Ren: An Autonomous AI Agent for Relationship Work

Ren is Syrena’s cross-channel AI assistant for relationship work. It gets to know a person’s goals, network, and history, then takes on the ongoing work of managing professional relationships. With access to Gmail and connected tools, Ren can spot relevant people, suggest warm introductions, and keep conversations moving across email and WhatsApp.

The engineering challenge

Relationship work unfolds over weeks and months. An agent has to remember who matters, why they matter, what was said, what changed, and which next step serves the user’s goal. It must be able to pick up work again without waiting for a fresh prompt, while knowing when a decision belongs to the person it serves.

Harness engineering

We engineered the harness that turns a goal into sustained work. It keeps the agent oriented around what the user is trying to achieve, which relationships matter, what is waiting for a response, and when a next step is due. Ren can wake itself up as new signals or follow-up moments arise, use connected context to reassess the situation, and continue without a new prompt.

The harness carries the same goal and relationship context across the web, email, and WhatsApp. It lets Ren prepare, act, and follow through over time while bringing consequential actions, such as sending an introduction, back to the user for approval.

The relationship memory layer

We integrated a memory layer into Ren’s harness to give it an evolving understanding of the user and the people in their network. It draws on conversations, connected email context, stated goals, and the outcomes of past interactions. It carries forward enduring details such as preferences, relationships, commitments, and working style, then brings the relevant context into focus when Ren needs to decide what to do next.

The memory improves on its own as new exchanges and outcomes arrive. It automatically consolidates what Ren has learned, updates its understanding of each person, and brings the useful context forward without someone manually maintaining a profile. Ren can recall why a connection matters, what has already been promised, how the user prefers to approach someone, and whether an earlier introduction led anywhere.

How it comes together

For a user looking for investors, Ren can connect the goal to existing relationships, identify a promising introduction path, explain why it fits, and prepare the message for approval. Later, the harness can bring Ren back to the conversation at the right moment, with the relationship history and next commitment in view. The complexity lives behind a simple exchange with the user.

The result

Conversations, introductions, and outcomes are part of one continuous piece of work. Ren can carry context from one channel to another, find a useful path through a person’s network, and use the response to decide what needs attention next. The user sees the recommendation and retains control over the consequential step.

The screens below show selected parts of the public product and Ren’s conversations across channels.