Flagship Launches iris labs With $20 Million for Personal AI Conversations

Flagship Pioneering launched iris labs on October 6, 2026, with an initial $20 million funding commitment and a first product built around personal conversations. Available through iMessage and RCS, iris chat is pitched as an AI thought partner for relationships, goals and everyday challenges—not simply a tool for getting work done.

Personal support through ordinary messages

The company says iris chat learns about a person’s experiences, priorities and aspirations through repeated exchanges. Its intended uses include spotting meaningful patterns, preparing for important conversations, clarifying priorities and taking concrete steps toward a goal. The experience is designed to adapt as a user’s relationships and circumstances change.

iris labs calls its guiding framework psychological fitness: developing the ability to navigate challenges and build more of what feels meaningful. It says that framework informs both product design and evaluation. The company was conceived more than two years ago and was founded within Flagship Labs.

Armen Mkrtchyan, iris labs’ CEO and co-founder, describes the aim as helping people understand what they are experiencing, find perspective and choose a next step. He emphasizes keeping people in charge of their own lives, framing the AI as support for personal judgment. The product can be accessed at chatwithiris.com, with conversations delivered through the messaging services.

The founding team connects the company directly to Flagship. Mkrtchyan is also an Origination Partner there, while iris labs co-founder and president Rockwell Shah is a Flagship Partner. Flagship founder and CEO Noubar Afeyan is another iris labs co-founder. The team also includes Ayşe Muñiz, a Principal at Flagship.

Afeyan describes iris labs as a platform at the intersection of psychology and intelligent systems, rather than simply an improvement to existing approaches. That ambition follows Flagship’s stated company-building model: creating platform businesses with the potential for multiple products. iris chat is the first product under that broader personal-AI remit; the launch names no additional products.

People aren’t just trying iris chat; they’re coming back to it time and again.

Rockwell Shah, co-founder and president of iris labs

Two kinds of AI oversight

iris labs draws a distinction between emotionally complex conversations and tasks such as coding or games, where success is easier to measure. Its stated challenge is to recognize nuance without reinforcing unproductive patterns. To address that, iris chat combines expert-informed guidance with AI judges—software agents that evaluate and refine other agents’ responses.

A separate safety layer has a different job. The company says each conversation is paired with a parallel AI safety agent that assesses the exchange in real time and guides responses when additional safeguards are warranted. That agent is designed to use conversational context to distinguish genuine concern from ambiguity, rather than apply the same response to every situation.

Return visits are not wellbeing results

Shah presents repeat use as evidence that people want AI for more than productivity. The company also says early evidence suggests users who interact with iris chat’s safety systems are much more likely to continue using the product. It frames that observation as support for the idea that safety and continued engagement can reinforce each other.

Those are engagement claims, not measured improvements in users’ lives. The launch announcement provides no retention rate, underlying figures for the safety-engagement comparison, or independently measured wellbeing outcomes. It describes how the product is intended to support people and how its oversight works, but does not quantify whether those mechanisms deliver the promised benefits.