# Olakai — The Enterprise AI ROI and Governance Platform

> Olakai is a vendor-neutral platform that measures AI ROI, governs risk, and controls cost across every AI tool and vendor an enterprise runs. It answers: "What value is our AI creating?", "What is it costing us?", and "What risks is it introducing?" — across two products: Olakai Assistive (workforce copilots and chatbots) and Olakai Agentic (AI coding tools and autonomous agents). This document is the developer surface — SDKs, the CLI, the REST API, and the browser extension used to get AI activity into Olakai.

## Current SDK & CLI Versions

| Tool | Version | Package | Install |
|------|---------|---------|---------|
| TypeScript SDK | 2.6.0 | @olakai/sdk | `npm install @olakai/sdk` |
| Python SDK | 1.6.0 | olakai-sdk | `pip install olakai-sdk` |
| CLI | 0.13.0 | olakai-cli | `npm install -g olakai-cli` |

Verify you're using these versions for compatibility with this documentation.

## For developers: essential tool to speed up the configuration of Olakai - no web UI needed after initial auth.
- [Olakai CLI Tool](https://docs.olakai.ai/docs/md/olakai/olakai-cli): `npm install -g olakai-cli` (v0.13.0)

## Core Concepts

Start here to anchor on the shared vocabulary before diving into a specific product. Olakai organizes AI activity under two products — Olakai Assistive (workforce copilots and chatbots) and Olakai Agentic (AI coding tools / Coding IQ and autonomous agents) — and measures ROI differently for each. Cost basis (per-token API billing vs. subscription/seat billing) is a cross-cutting concern that can attach to either.

- [Core Concepts & Glossary](https://docs.olakai.ai/docs/md/olakai/concepts-glossary): Glossary of platform terms: Agent, Workflow, KPI slot, traffic types, ROI vocabulary
- [How Olakai Measures AI ROI](https://docs.olakai.ai/docs/md/olakai/ai-roi-overview): How Olakai measures AI ROI across Assistive, Agentic, and Coding IQ — value vs. cost, per-token vs. subscription billing
- [AI Data Collection](https://docs.olakai.ai/docs/md/olakai/data-collection-overview): Decision tree: when to use Agentic vs Assistive vs Coding IQ, workflow vs prompt tracking, which collection method per source

## Olakai Agentic — AI coding tools & autonomous agents

Measure ROI, spend, and budgets across AI coding tools and autonomous agents. Two integration paths with very different mechanics:

### AI coding tools (Coding IQ)

For engineering leaders measuring the business ROI and spend of AI coding tools (Claude Code, Cursor, GitHub Copilot, OpenAI Codex, Google Gemini). No SDK — data comes from VCS pull-request analysis (GitHub, Bitbucket, GitLab), per-developer provider admin APIs, and desktop coding-agent hooks captured by the Olakai CLI.

- [Olakai Agentic Quickstart — AI Coding Tools](https://docs.olakai.ai/docs/md/olakai/coding-iq-quickstart): Fastest path to first Coding IQ dashboard populated (VCS + provider admin APIs)
- [Overview](https://docs.olakai.ai/docs/md/olakai/coding-iq-roi-report): AI Equivalent Engineers headline metric, before/after methodology, dollar-value business impact
- [AI Impact Methodology](https://docs.olakai.ai/docs/md/olakai/coding-iq-methodology): How AI tool usage is detected from VCS pull requests (GitHub + Bitbucket), cycle time calculation, adoption cohorts
- [Developer Data](https://docs.olakai.ai/docs/md/olakai/coding-iq-roi-developer-data): Per-developer ROI inputs: which PRs count, qualification rules, opt-out handling
- [Calibration](https://docs.olakai.ai/docs/md/olakai/coding-iq-roi-calibration): Tuning the uplift fraction, fully-loaded cost, weekly productive hours assumptions behind AI Equivalent Engineers
- [Analysis Details](https://docs.olakai.ai/docs/md/olakai/coding-iq-roi-analysis-details): Drill-down view: window sizes, learning-curve buffer, observed-vs-estimated split
- [Track Codex/Cursor/Claude Quickstart](https://docs.olakai.ai/docs/md/olakai/desktop-coding-agents): Track Claude Code, Codex CLI, and Cursor agent-mode sessions via per-turn hooks captured by the Olakai CLI
- [Developer CLI Self-Onboarding](https://docs.olakai.ai/docs/md/olakai/cli-self-onboarding-dev): Individual developer self-onboarding for the Olakai CLI: install, auth, send first event

### Autonomous agents & workflows

For SDK-instrumented AI agents and multi-step workflows (research agents, summarization, document processing) billed per token. Wrap the workflow with the SDK to aggregate tokens and time across steps.

- [Olakai Agentic Quickstart — Agent Governance](https://docs.olakai.ai/docs/md/olakai/agent-iq-quickstart): Fastest path to first agent workflow tracked end-to-end
- [KPIs for Agents](https://docs.olakai.ai/docs/md/olakai/agentic-kpis-guide): Raw Metrics, Metric Slots (Execution Cost, Time Saved, Value Created, Governance Compliance), Composites (ROI), Custom KPIs, and how cost basis differs for per-token vs subscription-billed agentic traffic
- [For AI Agents](https://docs.olakai.ai/docs/md/olakai/agentic-sdk-integration): Wrap your agent workflow with SDK, aggregate tokens and time across multi-step runs

## Olakai Assistive — workforce copilots & chatbots

Measure adoption, time saved, licensing ROI, sensitive-data exposure, and Shadow AI across employee chatbots, copilots, and AI-enabled SaaS (ChatGPT, Copilot, Gemini, Claude, and 850+ more) that respond to individual user prompts.

- [Olakai Assistive Quickstart](https://docs.olakai.ai/docs/md/olakai/assistive-iq-quickstart): Fastest path to first interaction tracked for chat/copilot apps
- [KPIs for Chat/Copilots](https://docs.olakai.ai/docs/md/olakai/assistive-kpis-guide): Adoption, OLA Index, Advanced Analytics time saved, subscription and licensing cost, engagement
- [For Chat Applications](https://docs.olakai.ai/docs/md/olakai/assistive-sdk-integration): Track individual prompts, conversation grouping by userEmail and chatId

### Browser Extension (Shadow AI / web-based capture)

For monitoring AI chat apps used directly in the browser (ChatGPT, Claude.ai, Gemini, etc.) without SDK integration. Installed via Chrome, Edge, or Firefox; optionally deployed at scale via MDM (see docs site for NinjaOne/Tanium runbooks).

- [Overview](https://docs.olakai.ai/docs/md/olakai/apps-monitored-via-extension): Canonical browser-extension overview: what it monitors, depth-of-capture facts, supported AI chat apps
- [Chrome Extension](https://docs.olakai.ai/docs/md/olakai/browser-extension-chrome): Chrome extension install and configuration
- [Microsoft Edge Extension](https://docs.olakai.ai/docs/md/olakai/browser-extension-microsoft-edge): Microsoft Edge extension install and configuration
- [Firefox Extension](https://docs.olakai.ai/docs/md/olakai/browser-extension-firefox): Firefox extension install and configuration
- [Automate Device Mapping](https://docs.olakai.ai/docs/md/olakai/browser-extension-device-map-link): Automating user-to-device mapping for extension deployments

## Detailed API Documentation

Method signatures, parameters, and examples are in the linked docs below.

- [Getting Started](https://docs.olakai.ai/docs/md/olakai/api-sdk-what-you-need): Pre-flight checklist: API key, user identity, which fields are required vs optional
- [TypeScript Client SDK](https://docs.olakai.ai/docs/md/olakai/api-sdk-typescript-client): Full API reference: OlakaiSDK class, wrap(), event(), error handling
- [Python Client SDK](https://docs.olakai.ai/docs/md/olakai/api-sdk-python-client): Full API reference: olakai_config(), olakai(), olakai_report(), olakai_monitor()
- [REST API Reference](https://docs.olakai.ai/docs/md/olakai/api-sdk-rest-api): REST endpoint, authentication, request/response schemas

## Reference

- [Complete KPI Reference](https://docs.olakai.ai/docs/md/olakai/kpis-reference): Metric Slots reference, output contracts, default formulas, override options
- [Formula Reference](https://docs.olakai.ai/docs/md/olakai/formulas): KPI formula language: operators, functions, classifier syntax
- [LLM Integration Patterns](https://docs.olakai.ai/docs/md/olakai/llm-integration-patterns): Common patterns for coding agents wiring up Olakai end-to-end
- [API Parameters Reference](https://docs.olakai.ai/docs/md/olakai/data-architecture): Complete API parameter reference
- [Custom Data Setup](https://docs.olakai.ai/docs/md/olakai/custom-data-setup): Per-agent custom data fields: define the schema you send via the SDK and reference it as variables in KPI formulas
- [Task and Subtasks](https://docs.olakai.ai/docs/md/olakai/task-categories-and-subtasks): Valid task/subtask values for the SDK and REST API
- [Policies](https://docs.olakai.ai/docs/md/olakai/governance-policies): Account-wide Governance Policies: customer-authored detection criteria (name, identifier, description) evaluated against all assistive, agentic, and coding traffic — distinct from employee-facing Acceptable Use Policies
- [Roles & Permissions](https://docs.olakai.ai/docs/md/olakai/roles-and-permissions): The four customer access levels (Employee, User, Analyst, Admin): what each is for, the server-enforced permission matrix, and why KPI tuning sits at Analyst while agents, workflows, integrations and policies stay at Admin

## Full Documentation

- [Complete Documentation](https://docs.olakai.ai/llms-full.txt): All content in single file

## Support

- [Documentation](https://docs.olakai.ai/)
- Email: support@olakai.ai