Good afternoon, everyone - Happy July. Second half of the year already, huh? Wild. In case you missed it, I wrote a fairly long thought piece on the future of higher education in the US and the downstream impact (I think) it will have on select student-housing towns across the country. If you need a read, you can check it out here:

On the deal front, we have an interesting one from our partners at Litquidity Ventures. This is one of the earliest-stage deals that’s come across the desk. The company is called Domain, and they are building Doma, an AI you add to your group chats the same way you would add a person.

It joins Signal, Telegram, Slack, and MMS today, with phone calls, Zoom, and Google integrations next. No app to download, no dashboard to log into, no prompt to write. Almost every AI product asks the human to go to the AI. Domain inverts that and puts the AI inside the conversations where work already happens - across every chat app at once, added like a person, with one memory that follows the relationship rather than a separate bot per workspace.

The company is raising a $3M SAFE at a $25M post-money valuation cap. Litquidity is participating, and we have already gathered $500K in commitments from early investors we walked through the product. This is a team-and-thesis entry at a price built for the stage, not a revenue-stage round.

Latest from the founders (June 2026): Doma is now in public beta and has grown roughly 10x in the last couple of months, entirely through organic word of mouth.

As always, I’d like to remind you to keep the contents of these emails strictly confidential. Please do not contact the company or affiliated parties as these emails are not meant to be shared with anyone outside of this list. Anyone found doing so will be removed.

If you'd like access to the NDA-gated data room and deal link, you can send me a text or reply to [email protected].

Important: If you received this email, you responded to our survey indicating that you are an Accredited Investor and/or Qualified Purchaser. This memo and its contents are strictly confidential and should only be viewed by express invitation by its author(s). It contains potential investment information that may only be viewed by invitees who qualify as Accredited Investors as defined by the United States Securities and Exchange Commission. If you have received this memo by mistake, or do not meet the qualifications of an Accredited Investor or Qualified Purchaser, please stop reading and disregard this memo immediately.

Domain is building Doma, an AI participant in your group chats. You add it the same way you add a person, and once it is in the chat, it remembers what was said across every conversation, surfaces what matters at the right time, connects people, and follows up on the commitments that get made and forgotten. It is live today on Signal, Telegram, Slack, and MMS, with phone calls, Zoom, and Google integrations on the near-term roadmap.

The bet is structural. Two years of underwhelming ROI AI pilots have shown that many consumers are hesitant to change where they work to use AI. $200B has been spent on AI, and almost nobody has captured the gains: 90% of firms report zero productivity impact (NBER, Feb 2026), and 95% of AI pilots showed no measurable ROI (MIT, 2025).

We still have high conviction in niche, practitioner-led AI products targeting outcome-driven solutions in high-ticket, high-trust industries like finance, law, and healthcare. However, broadly speaking, unconstrained and non-defined AI spend is receiving a bit of pushback lately.

To that end, this team's thesis is that the moat in AI is context, not capability, and that the richest context lives in the private group chats no foundation model can see and no enterprise tool indexes. 3.6 billion people use messaging apps every day. Doma meets that behavior where it already is.

Real early product signal, organic only. Doma launched March 11, 2026 with zero paid spend and is now in public beta, having grown roughly 10x in the last couple of months entirely through organic word of mouth - no app store, no push notifications, no drip campaigns. Users are adding Doma to their own group chats and bringing it into real work, and CAC is effectively $0 because using the product is how it spreads.

Founder-market fit is rare. The five co-founders have built and exited businesses across tech, media, and community for two decades. Dan Gardner co-founded Code and Theory (now part of a publicly traded marketing group with 20,000+ employees) and ON_Discourse, the invite-only network of business leaders that ran for two years and became Domain's training ground. Toby Daniels founded Social Media Week and sold it to PE in 2021. This is the team that has spent the longest time inside the group chats Domain is trying to serve.

The wedge is a real behavior, not a habit shift. Doma does not require a new download, account, login, or change in workflow. It is invited to a chat the same way a person is. The thing that has to be true for Domain to work is already true: that work happens in messages.

Context-as-moat is a credible thesis. Doma builds a separate, evolving context for every relationship and every group, then carries only the right insight forward. The more it is used, the more useful the next conversation becomes. If the team executes, the data graph is exclusive, built from private conversations no incumbent can index.

Capital efficiency is built into the plan. $3M targets 18-24 months of runway at ~$125-167K monthly burn. The use of funds is 45% product, 30% GTM, 15% AI infra, 10% ops. It is a product-led plan, not a hire-everyone round, and it defers enterprise ACV from the base case.

The price respects the stage. A $25M post-money cap on $3M of SAFE is a reasonable entry for a team of this caliber with a product already in users' hands and a believable path to a Series A. The valuation does not require the bull case to clear a strong return.

The category just got validated, in public. The strongest endorsement our thesis has had all year came from Anthropic. They shipped Claude in Slack - an AI you summon with an @mention that holds context per channel and follows up on its own. One of the best-resourced labs in the world just confirmed the core bet: AI belongs inside the room where work happens, not in a separate app you go visit.

It also draws the line cleanly. Claude in Slack is Slack-only, gated to Team and Enterprise plans, with memory scoped per channel by an admin - a feature inside one workspace. Doma is the inverse: one identity across Signal, Telegram, Slack, and iMessage, with context that travels with the person, not the channel. Most real work doesn't happen in one tool; it happens across five, and a Slack-locked product structurally can't follow. The deeper difference is relational - Claude in Slack is a worker you delegate to, Doma is a participant in the conversations where trust and decisions already live: client threads, investor groups, the Signal chats that never go anywhere near a Slack admin. Group creation, intros, onboarding, and proactive follow-through are product primitives for Doma, not edge cases - a surface a single-workspace incumbent can't reach without rebuilding from the messaging layer up.

Foundation models are finally good enough at long-horizon, multi-party reasoning to participate in a group chat without breaking the social contract. The current generation of frontier models can hold context across thousands of turns, attribute statements to speakers, and detect when not to interject. That is the unlock that makes "AI as participant" possible at all.

Messaging is the default work surface, not email. Signal, Telegram, Slack, and WhatsApp have absorbed the conversations that drive business outcomes. Email is where work goes to get archived; chat is where it gets done, and nothing has been built natively for that reality.

The market has rejected AI products that demand new behavior. The buyers who got burned on copilot subscriptions are now asking for AI that fits into work already happening. "Doma joins your existing chat" is exactly the framing this moment rewards. The risk in being early is being too early, and Domain has reduced that risk by launching directly into a behavior that is already universal.

The use cases the team is seeing in the public beta are a useful read on the concept: a founder, their fractional CFO, and Doma running monthly close, with Doma pulling the books, flagging runway changes, and staging the board update before anyone asks; a sales lead and two AEs working a strategic account in a thread where every call note and stakeholder map lives in the chat, so handoffs stop falling through; a hiring manager, recruiter, and panel interviewers in one thread where Doma holds the rubric and surfaces contradictions in feedback before debrief.

Every incumbent, and the new model-native entrants like Anthropic's Claude in Slack, starts inside exactly one messaging surface. A user's actual conversations span Signal, Telegram, Slack, and iMessage at the same time. Cross-surface continuity is the real product, and it is structurally hard for a single-surface incumbent to build.

Best-in-class founder team: Five operators with prior exits across Code and Theory, Social Media Week, ON_Discourse, and two decades of global media product. The team that has lived inside these conversations the longest.

Pure organic momentum: Roughly 10x growth in the last couple of months with zero paid acquisition. Users are bringing Doma into their own conversations and into real work, which is the signal that matters at this stage.

Entry price respects the stage: A $25M post-money cap for a shipped product with organic traction and a credible Series A path. We have seen thinner founder backgrounds raise at higher caps in this category.

The bull case is real: If the context-as-moat thesis works, the comp set is Notion, Slack, and the conversational-AI-as-default-surface category. Those are $10B+ outcomes.

Honest risk posture: Ten weeks of usage is not retention, the incumbents are now moving directly at this (Anthropic shipped Claude in Slack as we opened the round), and privacy is the whole game. The memo addresses each directly.