Trellis (Trellis Tech Inc.)

June 11, 2026

Public disclaimer: This is an independent public teardown based solely on publicly available information retrieved on 2026-06-10 and 2026-06-11. It is research support, not investment, legal, tax, or financial advice. Company-provided statements are labeled as claims, not verified facts. Trellis did not participate in or review this report. Errors and omissions are possible; final decisions remain with the reader.

The Trellis demo does the thing AI-agent decks keep promising: it opens Airbnb, files a $450 damage claim, orders the replacement TV, and dispatches the technician. If that works in production, this is not another guest-inbox copilot — it is the operating layer a property manager might actually rip workflows around. The timing is real too: Trellis has the Y Combinator P26 badge, and Guesty just launched its own agent platform for short-term rentals. The catch is what remains outside the public record: $500k ARR, 35 customers, and "200+ integrations" are still the company's numbers, while the visible connector catalog shows 37. The question is whether Trellis has a sanctioned, autonomous wedge before the incumbents turn agentic ops into a bundled feature. The answer is inside the teardown. Read it before the founder call turns into a priced round.

What The Company Does

Trellis positions itself as an AI-agent operations layer for short-term-rental operators managing roughly 10 to 1,000+ units (company claim). The claimed product surface (company claim, company-controlled pages):

AI inbox and guest comms across Airbnb, Vrbo, WhatsApp, and email — responding, handling requests, escalating only what needs a human.
Autonomous task execution — the homepage demo depicts an agent filing a $450 Airbnb damage claim via browser, ordering a replacement TV, messaging the guest, and scheduling a technician.
Team/ops coordination — cleaning, inspections, maintenance scheduled from live reservations, plus a separate Trellis Mobile field-ops app for workers (tasks, shifts, location).
Integrations — the homepage claims "200+ tools"; the public integrations directory lists 37 connectors across 7 categories.

Business model: B2B SaaS with a demo-led / free-trial motion (company claim). There is no public pricing page (source-backed fact — the pricing URL returns a not-found page). Independently supported facts are limited: Y Combinator batch P26 membership, two confirmed founders, a Delaware legal entity named on its own legal pages, thin mobile-app adoption, and the existence of docs/API/legal infrastructure.

Key Takeaways

Investor reaction
"Genuinely interesting wedge in a real, newly-open category — but every positive signal is the company's own telling, an incumbent already shipped the capability, and the architecture may sit on revocable platform access. I want one outside-the-room proof point before anything more than a call."
Current verdict
hold — first-call worthy, not deep-diligence ready.
Why investors might lean in
a real "why now" for autonomous short-term-rental (STR) operations, independently confirmed by an incumbent's agent-platform launch; operator-founders building for operators; and a claimed cross-stack agent layer that, if it truly completes tasks rather than chats, has optionality beyond messaging tools.
Why investors might pull back
every growth and autonomy signal is company-authored (company claim); independent usage proxies are thin (1 iOS review, ~150 Android downloads per a provider-derived estimate — source-backed fact); a Series-F incumbent (Guesty) already shipped a competing agent platform; and the product demo implies browser automation on Airbnb that the platform's terms prohibit.
Highest-leverage fix
bring one arm's-length customer who confirms billing, runs production tasks through Trellis with logged error/escalation rates, over a sanctioned (API/partner) connection. This single proof point collapses four separate investor objections at once.
Best next move
a single, tightly-scoped first call — gated on that proof point being supplied in advance.

What Makes This Potentially Fundable

The fundable version of Trellis is not "AI guest messaging" — that category is crowded and incumbent-contested. It is the system-of-action thesis: an autonomous agent layer that sits across the operator's entire stack (PMS, OTAs, WhatsApp, accounting, field teams) and actually executes operations — files damage claims, orders replacements, dispatches cleaners, reconciles payouts — rather than drafting replies. If that works in production, Trellis becomes the neutral operations brain that no single PMS vendor can replicate without favoring its own stack, accruing proprietary ops data and switching costs. From there, the same agent architecture could expand into adjacent SMB operations verticals, which is the only credible fund-returning path (inference).

The honest gap: nothing in the public record yet demonstrates that the autonomous layer works outside the company's own demo. The claimed ~$500k ARR, 35 customers, 3,000 listings, and a "customer running ops with no back-office humans" all originate from company-authored text (company claim). The demonstrated category — STR-only operations software — sizes below a typical fund-returning threshold on our labeled bottom-up estimate (inference, low confidence). The wedge is exciting; the evidence is not yet independent.

The Four Holes To Close Before Fundraising

Hole	Investor Fear	What To Bring
Connection-method legitimacy on Airbnb/Vrbo	The product demo shows a browser agent opening Airbnb's Resolution Center while Airbnb's terms prohibit bots and automated access; if access is unsanctioned, the platform can delete the wedge overnight	Signed API/partner agreement or sanctioned PMS pass-through, plus an architecture one-pager
No independent traction proof	~$500k ARR / 35 customers may be pilots or founder-affiliated operators; zero named customers appear anywhere public	Billing export, 2–3 arm's-length reference customers with affiliation disclosure, last-90-day net revenue retention
Autonomy is demo-grade in public evidence	A chat tool dressed as an agent is an explicit dealbreaker for this category	Customer-controlled task and error logs from production, not a founder-narrated demo
No funding or ownership record	Nothing public confirms any round beyond YC batch membership; investors cannot model ownership, dilution, or follow-on math	SAFE/round documents, post-money, available ownership

Decision Snapshot

One-sentence company description: Trellis (Trellis Tech, Inc., a Y Combinator P26 company) claims to provide AI agents that run vacation-rental operations — guest communications, team coordination, and cross-tool task execution — on top of an operator's existing PMS/OTA stack.
Screen: hold
Confidence: medium
Suggested next action: one gated first call; the gate is a pre-supplied arm's-length proof point (billing-verified customer + sanctioned-connection summary + reference call offer).
Why this matters now: the agentic-STR-ops window just opened — and the largest incumbent already entered it. Waiting six months likely means either the proof exists or the wedge is gone.
Investor-readiness diagnosis: first-call worthy on wedge and timing; not diligence-ready because every underwriting input (ARR, retention, autonomy, connection method, integration depth, cap table) is company-controlled or contradicted by independent data.
Best founder use of this report: treat the Claim Reconciliation and Diligence Findings sections as a pre-diligence self-audit, and assemble the proof packet in the Founder Action Plan before the next investor conversation.

IC Disagreement Map

Five partners reviewed the same evidence dossier independently, each through a distinct lens: asymmetric upside (power-law), category timing and market structure (prepared-mind), founder quality and execution (founder-jockey), critical flaws and unsupported claims (risk-reduction), and ownership, incentives, and what compounds over a long hold (long-horizon). All five voted hold at medium confidence — but for different decision-dominant reasons, and the devil's-advocate review disputed the consensus in both directions. That disagreement is preserved here rather than averaged away. Each partner also wrote the strongest contrarian case from their own lens, so the reader sees the best argument against the consensus, not just the consensus.

Power-law partner
holdmedium confidence
The demonstrated short-term-rental-only wedge sizes below a fund-returning outcome (roughly $105M–$490M per year for the US slice, labeled inference), every upside metric is a company claim, and the apparent browser-automation exposure is a clean strikeout risk.
Would flip on
Independent traction proof, a named expansion path beyond short-term rentals, and sanctioned platform access would move them to pursue. Confirmation that the wedge stays STR-only on revocable access would move them to pass.
Prepared-mind partner
holdmedium confidence
The "why now" is real but already contested: the incumbents own both the data and the distribution, no compounding moat has been demonstrated, and every piece of decisive proof is company-controlled.
Would flip on
Sanctioned API or partner status plus customer-controlled proof of autonomy and retention would move them to pursue. Confirmed unsanctioned automation would flip them to pass.
Founder-jockey partner
holdmedium confidence
The founders' identities are confirmed, but every founder-market-fit metric is self-reported, and the roughly seven-person team has no publicly visible technical lead for the hardest technical build.
Would flip on
Independent validation of the founders' prior operating depth plus a named, credible technical lead would move them to pursue. Failure to produce either would move them toward pass.
Risk-reduction partner
holdmedium confidence
The architecture appears unsanctioned and platform-revocable, every load-bearing claim is either company-controlled or contradicted by independent data, and the incumbents are positioned to compress the overlay from above.
Would flip on
A sanctioned connection method to the booking platforms plus independent proof of autonomy and revenue would move them toward pursue. Confirmed unsanctioned automation or revenue that reconciles to insiders and pilots would move them to pass.
Long-horizon partner
holdmedium confidence
There is no independent proof of a compounding moat, value capture is contested by AI shipping natively inside the property-management systems, and the absence of any funding or cap-table record means ownership and follow-on math cannot be modeled.
Would flip on
A verified integration or data moat, funding and ownership records, and staged proof triggers being met would move them to pursue. Continued inability to model ownership would keep them at hold or move them to pass.
Devil's advocate
challenges hold— confidence
Both directions are live. The first-call bar is genuinely met, which arguably supports pursuing to a first call; yet stripping out company claims leaves zero independent positive signal, which arguably supports a clean pass. The unanimous hold may be the least-committed of three defensible reads.
Would flip on
One binary, pre-callable question decides it: can the founder pre-supply a single arm's-length, customer-controlled proof point? If not, the committee should resolve to pass rather than schedule a courtesy call.

Two-round transparency: after the independent first-round memos, the committee ran one bounded evidence-request round (each partner named up to three pieces of evidence that could move their vote; the feasible ones were retrieved from public sources, the rest became the diligence questions below). All five partners then confirmed their hold votes unchanged. Because the final votes and confidence are identical, the committee applied its unanimity check: the agreement is evidential rather than social — each partner gates on a different decision-dominant gap, the contrarian reads above pull in opposite directions, and the second look it forced (re-testing whether platform-terms risk is overweighted vs revenue composition) reordered the first-call agenda without changing the verdict.

Scenario Range

Three scenarios, each with a falsifiable trigger an investor could actually watch. No probabilities — triggers and evidence, not point estimates.

Scenario	What The Company Looks Like In 3-5 Years	Falsifiable Trigger To Watch	Earliest Evidence
Strikeout	Dead or acqui-hired by a PMS vendor; the connector catalog is the only surviving asset. OTA enforcement breaks unsanctioned automation, incumbents bundle equivalent AI into contracts operators already hold, and reported revenue reconciles to pilots and insiders	Airbnb developer-policy tightening or broken-agent chatter; Guesty/Hostaway shipping equivalent autonomy at no extra charge; a billing export contradicting reported ARR	OTA policy changelogs; PMS release notes; the pre-call billing export
Base	A profitable niche STR-ops tool: real but modest revenue, mostly messaging-plus workflows with humans still in the loop, thin overlay pricing under incumbent AI pressure, no second vertical	Net revenue retention flat without expansion revenue; the integration registry turning out mostly Zapier-path/roadmap; no named expansion mechanism by the next raise	First-call task/error logs; the per-connector registry; the next-round deck
Home run	The neutral system-of-action for SMB operations: agents run operations across rival PMS/OTA stacks under sanctioned access, proprietary ops data compounds into reliability no copycat can match, and a second vertical is live	Signed OTA/PMS partner status; customer-controlled autonomy logs at scale; a live second-vertical deployment with paying customers	A partner agreement; arm's-length reference calls; an expansion announcement with verifiable customers

What Investors Will Test

What We Looked For	Current Read	Investor Implication	Founder Prep Priority
A falsifiable path from the STR wedge to a fund-scale outcome	Not stated anywhere public; STR-only slice sizes small (`inference`)	Without a named expansion mechanism, the ceiling caps conviction	High — articulate the wedge-to-platform thesis with an early signal
Evidence gathered outside the company's own telling	Only YC membership and thin app-store metrics are independent; all traction is company-authored	Traction cannot be underwritten as-is	Critical — billing export + arm's-length references
Why this category is winnable now	Enabling shift is real (incumbent launched agent platform) but contested	Timing argument works; durability argument doesn't yet	High — name the copy-resistant asset
The single disqualifying fact, if any	Conditional: browser automation vs Airbnb's anti-automation terms — method unconfirmed	A revocable-access architecture is a clean pass for many funds	Critical — document sanctioned connection method
Founder-market fit and execution depth	Identities confirmed; prior operating metrics self-reported; no visible technical lead	Jockey thesis is narrative until validated	High — references + named technical owner
Incentive alignment and value capture	Overlay economics vs PMS-native AI untested; no pricing public	Risk of margin compression to a thin overlay	Medium — win/loss and pricing evidence
Compounding advantage over years	"200+ integrations" unverified (37 in catalog); no demonstrated data/switching-cost asset	Moat is asserted, not evidenced	Medium — per-connector depth registry

Claim Reconciliation: Inconsistencies Investors Will Catch

Claim Or Metric	Where It Appears	Conflicting / Unreconciled Versions	Why Investors Flag It	How To Reconcile
Integration count	Homepage	Public integrations directory: 37 connectors in 7 categories	A ~5x gap on a moat-defining metric erodes trust in every other number	Publish a per-connector registry: first-party live vs Zapier-path vs roadmap, with owners
Product ratings	Site structured data	Apple: 1 review (5.0); Google Play: 0 reviews, ~150 downloads (estimate)	Schema markup that search engines display must match checkable platforms	Disclose the review source; remove or re-source the markup
Team size	YC profile	LinkedIn: 7 employees, 2–10 band	Minor, but signals loose data hygiene when other numbers are contested	Confirm current headcount once, use consistently
OTA action method	Homepage demo	Airbnb terms prohibit automated access; DPA lists a browser-automation subprocessor	Implies the flagship workflow may run on unsanctioned automation	State the connection method and partner status explicitly
Founding year	LinkedIn	A startup-directory snippet shows 2025 (unverified)	Trivial to fix; inconsistent basics invite deeper skepticism	Use the incorporation date everywhere
Prior-venture scale (Xenia)	LinkedIn company feed	AI Tinkerers judge bio: "scaled it to €3M ARR"	The jockey thesis leans on this number; figure drift across venues is a credibility tell on a claim with no independent trace	Use one checkable Xenia figure everywhere, with portfolio data or third-party references behind it

Diligence Findings: Issues To Fix Before You Raise

2 deal-breaker5 high3 medium

Flagship Airbnb workflow may run on browser automation that Airbnb's terms prohibit; no partner status is public
deal-breaker (conditional)
technical / legal
Fix
Document sanctioned connection method per channel; obtain or disclose partner status
All traction metrics ($500k ARR, 35 customers, 15% WoW) are company-authored with zero named customers anywhere public
deal-breaker (if unresolved)
commercial / financial
Fix
Billing export; named arm's-length references with affiliation disclosure
Potential customer-affiliation risk: founders previously operated in STR; reported customer base could include affiliated operators
high
financial / KPI
Fix
Customer roster with affiliation disclosure; references investors can call without founders present
"200+ integrations" vs 37-connector public catalog
high
commercial / technical
Fix
Per-connector registry (live vs Zapier vs roadmap); correct public copy
Autonomy evidence is demo-grade; no production task/error data public
high
technical
Fix
Customer-controlled task logs with error/escalation rates
Site schema rating (4.8/24) conflicts with app-store reality (1 review / 0 reviews)
high
KPI / commercial
Fix
Re-source or remove the markup; publish review provenance
No visible technical lead; 0 posted jobs; ~7-person team carrying an autonomy-heavy roadmap
high
team
Fix
Name technical leadership; show eng/integration hiring plan
No public pricing; monetization motion opaque
medium
commercial
Fix
Price sheet or ACV band, even if sales-led
No SOC 2 attestation; DPA contains only an audit-option clause while handling guest PII, worker location, and payment-adjacent actions
medium
legal / security
Fix
SOC 2 roadmap with date; guardrail documentation
Value capture vs PMS-native AI (Guesty Agent Hub, Hostaway AI) untested in any public win/loss evidence
medium
commercial
Fix
Win/loss notes vs incumbent AI; paid-substitution example

Data Room Readiness Checklist

Procurement-readiness score: not ready
Rationale: a DPA, subprocessor list, privacy policy, and data-deletion process exist publicly (good baseline for an SMB-sale motion), but there is no SOC 2 or equivalent attestation, no pen-test summary, no AI-governance/guardrail documentation, and app-store privacy labels are empty — gatekeepers at larger property-management buyers will stop there.

Data Room Area	What Investors Expect	Current Read	Status	Priority
Company overview & corporate	One-pager, incorporation, good standing, structure	Delaware entity named on legal pages; no registry pull public	partial	medium
Financials	Monthly P&L (18-24 mo), 3-yr model with assumptions, burn/runway	Nothing public; ARR claim unaudited	missing	critical
Cap table & funding history	Clean cap table, prior rounds, SAFEs/notes, 409A	No public funding record beyond YC batch	missing	critical
Legal & IP	Bylaws, board consents, IP assignments, material contracts	Public ToS/privacy/DPA/API terms exist; IP assignments unknown	partial	medium
Product & technology	Roadmap, architecture overview, security/compliance docs	Docs hub exists; architecture and connection method undisclosed	partial	critical
Team	Org chart, key employment/advisor agreements, vesting	Two founders confirmed; ~5–7 staff; no technical lead visible	partial	high
Customers & traction	Retention cohorts, ARR/MRR bridge, pipeline, 3-5 references	Zero named customers public; claims only	missing	critical
Market & competition	TAM/SAM/SOM, competitive landscape, pricing	Category exists but no defensible AI-ops segment sizing; no pricing	partial	high

First-Call Agenda For The Startup

Time	Topic	Founder Goal	Evidence To Bring
0–10 min	Connection method per channel (Airbnb, Vrbo, WhatsApp, PMS)	Remove the conditional deal-breaker	Architecture one-pager; partner/API agreements
10–25 min	Traction quality: paying vs pilot vs affiliated	Convert ARR claim into verifiable revenue	Billing export; customer roster with affiliation disclosure
25–40 min	Production autonomy walk-through	Prove task completion, not chat	One customer's task/error log; live workflow demo
40–50 min	Integration depth and moat	Reconcile 200+ vs 37 and show what compounds	Per-connector registry; ops-data accrual story
50–60 min	Team, round, and use of funds	Show the build is staffed and the round is modelable	Technical-lead intro; cap table summary; hiring plan

Business Model And Revenue Signals

B2B SaaS, demo-led, free trial / no credit card
company-controlled pagemedium
SourceTrellis homepage
Motion stated, not observed
~$500k ARR / 35 customers / 3,000 listings / ~15% WoW
company-authored YC textlow
SourceY Combinator directory YC launch page
Core revenue claim, unvalidated
Design-partner program recruiting operators
company-controlled pagemedium
SourceTrellis about page
Early-GTM signal

Revenue and profitability are not publicly knowable for Trellis. No scenario estimate is offered here because no pricing anchor exists; any ARR-quality judgment must wait for a billing export.

Revenue Quality Checklist

Before this moves from hold toward pursue, the company should be ready to show: a billing/Stripe export tying out to the claimed ARR; paying-vs-pilot-vs-affiliated customer mix with affiliation disclosure; gross and net revenue retention for the last two quarters; churn and expansion by cohort; CAC and payback if any paid acquisition exists; and per-unit pricing. Investors will rebuild CAC, LTV, payback, GRR, NRR, and churn from raw data — these numbers must tie out across any deck, model, and source documents.

Funding And Ownership Context

Item	Public Read	Evidence Label	Diligence Request
Total raised / rounds	Y Combinator batch P26 membership only	source-backed fact (membership)	Round history with amounts
Instruments (equity / SAFE / notes)	Unknown; YC standard deal is the only inferable baseline	inference	SAFE/note terms
Investors	Y Combinator (batch membership); others unknown	source-backed fact / unknown	Investor list
Post-money valuation	Unknown	unknown	Latest post-money
Cap table / ownership	Unknown	unknown	Cap table summary
Burn & runway	Unknown	unknown	Monthly burn, runway months
Next-round plan	Unknown	unknown	Target raise, milestones it buys

Crunchbase was checked and showed no matching Trellis profile. This absence carries moderate weight for ownership modeling since startup databases are often incomplete for very early companies.

Founder And Team

Lodovico Benvenuti
Co-founder, CEO; previously built an STR property-management software product (Xenia) per company narrative
Y Combinator directory LinkedIn X GitHub
Evidencesource-backed fact (identity); company claim (background metrics)
Confidencehigh / low
Jan Sahagun
Co-founder, COO; ~$2M STR sales in under two years per company narrative
Y Combinator directory LinkedIn STR Tech Report
Evidencesource-backed fact (identity); company claim (background metrics)
Confidencehigh / low
Team size ~5–7, San Francisco
Execution capacity
Y Combinator directory LinkedIn company page
Evidencesource-backed fact (band)
Confidencemedium

Founder Competency Coverage

Each cell is exactly evidenced (independent public trace exists), claimed (founder/company narrative only), absent, or unknown — never a score. This deliberately separates what the public record proves from what the founders say.

Domain depth (STR operations)
Lodovico Benvenuti (CEO)
claimed
Jan Sahagun (COO)
claimed
What The Evidence Is
Prior STR software (Xenia) and ~$2M STR sales are company narrative; a Milan accommodation brand named Xenia verifiably exists (Instagram), but the founding role and outcomes are not independently confirmed
Technical build capability
Lodovico Benvenuti (CEO)
evidenced (with caveat)
Jan Sahagun (COO)
absent
What The Evidence Is
Public GitHub footprint with 10 repositories (GitHub): 6 forks (mostly educational AI repos) and 4 original hobby-scale projects, the most recent a TypeScript dashboard updated in 2026 — real coding activity, not agent-scale engineering; prior employer Conduit verifiably exists as a YC "AI agents for hospitality" company (Y Combinator), employment itself self-reported; a targeted search found no public technical artifact for Jan (neutral for a commercial founder)
Product
Lodovico Benvenuti (CEO)
claimed
Jan Sahagun (COO)
claimed
What The Evidence Is
Shipped breadth (site, docs, API, two app stores) is evidenced; product judgment behind it is narrative
GTM / sales
Lodovico Benvenuti (CEO)
absent
Jan Sahagun (COO)
claimed
What The Evidence Is
Jan's ~$2M STR sales and prior commercial roles (a real company named Boom appears in public traces) are self-reported; no independent customer or revenue trace for either founder
Leadership / hiring
Lodovico Benvenuti (CEO)
unknown
Jan Sahagun (COO)
unknown
What The Evidence Is
~5-7 person team; zero public job posts; no third-party leadership trace
Fundraising history
Lodovico Benvenuti (CEO)
unknown
Jan Sahagun (COO)
unknown
What The Evidence Is
No funding record beyond YC batch membership exists publicly
Prior founding outcomes
Lodovico Benvenuti (CEO)
claimed
Jan Sahagun (COO)
claimed
What The Evidence Is
No registry, press, or acquisition record independently confirms any prior venture outcome for either founder

Autonomous agent reliability (production task completion at low error rates)
critical
Team coverage today
uncovered — no publicly evidenced agent-systems experience on the team
What would close it
Named technical lead with verifiable agent/systems history
Browser automation at platform scale (the flagship Airbnb workflow)
critical
Team coverage today
uncovered — CEO's GitHub footprint exists but depth on this build is unverified
What would close it
Architecture walkthrough + the engineer who owns it
Integration breadth maintenance (37+ live connectors)
high
Team coverage today
partially covered — shipped catalog is evidenced; maintenance ownership is not
What would close it
Per-connector registry with named owners

The pattern matters more than any single cell: the two builds the company's entire thesis depends on have no publicly evidenced owner. That is the founder-jockey partner's decision-dominant gap — and the contrarian read (textbook operator + engineer pairing with real shipping velocity) is exactly what the first call should test.

Public Professional Footprint

Supporting signals only — these checks of public technical artifacts, professional social presence, and education/credential traces carry low decision weight by design. They corroborate or weaken the coverage table above and sharpen founder-call questions; none of them alone changes the verdict.

Technical artifacts (repo level)
Lodovico Benvenuti (CEO)
evidenced
Jan Sahagun (COO)
absent
What The Public Record Shows
CEO's GitHub: 6 forks (educational AI repos plus one STR-domain CRM fork) and 4 original projects, most recently a TypeScript dashboard updated in 2026 — hobby/educational scale, no production agent system; a targeted search found no GitHub footprint for the COO
Professional social content
Lodovico Benvenuti (CEO)
evidenced
Jan Sahagun (COO)
evidenced
What The Public Record Shows
CEO: AI-engineering posts and a judge listing at an AI hackathon in Paris (bio text self-submitted), plus a self-published origin-story post anchoring his Milan hosting start at August 2021; COO: STR-industry media presence — a podcast episode and industry-event appearances — and original posts on the company's LinkedIn with real domain substance: a "100+ property-manager conversations in 2 months" discovery-cadence claim and operator visits naming checkable industry references. Post content is the founders' own narrative; notably, the "150-unit company with no back-office humans" claim originates in the COO's own podcast-announcement post — it must be validated by a customer, not by repetition
Education / credentials
Lodovico Benvenuti (CEO)
evidenced (HEC Paris); claimed (Cambridge)
Jan Sahagun (COO)
claimed (McGill)
What The Public Record Shows
An independent institutional page lists the CEO as HEC Paris on an awarded 2023 data-science team; his Cambridge claim and the COO's McGill degree appear only in self-controlled profiles and weak aggregator echoes
Publications / patents / certifications
Lodovico Benvenuti (CEO)
absent
Jan Sahagun (COO)
claimed (self-reported certificates)
What The Public Record Shows
No papers, patents, or independent certifications found for either founder

Net read: the footprint is consistent with the founders' stated roles (a coding CEO, a commercial COO) and adds one genuine verification (HEC Paris) plus named industry references worth calling on the first call — but it also surfaced the prior-venture figure drift now listed under Claim Reconciliation and confirmed that the headline autonomy claim originates in the founders' own posts. None of this substitutes for the arm's-length proof points the verdict is gated on.

Founder-Market Fit Read

What is promising: both founders' narratives are STR-native (operating software and direct rental sales), which matches the buyer's daily reality; shipping breadth for a ~7-person team (web product, mobile apps, docs, API, legal pages within the batch window) suggests velocity.
What is missing: every founder-market-fit metric is self-reported; one founder's public profile has an empty experience section and the other's is minimal; no third-party reference, press, or verifiable portfolio data confirms prior operating depth or outcomes (a podcast episode telling the "AI agents running a 150-unit company" story exists in an independent podcast directory, but the episode content is founder narrative); and no technical leader is visible for the hardest part of the build.
What to prepare: two or three third-party references from property managers who used the founders' prior products or services; verifiable portfolio numbers; a named technical owner.

Trellis Tech, Inc. (Delaware)
Operating entity per its own legal pages
Trellis Terms Privacy Policy
Confidencemedium
Follow-upRegistry confirmation
Xenia (prior STR venture per company narrative)
Founder's prior product/operation; a Milan accommodation brand named Xenia exists publicly
YC launch narrative Instagram: Xenia Milano
Evidencecompany claim (role/outcomes); source-backed fact (brand exists)
Confidencelow
Follow-upConfirm founding role and outcomes; disclose any customer overlap with Trellis's reported ARR
Conduit (CEO's claimed prior employer)
YC W24 "AI Agents for Hospitality" company — also a competitor in the agent-native segment
Y Combinator: Conduit
Evidencesource-backed fact (company exists); company claim (employment)
Confidencemedium
Follow-upConfirm role and tenure; note the competitive overlap
Trellis Mobile apps (bundle ID com.trellis.mobile)
Company's field-ops apps
Apple App Store Google Play
Evidencesource-backed fact
Confidencehigh
Follow-up—
Trellis AI (healthcare), Trellis Connect, Trellis Data, Gen Digital's Trellis
Unrelated homonyms — do not conflate
public web checks (Google Search)
Evidencesource-backed fact
Confidencehigh
Follow-upNone

Traction

Customer Status Table

No customer is named in any public source found in this research. The claimed base resolves as follows:

35 customers (unnamed, aggregate)
Company-authored text on the YC directory
No proofClaim onlyUnknownUnknownUnknownlow
"One customer running ops with no back-office humans" (unnamed)
Company-authored text on the YC launch page
No proofClaim onlyUnknownUnknownUnknownlow
Design partners (unnamed, recruiting)
Company about page
No proofClaim onlylikely unpaidUnknownUnknownlow

Traction Signals

YC batch P26 selection
source-backed facthigh
SourceY Combinator directory
None — selection signal only, not traction
Apple App Store: 5.0 from 1 review
source-backed facthigh
SourceApple App Store
Whether web product adoption (the core) is larger
Google Play: 0 reviews; ~150 downloads (provider-derived estimate)
source-backed factmedium
SourceGoogle Play
Same; the public Play page shows a bucketed download count
X footprint: @usetrellis has 51 followers and 5 posts (account created April 2026); the YC launch post drew ~37.4k views and 206 likes on YC's account
source-backed fact (platform metrics)medium
SourceX: @usetrellis YC launch post on X
Whether any distribution channel exists beyond the YC halo; launch engagement accrued to YC's 1.6M-follower account, not Trellis's

Product Hunt, G2, Capterra, and trade press searches were checked and showed no matching Trellis listing or coverage beyond Y Combinator pages. These absences carry little weight for early-stage companies still in a direct-sales motion.

Hiring And Org Momentum

Posted jobs — YC profile
source-backed fact0 open roles
SourceY Combinator directory
No public hiring velocity into an autonomy-heavy roadmap
Headcount
source-backed fact (band)5 (YC) vs 7 (LinkedIn), 2–10 band
SourceY Combinator directory LinkedIn
Very small team; consistent with pre-seed
Office
2261 Market St, San Francisco
SourceLinkedIn
US-first GTM

Indeed, Glassdoor, LinkedIn Jobs, and the Trellis careers page were checked and showed no matching listing. These absences carry little weight since job boards are often incomplete, and employee sentiment remains unavailable.

Role-mix read: no public postings exist on the checked job boards to interpret; the absence itself is the signal — either the round hasn't closed, or hiring runs through private channels.
Momentum read: unknown (no public hiring data on primary job boards); product-shipping breadth within the batch suggests velocity, but organizational growth is not visible.

Traction Quality Read

Traction Dimension	Current Status	Good Enough For First Call?	Needed For Deep Diligence
Revenue	claim-only ($500k ARR)	yes, as a claim to test	Billing export tying out
Customers	claim-only (35, unnamed)	yes, as a claim to test	Named, arm's-length references with affiliation disclosure
Autonomy in production	claim-only + demo	yes, as the core call topic	Customer-controlled task/error logs
Retention	unknown	n/a	Cohort GRR/NRR
Distribution	thin (no marketplace, SEO, or launch presence found)	yes, with the question why	Channel strategy and early channel proof

Competitive Landscape

Segment	Examples	Customer Alternative	Pressure On Company
Incumbent PMS with native AI	Guesty (Agent Hub), Hostaway AI, Hospitable AI	"My PMS already does this" — bundled into existing contracts	Highest: distribution + data owned by the PMS; overlay margin compresses
AI guest-messaging specialists	HostBuddy, Besty, Conduit (hostai.app), Enso Connect, Duve	Point tool for the inbox slice	High in messaging; lower in cross-tool task execution
Ops/turnover platforms	Breezeway, Turno	Cleaning/maintenance coordination without agents	Partial overlap on team-coordination slice
Human alternatives	VAs, call centers, manual inbox + templates	Trusted human judgment	Persistent at small scale; the labor line Trellis claims to replace

Category Visibility Snapshot

Distribution is one of the fatal-flaw factors this teardown tests, so the visibility claim is made reproducible rather than asserted: below are the exact category queries run on 2026-06-10, with every company that appeared in the captured page-one results (8–10 organic results per query as served). Names are listed in ranked order, in our own words. Visibility is not market share — the read is labeled inference, low confidence. Any reader can re-run these queries and check the claim in 30 seconds.

Google US"AI guest messaging vacation rental"

Captured page-one results (2026-06-10): 1. HostBuddy AI · 2. BoringHost · 3. RedAwning · 4. Breezeway · 5. Aeve · 6. Rentals United · 7. Guesty · 8. Escapia · 9. Hostfully
Was the company present?: No — absent from all captured results.

Google GB"AI guest messaging vacation rental"

Captured page-one results (2026-06-10): 1. Smoobu · 2. Hospitable · 3. BoringHost · 4. Aeve · 5. HostBuddy AI · 6. RedAwning · 7. Guesty · 8. NowiStay · 9. WelcomeScreen · 10. Rentals United
Was the company present?: No — absent from all captured results.

Google US"vacation rental AI agent automation"

Captured page-one results (2026-06-10): 1. Guesty · 2. RentalReady · 3. Aeve · 4. Conduit AI · 5. VirtualWorkforce.ai · 6. Enso Connect · 7. YouTube (category explainer) · 8. Hostaway · 9. PriceLabs
Was the company present?: No — absent from all captured results. Note: Conduit, the agent-native YC company, ranks on page one for the agent-flavored query.

Google US"AI agent vacation rental property management"

Captured page-one results (2026-06-10): 1. Guesty · 2. RentalReady · 3. Hostaway · 4. Aeve · 5. Y Combinator directory (ProhostAI — a different YC STR-AI company) · 6. Truvi · 7. Rent Responsibly · 8. YouTube (category explainer)
Was the company present?: No — and notably, when a YC page ranks for this query, it is another YC company's, not Trellis's.

ChatGPT"best AI tool for vacation rental guest messaging and operations"

Captured page-one results (2026-06-10): Hostaway (overall pick) · OnRepeat (AI-first automation) · Guesty (enterprise) · Hospitable (small hosts) · Conduit and ShortRental.ai (emerging AI-native). Assistant answers are generated text, not a ranking — listed here only as a visibility signal.
Was the company present?: No — absent from the answer entirely, including the "emerging AI-native" category where it would naturally belong.

Google Search (US and GB) and ChatGPT were checked for core category queries and Trellis did not appear in captured results. These absences carry little weight for early-stage distribution but are consistent with thin public funnel evidence.

Visibility read (inference, low confidence): across four category queries in two geographies and one AI-assistant answer, Trellis appeared zero times while at least 15 distinct alternatives did — including direct agent-native competitors. This is a one-day snapshot of visibility, not market share, but it is consistent with the rest of the distribution evidence (no marketplace listings, no review-platform profiles, no press): whatever demand Trellis is converting today, it is not arriving through the public category funnel.

Pricing And Competitive Benchmark

Alternative	What It Offers	Public Price Signal	Price Vs. This Company	Evidence Label
Guesty AI (within PMS)	Native AI suite inside the dominant PMS	Tiered PMS pricing; not public per-feature	Not comparable — Trellis has no public price	company claim
Hostaway AI (within PMS)	AI replies, unified inbox, task management	Quote/demo-led	Same	company claim
Hospitable	PMS + AI messaging bundle	Not on AI page	Same	company claim
Besty AI	"AI property manager" messaging	Demo/contact	Same	company claim
Enso Connect	Guest experience + AI chat + upsells	Pricing page exists; tiers not captured	Same	company claim
Breezeway	Ops, property care, messaging	Quote-led	Same	company claim
Turno	Turnover/cleaning automation	Self-serve pricing exists	Same	company claim

Price positioning read: the entire category is quote/demo-led; Trellis's missing pricing page is normal for the segment but removes the one chance to anchor value publicly. No independent benchmark is currently possible.
Price claims to correct or substantiate: none found — Trellis makes no public competitor-price comparisons (nothing to correct), but it also offers no value anchor of its own.

Competitive Wedge

What could become defensible (inference): cross-stack neutrality (agents acting across rival PMSs, OTAs, and field tools — structurally awkward for any single PMS to copy), an accruing corpus of operational decisions/error data that improves agent reliability, and switching costs once an operator wires payroll-replacing workflows through Trellis. What still looks copyable: AI inbox and message drafting (already commodity), task scheduling from reservations (PMS-native), and any integration breadth that turns out to be Zapier-path rather than first-party. The public record currently demonstrates the copyable parts and only asserts the defensible parts.

Risks And Open Questions

Risk	Severity	Evidence	What To Ask
Unsanctioned OTA automation that the platforms can revoke	critical (conditional)	Airbnb terms prohibit automated access (`source-backed fact`); demo + subprocessor imply browser automation (`inference`)	Connection method and partner status per channel
Traction claims fail independent verification	critical	All metrics company-authored on the YC directory; zero named customers (`company claim`)	Billing export; arm's-length references
Customer base partly founder-affiliated	high	Founders' prior STR operations per the YC launch narrative; unverifiable roster (`unknown`)	Affiliation disclosure per customer
Incumbent bundling compresses value capture	high	Guesty Agent Hub launched; Hostaway/Hospitable market native AI (`source-backed fact` / `company claim`)	Win/loss vs native PMS AI
Autonomy liability (agent errors with guest money/PII)	medium	ToS shifts AI-output liability to customers (`company claim`); no SOC 2 attestation in the DPA	Guardrails: human-in-loop, spend caps, kill-switch, incident runbooks
WhatsApp/messaging compliance	medium	Meta requires opt-in, 24h/template rules, human escalation (`source-backed fact`)	Compliance design for automated messaging
Key-person and team-depth risk	medium	~7 people on LinkedIn, no visible technical lead, 0 posted jobs on the YC directory (`source-backed fact`)	Technical ownership and hiring plan

Pre-Mortem: The Most Likely Obituary

Written as if the company is dead five years from now — the single most probable causal chain, not a list of generic risks. This is the discipline check on the bull case, and the founder's earliest-warning checklist.

Cause of death (one sentence): Airbnb revoked operator-delegated automated access during a platform-wide anti-bot enforcement wave, severing the highest-value action surface overnight, while incumbent PMS vendors had already bundled "good-enough" AI into contracts operators were never going to leave.
The causal chain (3-5 steps from today to the shutdown): (1) the company scales on browser-automation access to OTA surfaces without sanctioned partner status; (2) Guesty/Hostaway ship native agent features inside existing contracts, compressing the price to a thin overlay fee; (3) an OTA enforcement change breaks the agents' most valuable workflows for weeks and churn spikes among arm's-length customers; (4) the round needed to rebuild on sanctioned APIs cannot be raised because retention data now shows the dependence; (5) the team sells the connector catalog in an acqui-hire to a PMS vendor.
The earliest observable warning sign: support or changelog chatter about broken Airbnb actions, or an OTA developer-policy tightening — visible before any churn shows in metrics.
The question that defuses this chain today: "Show me the partner agreement or the exact API scope for every OTA action your agents take."

Decision-Critical Unknowns

Unknown	Why It Is Decision-Critical	Best Evidence	Decision Effect
OTA connection method and partner status	Single cleanest disqualifier	Partner agreements; architecture diagram	Unsanctioned access with no path to it means pass
Independent ARR/retention and customer mix	Underwrites everything	Billing export; references; NRR	Revenue reconciling to insiders and pilots reduces toward pass
Production autonomy with error rates	Category-defining test	Customer-controlled logs	Demo-only autonomy means pass
Live integration depth (vs 200+ claim)	Moat substance	Per-connector registry	A mostly Zapier-path or roadmap registry reduces conviction
Pricing/ACV and value capture vs PMS-native AI	Monetization durability	Price sheet; win/loss	Losing head-to-head against PMS-native AI reduces conviction
Cap table and round status	Ownership/follow-on math	SAFE/round docs	Unmodelable ownership keeps the verdict at hold
True professionally-managed unit count	Market-size sensitivity (±30%+)	Independent census data	A confirmed sub-scale ceiling reduces conviction

Diligence Questions

First Call

Question	Why It Matters	Good Evidence
How do Airbnb/Vrbo actions connect — official API, PMS pass-through, or browser automation?	Revocable access kills the wedge; cleanest disqualifier	Signed API/partner agreement; architecture one-pager
Provide independent proof of ~$500k ARR and the paying-vs-pilot-vs-affiliated mix	Core traction is entirely company-authored	Billing export; 2–3 arm's-length reference calls
Which customer runs fully autonomous ops, doing what tasks, at what error/escalation rate?	Separates task completion from chat	Customer-controlled task and error logs
Reconcile "200+ integrations" with the 37-connector catalog	Tests the moat claim and overall credibility	Per-connector registry with depth and owners
Who owns the technical architecture, and what is their track record?	Hardest build in the category; no visible CTO	Named technical lead with relevant history
Walk through each founder's prior STR operating scale — and reconcile the Xenia figure (~$2M revenue vs "€3M ARR" across public bios)	Jockey thesis rests on unverified narrative; figure drift is a credibility tell	Verifiable portfolio data; third-party references; one consistent checkable number
Disclose pricing/ACV and how value capture survives PMS-native AI	Monetization opacity + incumbent compression	Price sheet; win/loss vs Guesty/Hostaway

Follow-Up

Question	Why It Matters	Good Evidence
Architecture diagram + any Airbnb/Vrbo/PMS partner agreements	Confirms connection-method legitimacy in writing	Signed agreements; sanctioned API scopes
Billing export + reference intros + cohort retention	Converts claims into underwritable data	Stripe/billing data; arm's-length references; GRR/NRR
Review-source breakdown for the 4.8/24 site rating	Resolves the ratings inconsistency	Verifiable per-platform review provenance
Cap table, round size, post-money, ownership available	Models dilution and follow-on	SAFE/round documents
SOC 2 status + autonomous-action guardrail configuration	Guest PII + payment-adjacent autonomy	Attestation or dated roadmap; kill-switch/spend-cap config
Vrbo automation terms + Airbnb software-partner policy posture	Completes the platform-risk picture	Partner-status confirmation per channel
What is the concrete path from STR ops to a larger platform?	Fund-returning ceiling	Named expansion thesis with an early signal

Kill Criteria

Kill Criterion	Evidence That Would Trigger It
OTA actions run via unsanctioned browser automation with no partner path	Confirmed UI automation against platform terms; no written partner strategy
Reported customers/ARR are mostly founder-affiliated or unpaid pilots	Affiliation disclosure or billing export contradicting the claim
Autonomy is demo-only with no production task completion	No customer-controlled logs; human approves every action
Incumbents bundle equivalent AI into existing contracts and operators churn to it	Win/loss data showing native PMS AI substitution at lower price
No technical owner and no engineering hiring into an autonomy-heavy roadmap	Continued absence after direct questioning

Double-Down Criteria

Double-Down Criterion	Evidence That Would Justify More Diligence
Billing-verified ARR with positive net revenue retention	Export ties out; NRR cohort positive over 90+ days
Sanctioned OTA/PMS access	Signed partner/API agreements per channel
Customer-controlled autonomy proof at low error rates	Task logs from an arm's-length operator
Named, credible technical lead plus eng hiring	Verifiable background; open reqs
Falsifiable expansion thesis beyond STR with an early signal	Named mechanism; first cross-vertical proof
Clean cap table with ownership room	Round docs showing modelable entry

Founder Action Plan

Timeframe	Action	Output
Pre-call (now)	Document the OTA/PMS connection method and any partner agreements	One-page architecture + sanctioned-access proof
Pre-call (now)	Pull a billing export; identify 2–3 arm's-length reference customers with affiliation disclosure	Verifiable ARR + reference list
First call	Walk through one production customer's autonomous task log with error rates	Customer-controlled autonomy evidence
Week 1–2	Publish a per-connector integration registry (live vs Zapier vs roadmap) and align homepage copy	Reconciled integration claim
Week 1–2	Name the technical lead; share the engineering/integration hiring plan	Team-completeness evidence
Data room	Cap table, round size, post-money; SOC 2 status; guardrail configuration (human-in-loop, spend caps, kill-switch)	Modelable ownership + security posture

Decision

Screen: hold
Confidence: medium
Rationale: the wedge and timing are real — independently confirmed by an incumbent's agent-platform launch — and the founders and YC selection are verified. But every underwriting input (revenue, retention, autonomy, connection method, integration depth, ownership) is company-controlled, contradicted by independent data, or absent. This earns a structured first call, not deep diligence.
What would move this to pursue: one arm's-length, customer-controlled proof point (billing-verified paying customer + production task/error log + sanctioned connection), plus a named expansion path beyond STR and a modelable cap table.
What would move this to pass: confirmed unsanctioned browser automation with no partner path; traction reconciling to insiders/pilots; demo-only autonomy; or incumbent AI substitution evidence.
Recommended next step: a single gated first call. The gate: the founder pre-supplies the proof packet above. If it cannot be pre-supplied, resolve to pass rather than scheduling a courtesy call.
Founder preparation standard: assume every number on your website and YC page will be cross-checked against app stores, your own catalog pages, platform terms, and independent databases — because in this teardown, they were.

How We Would Miss This One

The expensive error in venture is the miss, not the loss — so here is the falsifiable case that this hold is wrong, with a date attached.

The miss scenario: the reported traction is real and arm's-length, the OTA connection turns out to be sanctioned or sanctionable, and the team quietly proves agent-run operations at production error rates before any institutional money marks the round. In that world, this teardown's hold was the moment the cheapest ownership in the asset was available — the category's why-now was already independently proven by the incumbent's own launch, and waiting for proof also waited for the repricing.
Flip conditions: billing-verified arm's-length ARR, customer-controlled production-autonomy logs, sanctioned OTA/PMS partner status, and a named expansion mechanism beyond STR — any three of the four reopen the verdict.
Revisit trigger / date: founder delivery of the pre-call proof packet, or 2026-09-10, whichever comes first. This decision is logged with these flip conditions and will be revisited against reality, not memory.

Source Log

All sources below were directly checked on 2026-06-10 or 2026-06-11 and are publicly accessible. Company-controlled pages prove what the company claims, not that the claims are true. Vendor market reports are directional estimates, not verified figures.

Trellis homepage
trellistech.com
Product positioning, 200+ integrations claim, autonomy demo, ICP, rating markup
Retrieved 2026-06-10company-controlled pagehigh
Trellis integrations directory
trellistech.com
37 public connectors across 7 categories
Retrieved 2026-06-10company-controlled cataloghigh
Trellis about page
trellistech.com
ICP, design-partner program
Retrieved 2026-06-10company-controlled pagehigh
Trellis Terms of Service
trellistech.com
Legal entity; AI-output liability allocation
Retrieved 2026-06-10company-controlled legal pagehigh
Trellis Privacy Policy
trellistech.com
Data practices; AI vendors; worker location data
Retrieved 2026-06-10company-controlled legal pagehigh
Trellis Data Processing Addendum
trellistech.com
Subprocessors (incl. browser automation); SOC 2 audit clause
Retrieved 2026-06-10company-controlled legal pagehigh
Trellis API Terms
trellistech.com
Public API program existence
Retrieved 2026-06-10company-controlled legal pagehigh
Trellis data-deletion page
trellistech.com
Deletion process existence
Retrieved 2026-06-10company-controlled policy pagehigh
Trellis documentation hub
docs.trellistech.com
Docs/API/changelog existence
Retrieved 2026-06-10technical documentationhigh
Y Combinator directory: Trellis
ycombinator.com
YC P26 membership; founders; team size; traction narrative (as company claim)
Retrieved 2026-06-10accelerator listinghigh (membership) / low (traction text)
Y Combinator launch: Trellis
ycombinator.com
Launch narrative; founder background claims
Retrieved 2026-06-10accelerator launch pagemedium
LinkedIn: Trellis company page
linkedin.com
Team size band; HQ; founder roles
Retrieved 2026-06-10public profile (self-reported)medium
LinkedIn: Jan Sahagun
linkedin.com
Founder identity
Retrieved 2026-06-10public profile (self-reported)medium
LinkedIn: Lodovico Benvenuti
linkedin.com
Founder identity
Retrieved 2026-06-10public profile (self-reported)low
X (Twitter): @lodobenvenuti00
x.com
CEO's personal public account (23 followers, 8 posts, joined July 2024)
Retrieved 2026-06-10social platform metrics (X)medium
Apple App Store: Trellis Mobile
apps.apple.com
iOS rating/review count; privacy labels
Retrieved 2026-06-10app-store platform metricshigh
Google Play: Trellis Mobile
play.google.com
Android downloads/review count
Retrieved 2026-06-10app-store platform metricshigh
X (Twitter): @usetrellis
x.com
Company social footprint: 51 followers, 5 posts, created April 2026
Retrieved 2026-06-10social platform metrics (X)medium
X (Twitter): Y Combinator launch post for Trellis
x.com
Launch visibility: ~37.4k views, 206 likes on YC's account
Retrieved 2026-06-10social platform metrics (X)medium
Instagram: @usetrellis
instagram.com
Owned-social footprint: private account, 0 followers/posts
Retrieved 2026-06-10social platform metrics (Instagram)medium
YouTube: @usetrellis
youtube.com
Owned-social footprint: 8 subscribers, 1 video, ~665 views
Retrieved 2026-06-10social platform metrics (YouTube)medium
Airbnb Terms of Service
airbnb.com
Prohibition on bots/automated access
Retrieved 2026-06-10platform termshigh
Vrbo Terms of Service
vrbo.com
Vrbo platform terms governing automated access
Retrieved 2026-06-10platform termshigh
Airbnb Help Center: software/partner policy
airbnb.co.uk
Airbnb's policy on third-party software and partner access (developer portal itself blocks automated retrieval)
Retrieved 2026-06-10platform policy (Airbnb help center)high
GitHub: lodobenvenuti
github.com
CEO's public technical footprint (10 public repositories)
Retrieved 2026-06-10developer platform metrics (GitHub)medium
GitHub: lodobenvenuti repositories
github.com
Repo-level footprint: 6 forks / 4 original projects, languages, last-update dates
Retrieved 2026-06-11developer platform metrics (GitHub)medium
AI Tinkerers Paris: Global AI Hackathon judges
paris.aitinkerers.org
CEO's judge listing and self-submitted bio (Conduit role, Cambridge claim, Xenia "€3M ARR" telling)
Retrieved 2026-06-11third-party event platform (bio self-submitted)medium
Hi! PARIS: Hi!ckathon 2023 awards
hi-paris.fr
Independent HEC Paris affiliation trace for the CEO (awarded team listing)
Retrieved 2026-06-11independent institutional pagemedium
Y Combinator directory: Conduit
ycombinator.com
Existence of the CEO's claimed prior employer (YC W24, AI agents for hospitality)
Retrieved 2026-06-10accelerator listingmedium
Getpodcast: "The Story of AI Agents Running a 150-Unit Company"
getpodcast.com
Existence of a founder podcast episode (content is founder narrative, not validation)
Retrieved 2026-06-10podcast directory (Getpodcast)medium
Instagram: Xenia Milano
instagram.com
Existence of a Milan accommodation brand named Xenia (linkage to founder unconfirmed)
Retrieved 2026-06-10brand-controlled social profilelow
STR Tech Report: Jan Sahagun profile
strtechreport.com
COO's media-platform profile (self-controlled)
Retrieved 2026-06-10self-controlled media profilelow
WhatsApp Business Messaging Policy
business.whatsapp.com
Opt-in, 24-hour/template, human-escalation rules
Retrieved 2026-06-10platform policyhigh
ShortTermRentalz: Guesty launches Agent Hub
shorttermrentalz.com
Incumbent agent-platform launch (why-now + competitive pressure)
Retrieved 2026-06-10STR trade press (ShortTermRentalz)medium
Travolution: Guesty launches Agent Hub
travolution.com
Second independent confirmation of the Agent Hub launch
Retrieved 2026-06-11travel trade press (Travolution)medium
Crunchbase: Guesty
crunchbase.com
Guesty scale, Series F, acquisitions
Retrieved 2026-06-10startup database (Crunchbase)medium
Crunchbase: Hostaway
crunchbase.com
Hostaway profile and backers
Retrieved 2026-06-10startup database (Crunchbase)medium
Crunchbase: Breezeway
crunchbase.com
Breezeway profile
Retrieved 2026-06-10startup database (Crunchbase)medium
TechCrunch: Guesty raises $130M Series F
techcrunch.com
Guesty round size and ~$900M valuation
Retrieved 2026-06-10tech press (TechCrunch)high
General Atlantic: Hostaway $365M investment
generalatlantic.com
Hostaway round size and lead investor
Retrieved 2026-06-10investor announcement (General Atlantic)medium
ShortTermRentalz: Breezeway secures Resurgens investment
shorttermrentalz.com
Breezeway growth investment (amount undisclosed)
Retrieved 2026-06-10STR trade press (ShortTermRentalz)medium
Guesty AI features
guesty.com
Incumbent native-AI positioning
Retrieved 2026-06-10competitor-controlled pagemedium
Hostaway AI
hostaway.com
Incumbent native-AI positioning
Retrieved 2026-06-10competitor-controlled pagemedium
Hospitable AI
hospitable.com
PMS + AI bundle positioning
Retrieved 2026-06-10competitor-controlled pagemedium
Besty AI
besty.ai
Direct AI-messaging competitor
Retrieved 2026-06-10competitor-controlled pagemedium
Conduit (hostai.app)
hostai.app
Direct AI-agent competitor
Retrieved 2026-06-10competitor-controlled pagemedium
HostBuddy AI
hostbuddy.ai
Direct AI-messaging competitor
Retrieved 2026-06-10competitor-controlled pagelow
Enso Connect
ensoconnect.com
Guest-experience AI competitor
Retrieved 2026-06-10competitor-controlled pagemedium
Duve
duve.com
Guest-journey competitor
Retrieved 2026-06-10competitor-controlled pagemedium
Breezeway
breezeway.io
Ops/turnover competitor
Retrieved 2026-06-10competitor-controlled pagemedium
Turno
turno.com
Turnover substitute
Retrieved 2026-06-10competitor-controlled pagemedium
Hostaway: 2024 STR market trends
hostaway.com
Operator professionalization and AI-adoption discourse
Retrieved 2026-06-10vendor blog (Hostaway)medium
Dataintelo: Vacation Rental Management Software Market
dataintelo.com
Narrow top-down estimate; listing/unit claims (directional)
Retrieved 2026-06-10vendor market report (Dataintelo)low
Market Research Future: Vacation Rental Software Market
marketresearchfuture.com
Broad top-down estimate (definition-sensitive)
Retrieved 2026-06-10vendor market report (Market Research Future)low
AirDNA
airdna.co
STR supply-data scale context
Retrieved 2026-06-10company-controlled pagemedium

Access limitations: Trellis's pricing and careers URLs return not-found pages (cited above as evidence of absence, not as content sources); Airbnb's developer portal blocks automated retrieval, so its software-partner policy is cited via the public help-center article above and the exact developer-terms text remains a diligence item; Apple Podcasts blocked retrieval of the founder podcast episode, which is cited via an independent directory mirror instead; no Crunchbase, funding-database, or state-registry record exists for Trellis itself; a 2019 Italian newspaper profile of the CEO surfaced in search results but its article body could not be retrieved, so it is excluded as snippet-grade material; education and credential claims (Cambridge, McGill) beyond the HEC Paris trace above remain self-reported; search-result and AI-assistant visibility snapshots informed competitive context but are excluded from this table as snippet-grade material.

Public disclaimer: This is an independent public teardown based solely on publicly available information retrieved on 2026-06-10 and 2026-06-11. It is research support, not investment, legal, tax, or financial advice. Company-provided statements are labeled as claims, not verified facts. Trellis did not participate in or review this report. Errors and omissions are possible; final decisions remain with the reader.

What The Company Does#

Key Takeaways#

What Makes This Potentially Fundable#

The Four Holes To Close Before Fundraising#

Decision Snapshot#

IC Disagreement Map#

Scenario Range#

What Investors Will Test#

Claim Reconciliation: Inconsistencies Investors Will Catch#

Diligence Findings: Issues To Fix Before You Raise#

Data Room Readiness Checklist#

First-Call Agenda For The Startup#

Business Model And Revenue Signals#

Revenue Quality Checklist#

Funding And Ownership Context#

Founder, Team, And Related Entities#

Founder And Team#

Founder Competency Coverage#

Public Professional Footprint#

Founder-Market Fit Read#

Related Entities And Founder-Associated Companies#

Traction#

Customer Status Table#

Traction Signals#

Hiring And Org Momentum#

Traction Quality Read#

Competitive Landscape#

Category Visibility Snapshot#

Pricing And Competitive Benchmark#

Competitive Wedge#

Risks And Open Questions#

Pre-Mortem: The Most Likely Obituary#

Decision-Critical Unknowns#

Diligence Questions#

First Call#

Follow-Up#

Kill Criteria#

Double-Down Criteria#

Founder Action Plan#

Decision#

How We Would Miss This One#

Source Log#