[Home](/) / Platform The platform # The Human in the Loop Platform Built for AI Teams Most AI agents do not fail because the model is weak. They fail at the step that needs a person — a review, a call, a judgement call with a name on it. A human in the loop platform exists to make that step callable, accountable, and fast enough to live inside an automated workflow. [What it does](#what) [How teams use it](#how) [Who it is for](#who) [Why a platform](#why) [Get started](#start) ## What the platform does hitl.ph turns human work into infrastructure. When an agent reaches a task only a person can finish, it sends one API request. The brief lands on a managed operations floor, a named specialist owns the outcome, and a structured result comes back — by poll or webhook — so the agent can continue. That is the whole product: not a dashboard your team babysits, and not an anonymous crowd. It is a managed workforce exposed the same way you would call any other service. The five task types the platform accepts map to the moments agents actually get stuck: - **Review a document.** A trained person reads and decides on something a model should not sign off alone. - **Make a phone call.** Confirm a booking, chase a supplier, verify a detail — the jobs that still need a voice. - **Validate ambiguous data.** Resolve the records and judgement calls that models guess at and get wrong. - **Take a photo, check the world.** Confirm something physically exists or eyeball a real-world state and report back. - **Make the call you can’t automate.** An accountable human makes, signs and stands behind the decision — not a confidence score. Each type is a literal `task.type` on the wire. See the [use cases](/use-cases/) for worked examples, or the [API docs](/docs/) for the request shape. ## How teams use the human in the loop platform In practice the integration is three moves. Your agent (or backend) posts a task with a type and a brief. A specialist picks it up end to end inside your working window. The result returns as structured data your workflow can act on. There is no queue for your engineers to triage and no spreadsheet of open exceptions — the human step is a dependency, not a process project. Teams usually place the call at an escalation point: low confidence, high stakes, a real-world action, or a decision that needs accountability. Everything before and after that point stays autonomous. The person is on call for the moments that need them, not sitting inside every loop slowing the system down. Some teams start with a single agent path — document review for a claims bot, phone confirmation for a bookings agent, a final approve/decline on refunds — then expand as more workflows hit the same wall. Others wire the platform into a shared tools layer so every agent in the organisation can escalate the same way. Either path works; the API stays the same. 1. **Call it.** One `POST /tasks` (or CLI) with the type and brief. 2. **A human owns it.** A trained specialist completes the step and is accountable for the result. 3. **It flows back.** Poll `GET /tasks/{id}` or take the webhook payload as structured data. ## Who it is for The buyers are the people building and shipping agents — not students writing essays about the concept. Three groups show up most often. ### AI builders and agencies If you ship agents for clients, you already know the demo breaks when the model needs a human. A callable human step lets you sell reliable workflows instead of promising “someone will look at it.” Agencies use hitl.ph as the shared escalation layer across client projects, with one key and one operations partner behind it. ### Product teams Product and platform teams that own agent features need a production-grade path for exceptions. Building an internal ops desk for every edge case is expensive; leaving those cases to the model is worse. The platform gives product a clear contract: define when to escalate, what context to send, and what structured answer comes back. ### Enterprises Enterprises evaluating a human in the loop company care about accountability, confidentiality, and who actually does the work. hitl.ph is operated by Yoonet — a managed outsourcing company with a delivery base in Balanga, Philippines, and ownership in New Zealand — not a marketplace of anonymous freelancers. Tasks stay with contracted specialists under confidentiality, with named ownership of outcomes. ## Why a platform instead of an ad-hoc ops queue Many teams invent a human step the hard way: a Slack channel, a shared inbox, a contractor on a spreadsheet. That works until volume arrives, someone is on leave, or compliance asks who signed off. A purpose-built human in the loop service treats the same work as infrastructure — versioned task types, SLAs inside your window, structured returns, and an organisation that can scale with demand. “Platform” here does not mean a bloated suite of dashboards. It means a durable interface between autonomous software and accountable people: one API, five task types, a real workforce, and a clear path from trial key to production volume. When you need someone in the seat full-time rather than on demand, the same team helps you [find FTE staff](/pricing/). If you are still mapping the concept itself, start with [what human in the loop means](/human-in-the-loop/). This page is the commercial counterpart — how to buy and run the capability, not only how to define it. ## Ready to add the human step Nothing here is theoretical. hitl.ph runs today as a human in the loop platform for AI teams: your agent sends the task, a named specialist owns it, and the result returns as data your workflow can use. Request access for an API key, read the [docs](/docs/), or skim the [use cases](/use-cases/) to see which task type matches the step your agent cannot finish alone. ## Give your agents a human to call. hitl.ph is onboarding its first agencies now. Request access and the team will send you a key. Or read [the use cases](/use-cases/) and [the docs](/docs/) first. *** Source: https://www.hitl.ph/human-in-the-loop-platform/ This is the markdown representation of that page, served to clients that send `Accept: text/markdown`. Machine-readable index: https://www.hitl.ph/llms.txt · Knowledge bundle: https://www.hitl.ph/okf/index.md