Frequently Asked Questions

Find answers to common questions below.

What is Eucera in one line?

Eucera is an agentic UX platform that ties user intent to outcomes by leveraging real-time context, product knowledge, and actions — acting as the end-user operator for your product and instantly simplifying the experience without requiring you to revamp your UI or backend.

How do Co-Pilot, Task Agents, and Chatterblocks differ?

• Co-Pilot → runs workflows end-to-end from plain-language intents (e.g., “Set up brand settings”).
• Task Agents (AI Functions) → prompt-less features (often a “Generate” button) that return structured outputs such as drafts, summaries, or auto-fill.
• Chatterblocks → in-thread UI components — forms, cards, tables, or charts — to review data, preview changes, and confirm actions inside the conversation.

How do you prevent hallucinations?

We ground every response in Product Knowledge, user context, and validated Actions. A built-in reranker selects the most relevant knowledge before generation. When confidence is low, the agent can escalate via HITL (human-in-the-loop) rather than guessing.

What is HITL and how does it work?

HITL (human-in-the-loop) is built into Conversations: you can review threads, train from feedback, or hand off to support when needed. This closes quality gaps, prevents bad outputs, and improves your agent over time.

Do we need to rewrite our UI or create a dedicated backend?

No. Eucera works with your existing product — reusing your UI components and acting against your existing APIs/workflows. You can deliver agentic experiences without rebuilding your front end or creating a parallel service layer.

How fast can we get to value?

Teams typically launch text-based actions within minutes; a richer experience with a Chatterblock usually takes 2–3 hours. You can ship meaningful outcomes the same day.

How does Eucera handle authentication and permissions?

Eucera honors your app’s existing permission model. The agent always acts as the authenticated user, so it can only access what that user is permitted to see or do. No bypasses of your access controls.

How do you verify user identity?

We use HMAC identity verification between your app and our SDK. Each request is cryptographically signed to the authenticated user, preventing impersonation and ensuring correct on-behalf-of actions.

What about sensitive data?

We provide PII masking/redaction so sensitive data never leaks into prompts, logs, or model inputs. You control exposure and retention. Built for enterprise compliance (aligned with SOC 2) and ready to support GDPR/HIPAA.

Where is data processed and stored?

Data is encrypted in transit and at rest. By default, processing occurs in the United States on GCP; sensitive data is not used for model training. Retention is under your control. Our roadmap includes additional EU/UK GCP regions.

What models do you support?

We support OpenAI, Google Gemini, and private models. Eucera can automatically choose models based on use case to balance quality, cost, and latency.

What is “Product Knowledge” in practice?

Product Knowledge is part of the Eucera brain alongside user context and actions. It connects to docs/help articles and other sources so answers are accurate and consistent. Beyond Q&A, it activates features with best practices baked in, guiding users toward successful outcomes.

How does Eucera interact with our APIs and workflows?

Through Actions that map directly to your APIs/workflows. The agent can create, update, and retrieve data on behalf of the user. With Chatterblocks, users can review info and confirm updates — all inside the conversation.

What about latency and quality trade-offs?

Eucera is optimized for low-latency intent matching. Optional components (e.g., reranker, multi-step tool use) can add ~100–300 ms depending on configuration, giving you control over the speed/precision balance.

Do you support single-page applications (SPAs)?

Yes. We synchronize component state with the agent, enabling Tandem Mode so agent and user share the same context in real time — even in complex, stateful SPAs.

How do we observe and debug agent behavior?

Use the Conversations dashboard for full observability: every query, response, and action is logged. Filter by sentiment, unanswered cases, or workflows. We also track user behavior and provide deep analytics on actions/events to measure adoption, debug edge cases, and improve experiences continually.