient is treated, the contract signs. Your OpenAI-compatible traffic passes through AqtaCore first: policies, spend limits, and safety checks run at the boundary, and every decision is cryptographically logged for cost attribution and governance exports.
No. Loop and runaway detection are one enforcement signal among several. The product is built to answer what serious teams need in production: who spent what, what policy applied, what was blocked or allowed, and what evidence you can show finance, security, or regulators, without rebuilding your stack.
Setup takes less than 5 minutes. Sign up, get your API key, add one line of code to your application, and you're protected. No complex configuration required.
Minimal changes required. Aqta acts as a proxy, you simply route your LLM requests through our gateway by swapping the base URL. Your existing code structure remains unchanged.
All data is encrypted in transit and at rest. We never train models on your data. We are GDPR compliant and our security controls are aligned with SOC 2 Type II Trust Services Criteria. Your prompts and responses are processed in real time and not stored beyond what's needed for threat detection.
Our infrastructure runs in EU data centres (GCP Europe-West1). We never transfer data outside the EU without explicit consent.
Yes. Your dashboard shows blocked and flagged traffic in real time: policy violations, budget stops, prompt-injection and jailbreak attempts, PII risk, rate-limit breaches, and runaway repetition. Each event includes forensics and context so engineering and security share one view.
We combine content and behaviour signals at the gateway: prompt injection and jailbreak patterns, PII and sensitive data risk, abusive or extractive traffic, rate anomalies, and runaway repetition from agents, alongside hard budget and policy rules. Network intelligence (where enabled) helps new patterns propagate without you waiting for a vendor release.
They are one class of production failure (expensive and embarrassing) but not the whole product. Aqta treats repetitive and pathological call patterns as inputs to the same enforcement layer that applies budgets and policies, so you are not relying on a single heuristic or post-hoc charts alone.
We scan prompts and responses for sensitive data (emails, phone numbers, card details, etc.) and automatically redact or block them before they reach your LLM or get logged.
Network intelligence is the framework for sharing anonymised threat patterns across deployments. Starter tier and above can opt in to contribute detected patterns. Propagation to connected deployments is part of the roadmap.
ROI Clarity shows your spend versus savings side-by-side with clear ROI multipliers. See immediately how much value Aqta provides, typically 2–5× return on investment through prevented incidents and optimised routing.
We calculate developer hours saved by preventing incidents that would require debugging. For example, each infinite loop prevented saves roughly 3 hours of debugging time. Configure your hourly rate to see the monetary value.
Instead of simple checkmarks, we provide a 0–100 compliance score with specific improvement suggestions. Track your progress towards major AI governance frameworks (including EU AI Act and GDPR). Pro tier includes PDF report export for audits.
Comparative benchmarks are on the roadmap for Pro tier. The goal is to let you see how your cost efficiency, enforcement rate, and compliance posture compare to similar organisations. Contact us if this is a priority for your procurement.
Yes. Invite team members with role-based access (Admin, Member, Viewer). Use @mentions to notify teammates, set up approval workflows for policy changes, and integrate with Slack for critical alerts. Starter tier includes up to 5 members, Pro tier is unlimited.
Yes. The free tier includes 500 requests per month, forever. Perfect for testing and small projects. No card required.
We charge based on requests processed through Aqta. Starter is €199/month for 50,000 requests. Pro is €499/month for 500,000 requests. Enterprise and Sovereign pricing is available for higher volumes and critical infrastructure use cases. GBP and USD also accepted.
We'll notify you at 80% usage. When your limit is reached, requests return a clear error and stop until you upgrade. No surprise charges. Upgrade your tier at any time from your dashboard to continue.
Yes. No long-term contracts. Cancel anytime from your dashboard. You'll retain access until the end of your billing period.
We support OpenAI, Anthropic (Claude), Google (Gemini), Perplexity (Llama, Sonar), and any OpenAI-compatible API. You can also bring your own API keys (BYOK) for full control.
Aqta adds minimal latency per request. Total end-to-end latency is dominated by the upstream LLM provider response time, not the gateway.
We provide integration examples for Python, Node.js, and cURL. Since Aqta is OpenAI-compatible, you can use any existing OpenAI SDK, just swap the base URL.
Enterprise and Sovereign customers can deploy Aqta in their own infrastructure, including air-gapped and sovereign mesh environments. Contact us for details.
Aqta includes built-in rate limiting per user, per API key, or globally. Configure limits in your dashboard to prevent abuse and control costs.
Our security controls are aligned with SOC 2 Type II Trust Services Criteria. Formal certification is planned. If your procurement requires an in-progress SOC 2 report or a specific timeline, contact us at legal@aqta.ai.
Yes. We're fully GDPR compliant. Data is processed in the EU, we have a DPA available, and we support data deletion requests.
Yes. Enterprise customers receive a Data Processing Agreement (DPA) as part of their contract. Contact us to request one.
Can't find the answer you're looking for? Our team typically responds within a few hours.
Contact Support