FAQ & Troubleshooting

Operanix is an Agent Operating System for enterprise teams. It is not a chatbot — it is a full platform for deploying, managing, and governing AI agents that work across your business.

Each agent specializes in a business domain (Sales, Support, Marketing, Finance, etc.) and is powered by your company's own knowledge, governed by compliance policies, and measured by continuous evaluations.

ChatGPT is a general-purpose chat interface. Operanix is purpose-built for enterprise agent management. Key differences:

• Company-specific knowledge — Agents are trained on your website, documents, and data, not generic internet knowledge
• Compliance & governance — Every response passes through a trust scoring engine with PII detection, content policies, and approval workflows
• Deployment — Agents can be deployed to your website as a chat widget, connected to Slack, or triggered via API
• Measurement — Continuous evaluations measure agent quality across 7 dimensions with actionable improvement suggestions

No. Everything in Operanix is configurable through a visual interface — agent setup, knowledge management, governance policies, workflow automation, and deployment.

The REST API (115+ endpoints) is available for teams that want programmatic access, but it is entirely optional.

30 minutes for your first agent: sign up, crawl your website, install an agent from the Marketplace, and start chatting.

7 days for a full deployment: configure all agents, assign knowledge, set up governance policies, run evaluations, and deploy to production channels.

Operanix ships with 11 pre-configured agent roles, each specializing in a business domain:

• Sales — Prospect research, outreach, deal analysis
• Support — Ticket triage, resolution, escalation
• Marketing — Campaigns, content, audience targeting
• CEO — Strategic briefings, cross-department synthesis
• Operations — Process optimization, vendor management
• Finance — Expense analysis, forecasting, budgets
• Legal — Contract review, compliance checks
• HR — Policy lookup, onboarding, employee Q&A
• Product — Feature prioritization, feedback synthesis
• Research — Market research, competitive analysis
• Strategy — Strategic planning, trend analysis

Yes. Navigate to Agent Workforce, select the agent, and open the Prompt tab. You can edit the system prompt, add few-shot examples, define response format requirements, and set tone/voice guidelines.

Go to Agent Readiness and click Run Evaluation. The evaluation framework scores your agent across 7 dimensions. A pass rate of 70% or higher indicates the agent is ready for production deployment.

Follow this debugging workflow:

1. Check the agent's Knowledge tab — make sure the relevant knowledge sources are assigned
2. Run an evaluation from Agent Readiness to identify which question categories are failing
3. Review the failed eval cases to understand what knowledge is missing
4. Add the missing knowledge (crawl additional pages, upload documents, or add manual entries)
5. Re-run the evaluation to confirm improvement

Any public website URL. Enter the URL in Knowledge Ops → Sources → Add Source → Web Crawl. The crawler will follow links within the domain and extract content from each page.

Operanix accepts the following file formats for knowledge upload:

• Documents: PDF, DOCX, PPTX, TXT
• Markup: MD, HTML
• Data: CSV

Every entity and document chunk passes through a compliance gate before being stored. The gate checks for:

• PII detection — Identifies and flags personal identifiable information (names, emails, phone numbers, SSNs)
• Harmful content — Screens for inappropriate, dangerous, or legally problematic content
• Policy violations — Checks against your custom governance policies

Flagged content is quarantined for human review rather than being silently discarded.

Pricing pages often use JavaScript rendering to display dynamic content (sliders, toggles, custom calculations). Standard crawlers may see an empty page.

Operanix uses Firecrawl with a 5-second wait for JavaScript rendering and priority URL seeding to handle this. If pricing is still missing, try uploading a PDF or screenshot of your pricing page as a document upload.

Yes. Operanix uses a multi-tenant architecture with Firestore tenant-scoped collections. All data is stored under tenants/{tenantId}/ paths. Cross-tenant reads are architecturally impossible — every query is scoped to your tenant ID at the database level.

Yes. Operanix supports SSO via:

• Google Workspace (OAuth 2.0)
• Microsoft Entra ID (OAuth 2.0 / OIDC)
• Okta (SAML 2.0 / OIDC)
• OneLogin (SAML 2.0)
• Any generic SAML 2.0 or OpenID Connect provider

Configure SSO in Settings → Authentication → SSO.

Operanix is built on SOC 2 compliant architecture with immutable audit logging, encryption at rest and in transit, RBAC, and tenant isolation. Formal SOC 2 Type II certification is in progress.

All data is stored on Google Cloud Platform in US regions. This includes Firestore (application data), Cloud Storage (documents and embeddings), and Secret Manager (credentials).

Yes. Navigate to Settings → Data Export to download your data. Exports include agents, knowledge, conversations, evaluations, and audit logs in JSON format.

Yes. Operanix offers a 14-day free trial with no credit card required. You get full access to all features during the trial period.

Yes. There are no long-term contracts. You can cancel your subscription at any time from Settings → Billing. Your data is retained for 30 days after cancellation in case you want to reactivate.

Operanix uses a progressive approach to usage limits:

• At 80%: Usage meters in the dashboard show a warning indicator
• At 100%: Features are throttled (slower response times) but not hard-blocked
• You will receive email and in-app notifications at both thresholds with upgrade options

This means no evaluation has been run yet. Navigate to Agent Readiness, select the agent, and click Run Evaluation. The evaluation takes 1–3 minutes depending on the number of test cases.

The trust score requires data from multiple sources. To populate it:

1. Run evaluations from Agent Readiness (feeds Agent Quality dimension)
2. Create governance policies in the Governance Center (feeds Governance Score)
3. Crawl your website in Knowledge Ops (feeds Knowledge Score)

The trust score will calculate automatically once data is available from at least 3 of the 5 dimensions.

Check these common causes:

• Embed code placement — The script tag must be placed before the closing </body> tag
• Tenant ID — Verify the data-tenant attribute matches your tenant ID in Settings
• Agent deployment — The agent specified in data-agent must be deployed to production (check Agent Workforce → Deployment tab)
• Domain allowlist — If you have domain restrictions enabled, add your website domain in Settings → Chat Widget

Some websites block automated crawlers or load content entirely via JavaScript. Try these alternatives:

• Upload documents — Save web pages as PDF and upload them in Knowledge Ops → Sources → File Upload
• Check robots.txt — The target site may disallow crawling in its robots.txt file
• Try specific URLs — Instead of crawling the entire site, add specific high-value page URLs as priority seeds

Response latency depends on the configured LLM model. Check and adjust in Agent Workforce → select agent → Identity tab:

• Fastest: Gemini Flash, GPT-4o mini — under 2 seconds typical
• Balanced: Gemini Pro, GPT-4o — 2–5 seconds typical
• Highest quality: Claude Opus, GPT-4 — 5–15 seconds typical

Also check if the agent has an unusually large knowledge base bound — more knowledge means more retrieval time. Consider narrowing the knowledge scope to the most relevant sources.