Pipeout is a B2B sales lead discovery and outreach automation tool. Officially, it is said to help developers and teams find potential customers suited to their product, and use AI to complete scoring, contact lookup, and personalized outreach. According to the official description, the entire process is divided into just three steps, and is fully automated throughout.
Purpose Pipeout is aimed at scenarios that require proactive outreach for customer acquisition. It first looks for companies that match the product and show buying intent, scores each company from 1 to 10 based on the ideal customer profile (ICP), and places the most suitable companies first. It then finds decision-makers within the target companies, verifies contact information, and prioritizes leads. Finally, it automatically executes LinkedIn and email outreach, including sending connection requests, running follow-up sequences, and pausing when someone replies.
Main capabilities In terms of discovery and scoring, it is officially said to look for buying signals, filter by size and industry, remove duplicates, and score companies so that the best matches are at the top. In terms of contact discovery, AI finds decision-makers within a company, verifies contact information, and prioritizes the best leads; it can display name, title, and email, sorted from highest to lowest score. In terms of model selection, users can choose OpenAI, Claude, or Gemini, use their own API keys, and select models per workspace. In terms of outreach, the tool sends LinkedIn connection requests and emails, runs follow-up sequences, and pauses campaigns when the other party replies; officially, it is said to use human-like pacing, comply with rate limits, and allow users to adjust timing for each campaign. In terms of accounts and security, LinkedIn access is session-based, and API keys are stored in an OpenBao-backed vault; keys are encrypted and stored separately per workspace, and officially, it is said that keys are never seen in plaintext; each workspace is fully isolated at the database level, with leads, campaigns, settings, and API keys kept separate from one another. In terms of agent-oriented automation, Pipeout comes with a CLI that can be driven directly by AI agents. Officially, it is said that the entire pipeline can be run with simple commands: find companies, find decision-makers, and launch outreach, without clicking in the interface. Agents make the decisions, the API handles execution at scale, and thousands of leads can be run on low-cost APIs instead of browsing web pages one by one and consuming tokens. Each command provides stable JSON output and exit codes, making it easy for agents to branch based on results and report back; workspace-scoped JWT authentication supports automatic re-login, which is convenient for unattended operation.
How to use Officially, theprocess given is: first apply to join the invite-only test; applications are manually reviewed, and after approval an invitation code is received by email, which is used to create an account and set up the first workspace within a few minutes. After that, features such as company scoring, contact discovery, lead and campaign viewing, and automated outreach can be used. For automation scenarios, its CLI commands can be used to drive the entire pipeline. When using outreach, you can choose OpenAI, Claude, or Gemini and bring your own API key.
Applicable scenarios and limitations Pipeout is suitable for teams that want to conduct B2B outreach via LinkedIn and email, and want to operate at scale with the help of AI agents or a CLI. Current limitations include: the product is in an invite-only beta stage, requiring application and waiting for review; officially, it is said to be free during beta with full features open, pricing details will be announced before the official release, and no official price is currently provided; no credit card is required to use the beta. Officially, it is emphasized that outreach pauses on replies and complies with rate limits, but no specific deliverability, conversion rate, or performance data is promised; this information is not currently provided. Users need to bring their own API keys for the selected model (OpenAI GPT, Anthropic Claude, Google Gemini).
