An agent-native GTM (Go-To-Market) platform designed to be driven from Claude Code, Codex, or any MCP client via a CLI and MCP server, with a web app for reviewing and approving state. It unifies TAM list building, data enrichment (30+ providers waterfall), multi-channel sequences (email, LinkedIn, WhatsApp), inbox, and CRM into one workspace. You describe the market and personas in plain language, and Oxygen sources accounts, finds roles, waterfalls contact data, scores fit, and produces a deduplicated table you own—all with a spend cap and approval gate before any paid or external write action executes. Pricing: Free workspace with 1,000 starter credits; paid plans reportedly start at $99/mo (per third-party, official full price list not surfaced). Every paid run requires approved: true and --max-credits; previews are free. Caveats: very early-stage (a Swiss GTM engineer hit a bug during setup), and the product is built for developers/agents rather than marketers seeking a visual dashboard.
Pros & Cons:
Pros: The agent-native CLI/MCP design is the core differentiator—you can build a TAM list, enrich it, and launch sequences entirely from Claude Code or Codex without switching to a browser, which fits the workflow of technical GTM teams; the unified workspace means a reply on email stops the LinkedIn step automatically, eliminating the ID-sync glue you’d otherwise maintain across Apollo + lemlist + Instantly + Clay; the approval and spend-cap gate is unusually rigorous—paid calls refuse to run without explicit approval and a credit ceiling, and previews are free, which addresses the “agent spends money without asking” problem; the free tier (1,000 credits) allows real evaluation without payment.
Cons: It is very early-stage—a hands-on reviewer hit a bug during setup, and the product is still being polished; pricing is not fully transparent—the official site does not surface a complete price table in search results, and the $99/mo figure comes from a third-party source; it is built for developers/agents, so marketers expecting a visual, click-through dashboard (like Apollo or Clay) will find the CLI-first workflow disorienting; credit consumption on data providers can scale, and while gates exist, you must actively set ceilings and approve actions to avoid surprises.
Who is it for?:Technical founders, GTM engineers, and RevOps teams that already use Claude Code or Codex and want to build TAM lists, enrich data, and run outbound sequences from the terminal without switching tools; teams that value unified approval rules and a single credit balance across lists, enrichment, and sequences; users willing to tolerate early-stage bugs in exchange for an agent-native workflow. Not suited for non-technical marketers seeking a visual dashboard (Apollo/Clay are more direct), teams requiring extensive independent user reviews before adopting, or anyone who needs a fully transparent published price list before a demo.