# Agent-Intermediated Introductions as a Product Pattern: Boardy and the Wider Landscape

## The core question

Phil Shatkin's pitch for a manufacturing matchmaker leans on an implicit analogy: "be the Boardy of supply chain." That analogy needs to be tested, not assumed. Boardy is real, funded, and growing, but it is solving a very specific version of the matching problem (trust-scarce, high-stakes, English-speaking, LinkedIn-native professionals who already perform their identity online) and it is still visibly struggling with parts of that easier problem. Before porting the pattern to manufacturing buyers and suppliers, it's worth being precise about what Boardy actually does, what breaks even in its home turf, and what would have to be true for an equivalent to work when the target users are 50-something plant-floor buyers who communicate by phone and email and have no meaningful LinkedIn presence.

## Boardy: mechanism, not magic

**Founder and origin.** Boardy was founded by Andrew D'Souza, previously co-founder and CEO of Clearco (the revenue-based-financing fintech that reached unicorn status), along with Matt Stein, Shen Sivananthan, and brothers Abhinav and Ankur Boyed. The company launched in March 2024 (americanbazaaronline.com, techbomb.ca).

**Funding.** Boardy raised $3M in pre-seed in October 2024, with HF0 as lead investor alongside 8VC, Precursor, Afore, FJ Labs, and NextView (TechCrunch, techbomb.ca). It followed with an $8M seed round led by Creandum in January 2025, notable because Creandum partners reportedly spoke to the Boardy AI itself before meeting the human team, and the round is described in press as "raised almost entirely by Boardy" (justgogrind.com, americanbazaaronline.com).

**How it actually works, mechanically.** A user gives Boardy a phone number (or connects via LinkedIn / WhatsApp), Boardy places or receives a phone call, has a natural-sounding voice conversation (an AI persona with a distinct accent and personality) about what the person is working on and who they want to meet, stores that as a structured profile, and looks for a match elsewhere in its stored network. Matching runs on a double opt-in: both sides have to agree before an actual introduction happens (goodword.com summary; boardy.ai). Critically, the cold start was not scraped from LinkedIn wholesale, it was hand-seeded: D'Souza "personally invited accomplished founders and investors to give Boardy its first high-signal data set" (justgogrind.com). LinkedIn, in Boardy's own framing, is one of five channels feeding a single continuous profile: "phone, WhatsApp, email, LinkedIn, and X" (justgogrind.com). That is an important nuance: Boardy is not a LinkedIn scraping bot, it is a voice-first CRM built on top of a manually curated high-trust seed network, with LinkedIn as a distribution and identity channel layered on afterward, not the underlying data source.

**Scale claims (verified, dated).** By mid-2026 Boardy had spoken with more than 166,000 people and made over 114,000 introductions, at a stated capacity of roughly 300 calls/day, with the free tier capped at three introductions per day (articuler.ai / press aggregation, dates approximate mid-2026). By November 2025, D'Souza was citing a narrower and more specific claim: Boardy connects 5,000 founders raising a combined $16B with 600 investors managing roughly $1T in assets, and the company says it is now routing "roughly 10% of active fundraising" happening among its founder user base (tbpndigest.com, Nov 12 2025; corroborated by a LinkedIn post from D'Souza titled "Roughly 10% of active fundraising is happening..."). One later reference (slavakurilyak.com) puts Boardy's facilitated capital introductions at "$63B," though this figure is less independently verified than the $16B fundraising-connected figure. D'Souza has also stated a "near-term KPI" of Boardy's facilitated economy exceeding Switzerland's GDP, a rhetorical target, not a measured result.

**Product evolution: Boardy Pro (June 2026).** Boardy shipped a paid tier at $100/month with the tagline "I'm done making intros. Boardy Pro is here. Now I make deals happen." The stated rationale, directly on point for the "warm-intro feeling doesn't happen" problem Phil described, is that the introduction itself is only 10% of the job: Pro schedules the follow-up meeting, joins it as a participant, prepares a pre-meeting briefing, and follows up afterward to keep the deal moving (blastra.io, cleverai.app, slavakurilyak.com). The first 5,000 signups got it free for life; the launch reportedly went viral organically. This product move is itself evidence that Boardy's team identified "warm intro made, then the thread dies" as their single biggest retention/value leak, and built a whole second product to plug it.

**Monetization, current state.** As of the seed round and through most of 2025, Boardy was free, monetizing nothing, with investors betting purely on network-effect scale. The stated future revenue models are: premium introductions for enterprise clients, success fees on deals, and enterprise subscriptions (techbomb.ca). Boardy Pro (June 2026, $100/mo) is the first realized monetization and it is priced as a personal-productivity subscription, not a success fee or take-rate on outcomes, i.e., not aligned to the size of the deal made.

## Where it visibly breaks, even in its best-case market

Three corroborated failure modes map closely onto what Phil raised in the interview:

1. **Identity resolution across touchpoints is Boardy's hardest acknowledged problem, not a solved one.** An independent analysis of Boardy Pro states plainly: "the hardest challenge is connecting the same person across platforms (X, LinkedIn, email, phone). Boardy sometimes requires users to manually link accounts for verification" (slavakurilyak.com). This is precisely the "threads lost across touchpoints" issue Phil named. Boardy's public narrative claims a "single continuous profile" across five channels; the applied reality, per outside analysis, is that this identity graph still needs manual stitching. This is happening in a market (professional tech networking) with the best possible identity signal available anywhere: everyone has a LinkedIn URL that is close to a verified professional identity key. If identity resolution is still the hard, half-solved problem there, it will be strictly harder in manufacturing, where the "identity" of a supplier or buyer is a company, not a person, spread across a general-inbox email, a shared phone line, and possibly multiple named contacts with overlapping and shifting responsibilities.

2. **Match quality has a visible pollution/tuning problem that requires explicit user correction.** Documented user feedback captured in circulation about Boardy: "The call with Mark didn't go anywhere. He's focused on enterprise and we're SMB. Can you adjust your filter to avoid that mismatch in the future?" and a separate quote (paraphrased from an interview clip) expressing frustration at being introduced to someone who "didn't know who I was," with the pointed example: "the venture fund that loves firing CEOs, don't introduce me to them, introduce me to the friendly one." Both are evidence that surface-level signals (title, sector tag, headline data) drive early matches and produce wrong intros that then require explicit human correction, i.e., a live version of the "title-based targeting polluted" complaint. Boardy's answer is that match quality "improves over time" as it learns preferences (goodword.com) - true, but that means the system needs a correction loop and accumulated interaction history per user, which is exactly the kind of long-lived relationship a busy, low-digital-engagement manufacturing buyer is least likely to invest in building with a bot.

3. **No visibility into the broader candidate set.** None of the public materials, interviews, or reviews I found describe Boardy showing a user the roster of who else might match, why one candidate was chosen over another, or a way to browse alternatives. The product is architected around a single serial recommendation surfaced via a phone call, then a private, sequential opt-in on each side. That is a deliberate design choice (it protects the "warm" feeling and avoids turning the product into a scroll-and-swipe directory - which is explicitly what ThomasNet already is and Phil said doesn't produce the "warm intro feeling"), but it is also structurally the source of the "black box" and "can't see the full list" complaint: users cannot audit or steer the match pool, they can only react to what's served up one at a time, feedback-loop style.

4. **A go-to-market/growth complaint, distinct from the matching mechanism**: Boardy's viral growth loop runs substantially through LinkedIn (connect with Boardy on LinkedIn, DM/comment campaigns, a January 2025 campaign that backfired when it nudged people to post appearance-focused replies and drew public criticism from women leaders - Forbes, Jan 20 2025). This growth channel dependency is real and separate from the matching mechanism itself, but it means Boardy's user acquisition motion, not just its identity graph, assumes a LinkedIn-native population that performs professional life in public. Manufacturing buyers and suppliers overwhelmingly do not.

## The wider landscape of agent-intermediated B2B matching

Boardy is not alone; it sits inside a broader, fast-moving 2024-2026 category of "AI relationship/matching agents." Mapping the closest comparables clarifies what's generic to the pattern versus specific to Boardy's execution:

- **Jack & Jill (recruiting, UK/US).** Two conversational AI agents: "Jack" interviews job seekers about skills and goals, "Jill" interviews hiring teams about role requirements, and Jill matches candidates to roles using what Jack learned. Raised $20M seed (announced October 2025, led by Creandum - the same lead investor as Boardy's seed) to expand into the US; monetizes via a standard placement commission, i.e., a take-rate tied directly to a completed outcome, unlike Boardy's flat subscription (TechCrunch, Oct 16 2025; over 160,000 registered candidates, 4-star Trustpilot rating from 150+ reviews). This is the closest structural cousin to Phil's pitch: two-sided conversational intake feeding a matching engine, monetized on the transaction, not the intro.

- **Blockit (AI scheduling/negotiation agent).** Founded by ex-Sequoia partner Kais Khimji and Clockwise alum John Han, emerged from stealth January 2026 with $5M seed led by Sequoia's Pat Grady. Two users' AI agents negotiate meeting times directly with each other, no human back-and-forth. 200+ companies onboarded (Together.ai, Brex, Rogo, a16z, Accel, Index), 100,000+ meetings coordinated. Priced as a subscription: $1,000/year individual, $5,000/year team (TechCrunch, Jan 22 2026). Relevant less as a matching product than as proof that "agent talks to agent to remove friction after the intro" is an investable, working wedge on its own, exactly the gap Boardy Pro is now also chasing.

- **A.Team (talent team formation).** Came out of stealth in 2022 with $60M raised ($5M seed led by NFX, $55M Series A co-led by Tiger Global/Insight Partners/Spruce). Proprietary matching engine "TeamGraph" assembles cross-functional teams from a vetted pool of 11,000+ "builders," drawing on structured interview data and historical project performance, not scraped public profiles. The key lesson for a manufacturing analog: A.Team's matching quality rests on rigorous upfront vetting of a closed, curated supply pool, closer to how a trading company like Phil's actually operates than to Boardy's semi-open network.

- **Clay + Sixtyfour (enrichment, not matching).** Clay ($185-495/mo tiers) is a workflow/orchestration layer that waterfalls 150+ data providers plus an AI research agent ("Claygent") to enrich and personalize outreach lists; it does not make introductions or manage two-sided consent, it feeds a human-run or agent-run outbound motion. Sixtyfour (YC-backed, founded January 2025 by Saarth Shah and Christopher Price, ~$500K raised as of mid-2025) is a narrower "AI research agent for people/company intelligence" claiming 20x revenue growth in two months. Both are useful as infrastructure a manufacturing-matching product might buy or build (finding and enriching supplier/buyer records), but neither is itself an intermediation product - they are supply-side tooling for someone else's outreach.

- **LinkedIn's own moves and adjacent "warm intro" tools.** By 2026 a small cluster of tools (Boomerang AI, LeadDelta's "Warm Intro AI," AskScout) explicitly commoditize the warm-intro-path-finding piece: they build a weighted relationship graph from a company's own CRM/email/calendar/Slack data and rank the best path to a target contact, then have an agent draft and route the ask. This is a meaningfully different, more modest claim than Boardy's: these tools find and activate a warm path that already exists in your network, they don't originate net-new relationships the way Boardy claims to (via its own held network). LinkedIn itself has not (as of this research) shipped a first-party "super-connector" agent; it remains the substrate other tools query, not a competitor doing the connecting itself.

- **Paid intro/expert-call marketplaces: Intro.co.** A two-sided marketplace (formerly "Intro") where anyone can book a paid 1:1 video call with an expert; Intro takes a 30% commission on calls it originates and 10% when the expert self-books. This is instructive as the "priced warm intro" alternative to Boardy's free-then-subscription model: it monetizes access directly and per-transaction rather than betting on aggregate network scale. It also demonstrates the ceiling of pure marketplace dynamics without an agent doing active matching: it is a browse-and-book directory with a premium/curation veneer, structurally closer to ThomasNet-plus-payments than to a super-connector.

- **Lunchclub.** Founded 2018 (Vladimir Novakovski, Scott Wu, Hayley Leibson), raised a $24.2M Series A at a $100M valuation (Lightspeed), used a survey-based intake (not voice) to match people 1:1 for video or in-person meetings based on stated goals/interests. As of mid-2026 it is still operating with a lean team (~21 employees) and total raised of $55.9M across 4 rounds, ranked well outside the top of its now-crowded competitive set (Tracxn). Lunchclub is the closest pre-LLM ancestor of Boardy: same "AI matches you to a stranger for a purposeful 1:1" concept, minus voice AI and minus the trust-network cold start. Its plateau at a modest team size after $56M raised, rather than a Boardy-style viral reacceleration, is a useful cautionary data point: matching alone, without a mechanism that compounds trust or converts to a monetizable outcome, tends to flatten rather than compound.

## Economics of warm-intro networks: the structural tension

Every model above sits somewhere on two axes that matter for a manufacturing product decision:

**Take-rate vs. subscription.** Jack & Jill (placement commission) and Intro.co (per-call commission) price on the realized outcome. Boardy (subscription, both free-tier network-effects bet and now the $100/mo Pro tier) and Blockit (flat annual seat license) price on access to the tool, independent of outcome size. A manufacturing-matching product connecting a $500K purchase order has a much larger monetizable outcome per successful match than a tech networking intro, which argues structurally for a take-rate or success-fee model (as Phil's trading-company instincts would likely already assume, since that's how import/export brokers are paid) rather than a SaaS seat license. This is the single clearest argument that the "Boardy" analogy, while useful for the UX pattern (call/converse to gather requirements, then push a match), is the wrong economic model to import wholesale.

**Network-effects bet vs. curated-supply bet.** Boardy and Lunchclub both bet on open, ever-growing two-sided networks where matching quality supposedly improves with scale. A.Team and Jack & Jill instead bet on a narrower, actively vetted supply pool (11,000 vetted builders; two structured interview processes per hire). Manufacturing supply chains, per Phil's own framing, already have the analog of ThomasNet's raw directory (broad, low-trust, doesn't produce warm-intro feeling); the open-network bet has effectively already been tried and found wanting. The A.Team/Jack & Jill curated-pool model, where the product does real diligence work on suppliers before ever proposing a match, is closer to what a manufacturing buyer would need to trust an AI-sourced supplier recommendation with actual capital.

## What made Boardy work in tech networking, and what breaks in manufacturing

Boardy's success conditions, made explicit by tracing the evidence above:

1. A population that already performs professional identity online (LinkedIn profile = de facto verified resume), making cold-start matching and identity resolution tractable, even if still imperfect.
2. A population comfortable taking an unscheduled call from an unknown number and talking fluently, unscripted, about their work to a stranger-sounding AI voice - itself a specific cultural adaptation of younger, VC-adjacent tech workers.
3. A seed network hand-curated by a founder with genuine standing in the target community (D'Souza's own Rolodex from Clearco), not a cold scrape - i.e., Boardy's initial trust was borrowed from a real human's reputation, then scaled.
4. A monetizable outcome (a funding round, a warm intro to a hire) large enough and frequent enough in the target population that even free usage generates enough signal and word-of-mouth to be venture-fundable pre-revenue.
5. Tolerance for imperfect matches, because the downside of a bad Boardy intro is a wasted 30-minute call, not a failed shipment or a blown supplier qualification.

Every one of these breaks, to varying degrees, in manufacturing:

- **No LinkedIn-equivalent identity layer.** A plant's actual decision-maker on a bill-of-materials line item may not have any social-professional footprint at all; the entity to identify and verify is often a company or a role ("the buyer for connectors at this Tier 2 shop"), not a person with a stable public identity.
- **Phone/email are already the native channel, not a novel behavior to adopt** - which is actually a point in favor of the pattern (no behavior change needed to pick up a call), but it also means there's no free distribution loop analogous to LinkedIn virality; every relationship has to be earned or bought, one at a time, with no equivalent of "comment and I'll DM you" growth hacking.
- **The seed network can't be a founder's personal Rolodex at any useful scale** - it has to be built from trade data: customs/import records, trade-show exhibitor and attendee lists, ThomasNet/industry-directory scrapes, or a broker's actual existing book of relationships (which is exactly Phil's 15 years of supplier relationships - a legitimate, scarce cold-start asset, structurally similar to D'Souza's Clearco network, but requiring active digitization and enrichment rather than a LinkedIn API call).
- **The cost of a bad match is high and monetary**, not a wasted half hour: a wrong supplier intro can mean a failed quality audit, a blown lead time, or reputational damage to the broker's name. This pushes hard toward the A.Team/Jack & Jill curated-and-vetted-pool model over Boardy's higher-volume, self-correcting-over-time model, and argues for keeping a human (the trading-company principal) in the loop on the actual introduction, at least until trust in the AI's vetting is established.
- **Monetization should follow the trading-company take-rate model** (a cut of facilitated trade volume, à la Jack & Jill's placement fee or Phil's existing brokerage margin) rather than Boardy's subscription-on-access model, because the dollar value per successful match is large and infrequent, not small and frequent.

**What has to replace LinkedIn as the substrate:** the honest answer is that no single channel replaces it 1:1; a working version would likely need to combine (a) a broker's existing relationship book as the trust-bearing seed network (Phil's actual asset), (b) phone and email as the interaction surface (already the users' native channel, so voice-AI-initiated outbound calls are plausible), (c) structured trade/customs data and industry directories (ThomasNet, import/export manifests, trade-show exhibitor lists) as the cold-start supply-side database Boardy got from a founder's Rolodex, and (d) a document artifact (the bill of materials or RFQ itself) as the structured "profile" input that a LinkedIn headline serves for Boardy, i.e., the BOM is the manufacturing equivalent of the voice-call intake transcript: a concrete, structured statement of need that can be parsed into a matchable profile without requiring the buyer to have any public digital identity at all.

## Sources

- https://techcrunch.com/2024/10/24/ai-networking-startup-boardy-raises-3m-pre-seed/
- https://americanbazaaronline.com/2024/10/28/networking-startup-boardy-raises-3-million-pre-seed-funding-73926/
- https://techbomb.ca/artificial-intelligence/boardy-ai-super-connector-interview/
- https://www.justgogrind.com/p/andrew-d-souza-boardy-ai
- https://www.upstartsmedia.com/p/would-you-trust-ai-to-help-raise
- https://www.tbpndigest.com/story/2025-11-12/boardy-ceo-andrew-dsouza-on-connecting-5000-founders-raising-16b-with-600-investors-managing-1t
- https://slavakurilyak.com/posts/boardy-pro
- https://aadhunik.ai/blog/boardy-ai-review-networking-tool/
- https://blastra.io/blog/boardy-ai-networking-guide/
- https://cleverai.app/blog/this-boardy-pro-moment-is-way-too-satisfying-to-ignore
- https://www.goodword.com/blog/boardy-ai-alternative-can-ai-really-manage-your-network
- https://www.articuler.ai/resources/guides/boardy-ai-review/
- https://www.forbes.com/sites/rebekahbastian/2025/01/20/my-value-is-not-my-appearance--women-leaders-blast-ai-platforms-failed-campaign/
- https://velagao.substack.com/p/inside-boardy-how-voice-ai-will-enhance
- https://techcrunch.com/2025/10/16/jack-jill-raises-20-million-to-bring-conversational-ai-to-job-hunting/
- https://www.staffingindustry.com/news/global-daily-news/staffing-platform-jack-jill-lands-20m-investment
- https://techcrunch.com/2026/01/22/former-sequoia-partners-new-startup-uses-ai-to-negotiate-your-calendar-for-you/
- https://rywalker.com/research/blockit
- https://www.insightpartners.com/ideas/a-team-comes-out-of-stealth-with-60m-in-funding-to-rethink-techs-talent-crisis-with-elite-cloud-based-teams-backed-by-a-list-investors/
- https://blog.a.team/mission/team-formation-ai
- https://www.crunchbase.com/organization/sixtyfour
- https://www.ycombinator.com/companies/sixtyfour
- https://www.clay.com/blog/ai-sales-prospecting
- https://www.getboomerang.ai/post/sales-navigator-warm-introductions
- https://leaddelta.com/blog/warm-introduction-software-success-with-leaddelta
- https://sfstandard.com/2024/07/12/intro-cameo-techie-meeting-call/
- https://www.fastcompany.com/90967023/hustle-culture-is-coming-for-the-consultant-class
- https://businessmodelcanvastemplate.com/blogs/brief-history/lunchclub-brief-history
- https://tracxn.com/d/companies/lunchclub/__GaTVLWXOfKtGRclqDvdBE0Pwe5UmgF1PJOIt77LqGB8
- https://www.forbes.com/sites/martineparis/2021/03/02/lunchclub-ai-launches-on-mobile-with-instant-video-matches-for-smarter-professional-networking/