DePIN x Networking

Welcome back to DePIN Snacks! Each week, we cover a recent web2 startup fundraise and explore why crypto-based networks will win long-term.

Today’s topic: professional networks.

Last week, Boardy announced a $3m pre-seed for a new type of AI-powered professional network. It works like this:

• You send your phone number to Boardy via a LinkedIn DM.

• Boardy, an AI agent, calls you to chat about your career trajectory and goals for ten minutes—much like a traditional headhunter screening call.

• Boardy facilitates double opt-in intros with other professionals in its network.

This weekend, I used Boardy to source an engineer candidate for one of our portfolio companies. It’s obvious to me that - current limitations notwithstanding - AI agents are the future of how VCs support their portfolio companies.

As an early-stage VC, I send dozens of double-opt in messages every week connecting founders to potential hires, customers, investors, etc. Over the past year, we’ve rolled our own CRM at EV3 to help us do this systematically, which involves:

• 1) ingesting data from Google / LinkedIn / Telegram / Twitter / Otter / Luma…

• 2) cleaning the dataset via de-duplication / enrichment / pruning / etc

• 3) training human-in-the-loop verifiers to manually handle edge cases

• 4) running LLM inferences to query and extract insights from the dataset

• 5) surfacing those insights through simple / fast / familiar user interfaces

We’re not the only VC to invest in building these types of tools - YC built Bookface, Sequoia built Grove, 776 built Cerebro - today, these tools use software-based interfaces that are structurally limited in how they incorporate external context. What does that look like a world of agent-based interfaces?

The first-order startup idea is whitelabeling Boardy: a company that helps VCs create “custom Boardys” to facilitate warm connections within their existing network. VCs want scarce, high-value intros to benefit the top-performing and/or highest-ownership companies in their portfolio. Unlike software-based interfaces, agents like Boardy can incorporate external context (e.g., private notes) to prioritize making connections for companies that can meaningfully drive fund returns—without making other founders feel like they are getting a second-tier experience. Similar custom Boardys can be sold to accelerators, L1 foundations, alumni associations, etc.

This idea has a few critical drawbacks, first and foremost privacy: as a VC, how can I know that the company running my private Boardy won’t farm my network for its own benefit later on, e.g. by introducing companies that I’m not invested in? Maybe by training and hosting my custom Boardy locally using infrastructure like Exo Labs.

The second-order startup idea is Boardy as an investment DAO: instead of launching a product for VCs, why not compete with them directly? The Boardy DAO could charge a large ($1k-$1m) buy-in or onboarding fee that gets invested into what the agent believes to be the most promising startups it meets—i.e., the AI-powered Nouns. Inevitably, the number of people in Boardy-the-DAO’s network will become orders of magnitude larger than the networks of even the most successful individual VC firms.

This idea also has obvious drawbacks, namely that it conflates capital and value: your share of the DAO is determined by capital contributions, rather than how much value your network brings to the community. This creates adverse selection, disincentivizing members whose networks punch above their weight relative to how much money they can contribute. Ultimately, even for AI agents it’s difficult to quantify - and even harder to predict - the long-term value created by any one specific introduction.

The generalizable insight behind Boardy’s early success is that agent-based interfaces - specifically using phone calls to facilitate human connections - lowers the barrier for users to reveal their valuable ‘secrets’: in the initial interaction, because users realize they’re talking to an agent, and in subsequent (human) interactions, because both users know they’ve been mutually qualified by the agent in advance. In other words, the familiarity and intimacy of a phone call UI allows agents like Boardy to collect and act on data that would otherwise never make it onto the internet.

Boardy is initially focused on the professional domain: taking social graph data from LinkedIn, enriches it with knowledge collected from 1-on-1 calls with professionals, and facilitates higher-signal connections than would be possible using LinkedIn data alone. We expect Boardy-like products to emerge in other domains where people face similar challenges but can’t/won’t reveal the specifics of those challenges publicly, e.g.:

• Boardy-for-researchers that takes data from universities/think-tanks/journals, enriches it with knowledge connected from 1-on-1 calls with researchers, and connects them with other researchers focused on adjacent problems.

• Boardy-for-singles that takes Facebook/Instagram/TikTok data, enriches it with knowledge collected from 1-on-1 calls with users, and connects them with other people who are looking for friends/dates/etc in the area.

• Boardy-for-unions that takes data from employer/payroll/HR platforms, enriches it with knowledge connected from 1-on-1 calls with employees, and connects them with other union members facing similar workplace issues.

• Boardy-for-parents that takes data from schools/activities/social media, enriches it with knowledge connected from 1-on-1 calls with parents, and connects them with other families facing similar parenting-related issues.

If you’re working on a Boardy-like agent for other domains, reach out to me on twitter @danconia_crypto or telegram @salgala 🌎️