Question 1

What is Cloptima?

Accepted Answer

Cloptima is an AI governance platform. At its core is a governed, OpenAI- and Anthropic-compatible AI gateway with BYOK, guardrails, budgets, provider failover, exact and semantic response caching, and LLM FinOps. The same platform also covers cloud cost: it analyzes code changes for cost impact, optimizes your data warehouse queries, and attributes costs to teams automatically — so model spend and cloud spend get the same operating discipline.

Question 2

Who is Cloptima for?

Accepted Answer

Cloptima serves engineering teams (AI/ML platform, platform, DevOps, SRE) who want governed model access and actionable optimization, and finance/FinOps teams who need cost visibility and chargeback — across startups, SMBs, and mid-market companies. AI and cloud costs are shaped by the same architecture, model, and query decisions these teams already own.

Question 3

Which model and cloud providers does Cloptima support?

Accepted Answer

For LLM providers, Cloptima supports OpenAI, Anthropic, Google Gemini/Vertex AI, and Amazon Bedrock, with OpenRouter, Mistral, Cohere, Groq, and other Tier 2 providers supported for pricing and telemetry. For cloud cost, Cloptima supports AWS and Google Cloud Platform, with Azure support coming soon. For Kubernetes, we support any distribution — EKS, GKE, AKS, and self-managed clusters.

Question 4

How is Cloptima different from other tools?

Accepted Answer

Cloptima leads with enterprise AI Gateway governance and keeps multi-cloud FinOps in the same platform: (1) A low-latency AI Gateway with real-time prompt guardrails, budgets, provider failover, and exact and semantic caching built directly into the request path. Budgets reserve the worst case across every allowed attempt, so failover cannot push spend past a hard limit. (2) LLM FinOps connecting token spend and agent sessions to business units and product workflows. (3) AI-native tools including an MCP server and CLI to query and manage your cost data. (4) Data warehouse query intelligence catching scan spikes across BigQuery, Snowflake, and Redshift. (5) Automated PR cost impact in CI commenting on infra and workload changes before merge.

Question 5

How long does it take to set up?

Accepted Answer

Under 5 minutes for AI Gateway — point your client's base URL to Cloptima, generate a virtual key, and your requests are governed immediately with zero code changes. For cloud and Kubernetes, connect your AWS/GCP account with a read-only IAM role and deploy the lightweight K8s Helm agent in under 15 minutes.

Question 6

What is the Cloptima AI Gateway?

Accepted Answer

An OpenAI- and Anthropic-compatible gateway that sits between your apps and model providers. Point your existing client at Cloptima's endpoint, use a Cloptima virtual key instead of a provider key, and get budget enforcement, model policy, attribution, and no-retention prompt handling with no application rewrite.

Question 7

Do you support caching LLM responses?

Accepted Answer

Yes, two ways. Exact response cache serves identical, policy-scoped requests without calling the provider. Semantic response cache goes further and matches requests that mean the same thing even when worded differently, gated by similarity score, source freshness, and content class. Both support an observe mode that shows hit rate and savings before you ever enforce anything.

Question 8

How do guardrails work, and do they slow requests down?

Accepted Answer

Built-in credential rules and your own personal-data rules run inside the gateway with no extra network call. Provider scans (Azure AI Content Safety, AWS Bedrock Guardrails, Google Model Armor, or a webhook) add a call, so you can cap their cost per request and skip them when a request is already slow. Every profile change is versioned and audited.

Question 9

Can we apply our negotiated provider contract rates?

Accepted Answer

Yes. Contract price sheets are customer/org-scoped and forward-only, with cached-token and special-token rate rows, a commitment and credit ledger, and a real-data impact preview computed from actual usage before approval. Historical usage is never repriced, and approved rates propagate platform-wide within 24 hours.

Question 10

Do you support prompt caching cost savings from providers directly?

Accepted Answer

Yes. Cloptima observes provider-native prompt cache usage (cache reads, writes, and savings reported by OpenAI, Anthropic, Gemini, and others) automatically, and can recommend prompt-prefix changes to increase cache-hit rates further.

Question 11

Do you offer a trial?

Accepted Answer

Yes. Starter plans include a free trial with no credit card required — AI Starter for 90 days, Cloud Starter and Platform Starter for 14 days. Pro and Growth plans start with checkout.

Question 12

How is Cloptima priced?

Accepted Answer

Three self-serve tracks — AI FinOps, Cloud FinOps, and bundled Platform — each with Starter, Pro, and Growth tiers, plus custom Enterprise. AI and Cloud FinOps start at $49/month and bundled Platform at $89/month; higher tiers add LLM events, accounts, analyzed spend, retention, and governance. Annual billing saves 20%. See the pricing page for current limits.

Question 13

Is Cloptima useful for startups?

Accepted Answer

Yes. Startups often see the strongest ROI because AI and cloud spend directly impact runway. Cloptima stops AI overspend with budgets enforced in the request path, and helps teams catch cost-impacting changes early and automate savings in daily workflows.

Question 14

What happens after the trial?

Accepted Answer

At the end of your trial, you can choose a paid plan to continue intelligence runs and integrations. We send reminders before trial expiry.

Question 15

Does Cloptima store my data?

Accepted Answer

Cloptima stores cost metrics, resource metadata, and aggregated usage data. No-retention prompt policies are available for teams that do not want prompt or response payloads persisted. We do not store your application code, secrets, or the actual data in your databases. All data is encrypted at rest (AES-256) and in transit (TLS 1.3).

Question 16

Is Cloptima SOC 2 compliant?

Accepted Answer

We are currently pursuing SOC 2 Type II certification. Talk to our founders for our current security documentation and compliance roadmap.

Question 17

What access does Cloptima need to my cloud account?

Accepted Answer

Cloptima uses a read-only IAM role. It can read cost/billing data, resource metadata, and CloudWatch/Monitoring metrics. It never has permission to create, modify, or delete any resources in your account.

Question 18

What is the MCP server?

Accepted Answer

MCP (Model Context Protocol) is an open protocol for connecting AI assistants to data sources. Cloptima's MCP server lets you query AI governance and cost data from Claude, ChatGPT, or your own AI agents via natural language — connect directly to our hosted endpoint, or run it locally through the Cloptima CLI.

Question 19

What is the Cloptima CLI?

Accepted Answer

The Cloptima CLI is a terminal tool for querying LLM spend, costs, recommendations, anomalies, budgets, and account health. Install with Homebrew (brew tap cloptima/tap && brew install cloptima-cli) or use Docker. It supports JSON, table, and CSV output for scripts, CI/CD, and AI agents.

Question 20

What integrations does Cloptima support?

Accepted Answer

For AI: the gateway connects to OpenAI, Anthropic, Google Gemini/Vertex AI, and Amazon Bedrock, and guardrail profiles can call Azure AI Content Safety, AWS Bedrock Guardrails, Google Model Armor, or a webhook. For workflows: GitHub (PR cost comments), Slack (alerts), and Jira (ticket creation). BigQuery for query optimization. We're building Redshift, Snowflake, and Databricks integrations.

Question 21

How does the K8s agent work?

Accepted Answer

The Cloptima K8s agent is deployed via Helm chart. It collects resource usage metrics (CPU, memory) and entity metadata (deployments, pods, nodes) with read-only cluster access. It runs as a lightweight pod (~50MB memory, <0.1 CPU).

Question 22

Can I monitor multiple clusters?

Accepted Answer

Yes. Each cluster gets its own agent with a unique API key. All clusters report to a single Cloptima dashboard where you can compare costs, set budgets, and manage teams across clusters.

Question 23

Does it work with Karpenter or Cluster Autoscaler?

Accepted Answer

Yes. Cloptima monitors costs regardless of your autoscaling approach. It actually helps you optimize autoscaler configurations by showing the cost impact of scaling events.

Frequently Asked Questions

General

AI / LLM FinOps

Pricing & Billing

Security & Privacy

Integrations & Tools

Kubernetes

Still Have Questions?