Innovation & Tech

Reply.io releases Jason Harness with nine roles and MIT license

Reply.io’s open-source alpha gives AI agents nine sales roles, local campaign memory and human send approvals, while external models and delivery services still handle key tasks.

By Grace Kim

4 min read

Reply.io Launches an Open-Source AI Agent Harness for Sales Outreach
Reply.io Launches an Open-Source AI Agent Harness for Sales OutreachPexels / pixabay

What's News

  • Reply.io opened the complete runtime source on November 1 and announced the alpha publicly on November 17.
  • Jason Harness defines nine roles for one host agent, not necessarily nine separate models.
  • Campaigns, contacts and history remain in a local database, and every message needs human approval before sending.
  • The runtime and extension-system code is available under the MIT license.
  • Reply.io’s launch material provides no independently validated deliverability, qualified-reply or meeting results.

By Grace Kim

Reply.io’s Jason Harness assigns nine sales-outreach roles to one host agent, but every message still requires human approval before an external sending service delivers it.

Reply.io announced the open-source alpha publicly on November 17. Its changelog opened the complete runtime source on November 1, ahead of the broader launch.

The tool turns general-purpose AI agents into a sales-outreach workspace. It gives tools such as Claude Code or OpenAI Codex distinct roles for researching prospects, planning follow-ups and drafting messages. Jason Harness keeps campaign state on the operator’s computer and leaves delivery to an external sending service.

What does Jason Harness actually add?

Jason Harness is separate from Reply.io’s managed Jason AI sales development representative and ships no model of its own. It packages instructions, campaign records and approval steps so an existing agent can resume a campaign across sessions. Reply.io calls the release an “alpha” and says “a person must approve messages before they go out.”

Reply.io describes nine roles covering campaign management, planning, research, writing, personalization, critique, response handling, analytics and deliverability. That designation does not necessarily mean nine different models are running. The roles define jobs for the host agent: a planner can propose a sequence, a researcher can gather context, and a critic can flag an unsupported claim before a draft reaches a human reviewer.

Keeping campaign state outside the chat addresses a practical limit of general-purpose assistants. A seller who starts a fresh conversation still needs to know which pitch a prospect saw, what the prospect said in response and whether a colleague already approved a follow-up.

Reply.io stores campaigns, contacts and history in a local database, with rules and memory in local files. The code for the runtime and extension system is available under the MIT license, according to Reply.io’s open-source documentation.

In plain terms, the harness lets an agent recover the campaign record instead of rebuilding it from scratch.

Does local storage make outreach private?

No. Local storage does not mean everything runs offline. Reply.io says the underlying agent processes requests according to its own provider and plan. Sending runs through Reply.io or another provider connected through an extension.

Connected services “may receive the data they need to perform a task.” A local database therefore is not a blanket privacy guarantee. Organizations evaluating the alpha would need to trace what goes to each service.

Local campaign storage limits where the record sits; it does not remove the external services required to generate and send messages.

Why does human approval still matter?

Putting drafting and research in the same workspace may reduce context switching. It does not remove the cost of reviewing what the AI produces. A message can sound specific while relying on an outdated job title or an unsupported claim about the recipient’s company.

A reviewer also needs to know whether someone opted out, whether a proposed follow-up repeats an earlier touch and whether the send still fits the original campaign brief. Human approval becomes more useful when those facts appear alongside each draft.

The approval gate protects the send only when reviewers can check the record, not just the wording.

Can open-source code make outreach cost-free?

The open-source release lets teams inspect and adapt the harness, but free access to the code is not the same as free outreach. AI subscriptions may have usage limits, while sending accounts or prospect data can carry separate costs.

Reply.io has described the product as an alpha; its launch material provides no independently validated results for deliverability, qualified replies or meetings compared with a conventional workflow.

On the available evidence, the alpha’s cost and performance case is still unproven.

How should teams measure the alpha?

The clearest test is whether the harness can reduce the time spent reconstructing campaign context without increasing the time needed to correct drafts. Teams comparing it with an existing workflow can track three measures: review time per message, how often drafts are rejected for inaccurate claims, and whether replies are passed back to the right person.

Those operating results would say more about the tool’s value than its nine-role designation. Until teams produce them, Jason Harness remains an inspectable alpha rather than proof of better sales performance.

The next milestone is evidence that reviewers spend less time rebuilding context and correcting messages than they do in a conventional outreach workflow.

Source: unite.ai

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,447 articles

Related articles

« Previous article