You open Digipals expecting another chat app, and the first thing that happens is the chat opens back at you. A widget surfaces a Thai place equidistant from three friends, a Saturday window everyone has open, and a tappable hold on a 7:30 table. The thread has done the planning before anyone has typed a sentence. That, more or less, is the pitch.
Digipals, a San Francisco company in Y Combinator's Fall 2025 batch, is building what its founders describe as an AI-native group chat app where ambient agents coordinate in-real-life hangouts [Y Combinator, 2025]. The product reads location, calendar, and photo signals to suggest times, propose venues, hold reservations, and split the bill afterward [Y Combinator, 2025]. The waitlist is live at digipals.app; the company has not launched publicly [digipals.app, 2025].
The wedge inside the thread
The interesting move here is positional. Most consumer AI is being shipped as a standalone assistant you open on purpose. Digipals is trying to embed the agent inside the surface where friend groups already negotiate plans, the group chat itself, and let it volunteer suggestions as widgets rather than wait to be summoned [digipals.app, 2025]. The company frames this as the social operating system for the AI era, with the explicit goal of pushing people from digital threads into IRL meetups [Y Combinator, 2025].
The bet is that the friction of group coordination, the eight-message back-and-forth about whether Tuesday works and whether anyone has been to the new ramen spot, is the actual tax keeping hangouts from happening. Remove the tax, the theory goes, and frequency goes up. It is a thesis that has animated a long line of social products, from Evite to Partiful, but Digipals is the first batch of YC founders trying to answer it with an agent that does the legwork instead of a form the host fills out.
Two founders, a small team, a big category
The company is led by Peggy Wang, founder and CEO, who previously co-founded ego in YC's Winter 2024 batch and worked as an AI engineer at Meta's Reality Labs after studying CS and AI at Stanford [X, 2026]. Her co-founder Mathew Matakovic is CTO, with a background in web gaming and a degree from Purdue [Crunchbase, 2026] [LinkedIn, 2026]. Headcount sits at three [Y Combinator, 2025].
Funding is a standard YC seed, undisclosed in size, with Y Combinator as the named backer [Y Combinator, 2025]. No other investors, customers, or usage figures have been disclosed.
| Person | Role | Prior |
|---|---|---|
| Peggy Wang | Founder, CEO | ego (YC W24); AI Engineer, Meta Reality Labs; Stanford CS/AI |
| Mathew Matakovic | Co-founder, CTO | 5+ years web gaming; Purdue |
The honest counterfactual
Consumer social is the genre with the highest mortality rate in software, and an AI layer does not change the underlying physics of how a network gets to critical mass. The risks are worth naming plainly:
- Cold-start gravity. A group chat app is only useful when your group is on it. Digipals will need to convince entire friend clusters to migrate from iMessage, WhatsApp, or Discord, not just individuals to sign up [Y Combinator, 2025].
- Trust surface. The agent's value depends on access to calendar, location, and photos, the most sensitive data on a phone. Persuading users, and their friends, to grant that on a pre-launch product is a real lift [Y Combinator, 2025].
- Suggestion quality. Restaurant recommendations, reservation availability, and bill splitting are each well-served by incumbent apps. The agent has to be meaningfully better than tapping Resy, not merely present [digipals.app, 2025].
The most plausible answer the team can offer is that the bundle itself is the product. No single existing app does plan-to-photo-memories end to end inside the conversation, and if the agent feels closer to a competent friend than to a chatbot, the social proof inside one cluster could carry it to the next.
What the next year tests
The milestones to watch are mundane and decisive. Does the waitlist convert into active groups, not active users. Do those groups hang out more often after installing it, measured against their own baseline. Does anyone outside the founders' immediate network invite a second cluster. YC Demo Day in early 2026 will be the first public read on whether the early cohorts behave the way the thesis requires.
Which is, in the end, the question Digipals is actually asking. If the friction of planning is the thing keeping us indoors and on our phones, what does friendship look like when a small piece of software quietly removes it.
Sources
- [Y Combinator, 2025] Digipals: The future of Social in the age of AI | https://www.ycombinator.com/companies/digipals
- [digipals.app, 2025] Digipals, Building the future of Social in the age of AI | https://www.digipals.app/
- [Crunchbase, 2026] Mathew Matakovic, Co-founder and CTO at Digipals | https://www.crunchbase.com/person/mathew-matakovic
- [LinkedIn, 2026] Mathew M., Digipals (YC F25) | https://www.linkedin.com/in/justatree/
- [X, 2026] Peggy Wang (@peggy_wang) | https://x.com/peggy_wang