BlitzGraph (Blitz Holdings Inc)

June 24, 2026

Public disclaimer: This is an independent public teardown based solely on publicly available information retrieved on 2026-06-24. It is research support, not investment, legal, tax, or financial advice. Company-provided statements are labeled as claims, not verified facts. BlitzGraph did not participate in or review this report. Errors and omissions are possible; final decisions remain with the reader.

Most pre-revenue infrastructure founders ask you to take their word for it. This one left a paper trail you can read without taking the meeting: years of public bug reports filed against a leading open-source graph database, the rare proof an investor can check from the outside. Almost everything else sits on the company's side of the line. BlitzGraph pitches a graph-native "AI-native backend" for AI agents — public beta, wired into the agent runtime, riding a $1.3 trillion agentic wave — yet whether a single developer actually reaches for it, and whether it can keep the value it creates, is still unproven. Five partners spent one day on the public record and landed on a single question: is this the backend agents get built on, or a convenience the stack around it absorbs? The public sources cannot close it. A founder call can. Read the teardown before that call prices the round.

What The Company Does

BlitzGraph describes itself as an "AI-native backend" — a hosted, graph-native database and backend-as-a-service built for AI agents. Instead of SQL, joins, or object-relational mappers, it models data as a typed graph queried through a JSON-based language the company calls BQL, exposed over an HTTP API with TypeScript and React software development kits, a studio user interface, and hosted frontends (BlitzGraph homepage; BlitzGraph docs) company claim. Its distribution wedge is a remote server speaking the Model Context Protocol, the emerging standard for letting AI coding agents such as Claude Code and Codex call external tools directly (BlitzGraph homepage; BlitzGraph changelog) company claim.

The buyer is a developer or engineering, data, or AI team building agent applications. The product is self-serve, with sign-up through the dashboard and a "book a demo" path (BlitzGraph homepage) company claim. The business model is usage-based billing through Stripe, structured by a "space tier" plus usage (BlitzGraph terms of service) company claim. Two facts about the commercial model are independently observable rather than asserted: there is no public pricing page, and the founder has stated that payments are not yet enabled (Hacker News) company claim for the payments status, with the missing pricing page confirmed directly. The current public release is an active beta with frequent prior releases (BlitzGraph changelog) company claim. The legal operator named on the company's own legal pages is Blitz Holdings Inc, San Francisco (BlitzGraph privacy policy) company claim.

Key Takeaways

Investor reaction
BlitzGraph pairs a genuinely domain-deep, Y Combinator-lineage technical founder with a non-obvious distribution idea on a very large agentic tailwind — but every load-bearing commercial pillar (paying customers, value capture, team coverage, ownership, benchmarks) is company-stated or unknown, and a dedicated second pass over the public record upgraded none of them.
Current verdict
hold. This is worth one disciplined founder call to resolve the gating questions, not a check and not a fast no.
Why investors might lean in
The founder has independently corroborated, multi-year graph-database domain depth (GitHub — SurrealDB issues) source-backed fact; the product is purpose-built for AI agents and reaches them inside the agent runtime, so distribution and lock-in could compound together if the wedge converts; and the entry would be the cheapest possible option on a potential default-infrastructure outcome before it is consensus.
Why investors might pull back
There is no independent evidence any developer reaches for BlitzGraph — it is absent from category search results, has at most one downstream package dependent (npm — @blitzgraph/client-core) source-backed fact, drew a modest launch reception, names no customers, and has no public pricing with payments off; the value it creates can be assembled by composing an agent-memory layer onto an installed backend; the engine is closed with a single publicly visible builder; and the "Backed by Y Combinator" badge maps to the founder's prior company, not to a BlitzGraph listing, leaving ownership unreadable.
Highest-leverage fix
Produce one independently referenceable paying or production customer with retained usage, a named core-engineering team, and a one-page entity and cap-table answer. This single bundle discharges the three deepest holes at once.
Best next move
Run one decisive, time-boxed founder call with pre-committed pursue-or-pass triggers; advance to deeper diligence only if a referenceable paying customer, a value-capture answer, a named core team, and a clean cap table all appear, and commit now to a pass — not a third hold — if the demand and value-capture gates stay empty.

What Makes This Potentially Fundable

The fundable version of BlitzGraph is not "one more graph database." It is the default data and memory layer that AI agents get built on. If agent state is genuinely graph-shaped, and if the native access path into the agent runtime (the Model Context Protocol server) becomes the way agents reach their backend, then a company that owns that layer accrues switching costs as agents accumulate graph state and history. The relevant ceiling would stop being "graph database" — a pool a public incumbent reached at a little over $200 million in annual recurring revenue after roughly twenty-five years (Neo4j press release) company claim — and become "default backend for the agentic build-out," where agentic information-technology spend is projected to reach $1.3 trillion by 2029 (IDC) source-backed fact, and where public database-as-a-service comparables show software-like gross margins are achievable at scale (MongoDB Q3 FY2026 results) source-backed fact.

The non-obvious path from wedge to advantage is that distribution and lock-in could arrive together: if the agent reaches for BlitzGraph at build time, the company captures both the integration and the accumulating graph state. The honest gap that keeps this report at hold is that the mechanism is entirely unmeasured. The same forecast that sizes the tailwind also warns that roughly 80% of agentic infrastructure spend concentrates with large service providers, and that more than 40% of agentic projects will be canceled by the end of 2027 (IDC; Gartner) source-backed fact. Until there is evidence the wedge converts, the upside is a category option, not an underwritten outcome.

The Four Holes To Close Before Fundraising

Hole	Investor Fear	What To Bring
No independent demand pull anywhere.	An asymmetric bet with confirmed zero pull is a narrative, not an option.	Named, referenceable paying or production accounts and a retention or usage cohort.
No legible value capture against a composable stack.	A developer can compose an agent-memory layer onto an installed backend, so the value BlitzGraph creates is captured by the layers around it — a feature, not a company.	Live pricing, a value metric, and a named win where a developer chose BlitzGraph over composing alternatives, with willingness to pay.
A single visible builder on an infrastructure-grade custom engine.	Key-person risk: the hardest part of the company rests on one person.	Named core-engine engineers, a commit-author log, or active headcount on the operating entity.
Unreadable entity and cap table, and the accuracy of the accelerator badge.	Ownership cannot be priced and the accelerator claim cannot be trusted.	Incorporation documents, a current cap table, and confirmation of which entity holds the Y Combinator relationship.

Decision Snapshot

One-sentence company description: BlitzGraph is a public-beta, graph-native backend-as-a-service that positions itself as an AI-native data and memory layer for agents, reached through a remote Model Context Protocol server.
Screen: hold.
Confidence: medium.
Suggested next action: One tightly scoped, time-boxed founder call, pre-committed to kill criteria, focused on independent demand pull, value capture, core-engine team coverage, and the entity and cap-table picture.
Why this matters now: Agent applications are being built faster than the data layer beneath them has standardized, so the category window is real, but a 2027 project-cancellation wave and provider concentration mean a pre-revenue entrant must show pull before the window closes.
Investor-readiness diagnosis: Enough founder and category signal to earn a call, not enough independent proof to earn diligence momentum.
Best founder use of this report: Treat it as a proof-packet checklist before investor enthusiasm turns into a customer, value-capture, team, and ownership interrogation.

IC Disagreement Map

Evidence dossier review was independent across five public partner lenses, each reading the same dry evidence: the power-law lens tested fund-returning upside, the prepared-mind lens tested category timing and product insight, the founder-jockey lens tested team fit and execution, the risk-reduction lens tested critical flaws and value capture, and the long-horizon lens tested portfolio fit and ownership durability.

Power-law partner
First-call worthylow confidence
One cheap call tests an asymmetric agent-backend option on a $1.3 trillion wave carried by a domain-deep founder, and pre-killing it is the classic power-law error of omission.
Would flip on
Independent demand pull, namely referenced customers, retained usage, or organic adoption, plus one value-capture win, would move them toward deeper diligence; a confirmed solo core-engine maintainer or an aspirational accelerator and entity attribution would move them to drop it.
Prepared-mind partner
holdmedium confidence
A prepared-mind insight may have picked the right battleground at the right moment, but there is no observable category pull yet and the positioning is contested by graph and memory peers.
Would flip on
Referenced production users plus an independent benchmark plus early ecosystem lock-in (any two) would move them to pursue; a confirmed solo engine, buyers preferring to compose alternatives, or an incumbent shipping an equivalent agent-native graph layer would move them to pass.
Founder-jockey partner
holdmedium confidence
The single best-evidenced fact in this research is a team fact, the founder's multi-year domain depth, but core-engine team coverage and the leadership track record are unconfirmed on the hardest possible build.
Would flip on
A named credible second core engineer plus founder-authored engine commits plus a validated prior leadership record would move them to pursue; a confirmed solo-maintained engine paired with an inflated founder narrative would move them to pass.
Risk-reduction partner
holdmedium confidence
Founder depth is real, but every load-bearing commercial fact is absent or company-stated and three structural risks stack: value capture against a composable stack, a single-visible-builder engine, and multi-tenant server-side code execution.
Would flip on
Named referenceable paying or production customers with retention, a named core team, and a credible compliance and reliability roadmap would move them to pursue; persistent absence at the next milestone, unsafe code isolation, or an encumbered cap table would move them to pass.
Long-horizon partner
holdmedium confidence
Ownership is unreadable and no asset compounds yet, so the cap-table and entity structure is the decisive variable over a long hold, and the company could be un-investable on ownership grounds regardless of product.
Would flip on
A clean cap table with founder majority plus first paying cohorts with retention plus a forming compounding asset would move them toward pursue; stale preferred rights from the prior business, or zero conversion to paid after payments enable, would move them to pass.

After the committee's evidence-request round, four partners confirmed their votes unchanged and one (the power-law lens) lowered confidence from medium to low because the focused second pass surfaced no independent demand and no value-capture win, thinning the asymmetry without refuting it. The votes were not unanimous — four hold and one first-call-worthy, with no partner reaching pursue or pass. The near-consensus is evidential rather than social: the four holders reached hold from four different lenses, each dead-ending on a different publicly unanswerable variable that traces to the same root — a thin, single-visible-builder, closed, pre-revenue entity whose load-bearing commercial facts are all company-controlled — while disagreeing sharply on which way the case ultimately leans.

Scenario Range

Scenario	What The Company Looks Like In 3-5 Years	Falsifiable Trigger To Watch	Earliest Evidence
Strikeout	A dormant repository or an acqui-hire: a respected data model and an articulate, domain-deep founder, but the access wedge never converted to retained paid usage, the composable stack and incumbents absorbed the workload, and the company ran out of money and patience in the 2027 agentic air-pocket.	Adoption stays flat while the founder keeps shipping — release cadence and community activity continue, but downstream package dependents stay at a handful, there is still no independently referenceable paying customer, and payments stay off.	Within 6-12 months: shipping effort rises while revenue stays `unknown`, at most one package dependent persists, and there is still no public pricing — which is already the state today.
Base	A technically interesting, well-architected niche graph backend with a thin trickle of low-value indie and hobby paying spaces once payments turn on; sustainable at best but not category-defining, a capped pool that does not on its own return a fund.	A handful of paying spaces and live pricing appear, but no reallocated named budget line, no win displacing an incumbent at scale, and enterprise gates stay closed.	Within 12-18 months: live pricing and a first paid cohort appear without a category-defining migration pattern or any enterprise-readiness artifact.
Home run	BlitzGraph is the default agent backend and memory layer for a meaningful share of agentic apps — a sticky system of record with high switching costs, durable subscription margins, and compounding schema and ecosystem gravity on the agentic build-out.	Buyers reallocate a named existing database or backend budget line to BlitzGraph for the agent-state workload, win-loss shows durable wins over the composed stack, and downstream dependents and retained-usage cohorts grow tracking releases rather than lagging them.	Within 12-18 months: at least one production migration naming a displaced vendor, a third-party-reproducible traversal benchmark advantage on agent-state workloads, and retention cohorts (not community buzz) showing a switching-cost asset accruing.

What Investors Will Test

What We Looked For	Current Read	Investor Implication	Founder Prep Priority
A falsifiable path to a fund-returning outcome, not generic excitement.	The home-run mechanism (the access wedge converting to a default backend) is hypothesis-only; today's state is exactly its kill condition.	Investors will not underwrite "interesting technology plus accelerator plus protocol" as upside.	Name the single adoption mechanism and the observation in the next 12-18 months that would falsify it.
Independent proof outside the company's own surfaces.	Only the founder's domain depth survived independent corroboration; customers, benchmarks, revenue, headcount, and pricing did not.	The first call should be a proof test, not a narrative pitch.	Bring referenceable users, a retention cohort, and any third-party benchmark.
Founder-market fit on the hardest build.	Domain depth is strongly evidenced; the ability to ship, operate, staff, and sell infrastructure-grade reliability is unproven.	Founder quality earns a call but does not clear team or execution diligence.	Document core-engine authorship and a named second engineer.
Durable value capture against a composable stack.	The value created can be assembled from composable parts, and most agentic spend may concentrate with providers.	Value creation without value capture reads as a feature, not a company.	Show a named win and willingness to pay against the composed alternative.
Underwritable ownership.	The entity and cap table are unreadable, and the accelerator badge maps to a different entity.	Ownership math cannot be done, which can stall a deal regardless of product.	Provide incorporation documents and a current cap table.

Claim Reconciliation: Inconsistencies Investors Will Catch

Claim Or Metric	Where It Appears	Conflicting / Unreconciled Versions	Why Investors Flag It	How To Reconcile
Who actually holds the Y Combinator relationship.	BlitzGraph homepage Y Combinator — Blitz	The homepage shows BlitzGraph as "Backed by Y Combinator," but the accelerator lists only Blitz (Summer 2022) and no BlitzGraph company page.	A badge that maps to a different entity reads as borrowed credibility at best and misattribution at worst.	Confirm which legal entity and batch holds the relationship and show the cap table.
Funding attribution and the operating entity.	Y Combinator — Blitz SEC EDGAR — company search	The prior company shows a single pre-seed round; there is no BlitzGraph-specific raise on the public record and no securities issuer surfaced.	Runway, dilution, and ownership math all depend on whether BlitzGraph is a pivot inside the prior holding company or a new entity.	Provide incorporation documents and confirm whether BlitzGraph is a new company or a pivot.
Team size on the operating entity.	BlitzGraph homepage	The prior company shows a small team historically, but only one builder is publicly visible on BlitzGraph today.	A single-builder infrastructure-grade engine is a key-person risk investors will probe.	Provide a commit-author log and the roster mapped to the BlitzGraph entity.
The depth of the founder's issue-filing history.	Hacker News GitHub — SurrealDB issues	The founder cites "120-plus issues across graph databases," while 56 against one database are independently visible.	Even a sympathetic count gap erodes trust in other self-reported numbers.	Export issues across all named repositories so the full count is verifiable.
Beta data durability.	Hacker News BlitzGraph terms of service	The founder states daily backups with no wipes yet, while the terms warn beta data may be wiped.	A durability statement that conflicts with the terms is a procurement and trust flag.	Publish a backup-and-restore commitment and reconcile it with the terms.

Diligence Findings: Issues To Fix Before You Raise

2 deal-breaker4 high2 medium

No independently referenceable paying or production customer exists after a public launch.
deal-breaker if unresolved after the first call.
Commercial diligence.
Fix
Provide a referenceable retained paying or production account with a reference call.
Value capture against a composable agent-memory plus backend stack is unproven.
high
Market and strategy diligence.
Fix
Show a named win and willingness to pay against the composed alternative.
The core engine has a single publicly visible builder and the engine is closed.
high
Team and technical diligence.
Fix
Provide a commit-author log and a named second core engineer.
The entity and cap table are unreadable and the accelerator badge maps to a different entity.
high, escalating to deal-breaker if the badge is materially misattributed or ownership is encumbered.
Financing and legal diligence.
Fix
Provide incorporation documents, a cap table, and accelerator confirmation.
Server-side JavaScript runs in the shared multi-tenant backend with isolation untested externally.
high
Technical and security diligence.
Fix
Provide a tenancy isolation architecture and an independent penetration test.
The company is pre-revenue with no public pricing and payments off.
high
Financial diligence.
Fix
Enable payments, publish pricing and a value metric, and capture a first paid cohort.
There is no compliance or production-readiness posture for regulated or enterprise buyers.
medium
Legal and procurement diligence.
Fix
Provide a compliance roadmap, a regional data-residency option, and a service-level commitment.
The performance differentiation is self-asserted and unbenchmarked.
medium
Technical diligence.
Fix
Provide a third-party-reproducible benchmark on agent-state workloads.

Data Room Readiness Checklist

For a developer-infrastructure product that will eventually sell into mid-market, enterprise, and regulated buyers, the security and compliance gatekeepers will demand artifacts that do not yet exist publicly.

Procurement-readiness score: not ready.
Rationale: Public evidence shows United-States-only hosting with no regional data-residency option or signed data-processing agreement, no compliance attestation such as a security audit report, no service-level commitment, beta data-durability warnings, and server-side code execution in a shared multi-tenant backend — the cluster of items a regulated or enterprise buyer's gatekeepers screen first (BlitzGraph privacy policy; BlitzGraph terms of service; BlitzGraph docs).

Data Room Area	What Investors Expect	Current Read	Status	Priority
Company overview & corporate	One-pager, incorporation, good standing, structure	Blitz Holdings Inc is named on the legal pages, but the BlitzGraph entity mapping is unresolved (BlitzGraph privacy policy).	`partial`	High
Financials	Monthly profit and loss (18-24 months), three-year model with assumptions, burn and runway	No public financials; pre-revenue with payments off.	`missing`	High
Cap table & funding history	Clean cap table, prior rounds, simple agreements or notes, 409A valuation	A pre-seed round exists on the prior company; no BlitzGraph-specific raise or cap table is public.	`missing`	Critical
Legal & IP	Bylaws, board consents, intellectual-property assignments, material contracts	Legal pages exist; intellectual-property and assignment posture is not public.	`partial`	High
Product & technology	Roadmap, architecture overview, security and compliance documents	Architecture and a changelog exist; no security or compliance package, and a multi-tenant code-execution surface is open (BlitzGraph docs).	`partial`	Critical
Team	Org chart, key employment and advisor agreements, vesting	One visible builder; no roster, agreements, or vesting visible.	`missing`	Critical
Customers & traction	Retention cohorts, revenue bridge, pipeline, references	No customers, cohorts, or references found.	`missing`	Critical
Market & competition	Market model, competitive landscape, pricing	Competitive landscape is mappable; the company's own pricing and served-market evidence are missing.	`partial`	High

First-Call Agenda For The Startup

Time	Topic	Founder Goal	Evidence To Bring
0-10 minutes	Frame the exact wedge.	Make clear which agent-state workload BlitzGraph wins first and why a graph backend serves it better than the alternatives.	A concrete workload, the buyer, and the displaced alternative.
10-25 minutes	Demand pull.	Prove real adoption beyond a launch thread.	A referenceable user, a retention or usage cohort, and adoption metrics.
25-35 minutes	Value capture.	Show why a developer pays BlitzGraph rather than composing alternatives.	A named win, pricing, a value metric, and willingness to pay.
35-45 minutes	Team and engine.	Show the hardest build has more than one owner.	A commit-author log and named core engineers, including the co-founder's status.
45-55 minutes	Entity and ownership.	Make ownership underwritable and the accelerator claim verifiable.	Incorporation documents, a cap table, and accelerator confirmation.
55-60 minutes	Triggers.	Agree what proof moves the decision to pursue or to pass.	Pre-committed pursue-or-pass triggers and a revisit date.

Business Model And Revenue Signals

Usage-based billing by space tier plus usage, through a third-party payment processor.
Company-controlled terms.medium
SourceBlitzGraph terms of service
The model is described but no dollar tiers are published.
No public pricing page.
source-backed factCompany-controlled pages.high
SourceBlitzGraph homepage
Blocks revenue and value-capture diligence.
Payments not yet enabled.
Founder forum post.medium
SourceHacker News
Pre-revenue; no basis to model economics.
Software-like margins are achievable for a database company at scale.
source-backed fact for the comparable; inference for BlitzGraph.Public company results.medium
SourceMongoDB Q3 FY2026 results
The comparable does not prove BlitzGraph's margin.

No public source discloses BlitzGraph revenue, paying-space counts, retention, or unit economics. Any revenue figure before founder disclosure would be an estimate, not a fact.

Revenue Quality Checklist

Before this can move from hold to pursue, the company should be ready to show, once payments are live, paying-space counts and conversion from beta, retention and repeat usage, customer concentration, and the value metric pricing is based on. Investors will rebuild customer acquisition cost, lifetime value, payback, gross revenue retention, net revenue retention, and churn from raw billing and usage data, and those 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	A single pre-seed round is recorded on the prior company; no BlitzGraph-specific raise is public (Y Combinator — Blitz).	`source-backed fact` for the prior round's existence; `unknown` for any BlitzGraph raise.	Provide a financing history and reconcile the BlitzGraph entity to it.
Instruments (equity / SAFE / notes)	Not publicly disclosed.	`unknown`	Provide instruments and conversion terms.
Investors	The prior company's pre-seed names an accelerator and one venture investor (Y Combinator — Blitz).	`source-backed fact` for the prior company; `unknown` for BlitzGraph.	Provide the investor schedule and any prior-investor rights.
Post-money valuation	Not publicly disclosed.	`unknown`	Provide valuation and available allocation.
Cap table / ownership	Not publicly disclosed.	`unknown`	Provide a fully diluted cap table, option pool, and founder vesting.
Burn & runway	Not publicly disclosed.	`unknown`	Provide monthly burn, cash balance, and runway.
Next-round plan	Not publicly disclosed.	`unknown`	Provide a financing road through the 2027 air-pocket.

SEC EDGAR was checked and showed no matching securities issuer for the operating company. This absence does not disprove the prior raise, but it means public financing records do not support ownership or valuation modeling.

Founder And Team

Loïc Veillard Garoz
Founder and the single publicly visible builder; independently corroborated graph-database domain depth.
GitHub — Loïc Veillard GitHub — SurrealDB issues LinkedIn — Loïc Veillard Hacker News
Evidencesource-backed fact for domain depth and identity; company claim for the leadership and employment history.
Confidencehigh for domain depth; medium for the broader profile.
Clement (named co-founder of the prior company)
Possible co-founder; current BlitzGraph role unknown.
Y Combinator — Blitz
Evidencesource-backed fact for the prior-company roster; unknown for the current role.
Confidencemedium for the prior role; low for the current role.
Published client packages
TypeScript and React software development kits published on a public registry.
npm — @blitzgraph/client-core
Evidencesource-backed fact
Confidencehigh

Founder Competency Coverage

Domain depth
Loïc Veillard Garoz
evidenced
Clement (prior co-founder)
unknown
What The Evidence Is
56 public issues against a leading graph database independently corroborate multi-year graph-database depth for the founder; nothing public covers the co-founder.
Technical build capability
Loïc Veillard Garoz
claimed
Clement (prior co-founder)
unknown
What The Evidence Is
Client packages are published, but the core engine is closed and its authorship is not publicly verifiable.
Product
Loïc Veillard Garoz
claimed
Clement (prior co-founder)
unknown
What The Evidence Is
The product surface is described only on company-controlled pages.
GTM / sales
Loïc Veillard Garoz
unknown
Clement (prior co-founder)
unknown
What The Evidence Is
No public evidence of a go-to-market or sales motion.
Leadership / hiring
Loïc Veillard Garoz
claimed
Clement (prior co-founder)
unknown
What The Evidence Is
A prior leadership history is self-submitted and uncorroborated; no BlitzGraph hiring is indexed.
Fundraising history
Loïc Veillard Garoz
claimed
Clement (prior co-founder)
unknown
What The Evidence Is
The prior company raised pre-seed; the founder's specific role is not separately evidenced.
Prior founding outcomes
Loïc Veillard Garoz
claimed
Clement (prior co-founder)
claimed
What The Evidence Is
Both are recorded as founders of the prior company; the outcome of that company is undisclosed.

A custom graph engine, query planner, semantic search, and cloud-authentication layer.
high
Team coverage today
A single publicly visible builder; engine authorship is unverified.
What would close it
A commit-author log showing founder-authored engine work plus at least one named systems engineer.

Public Professional Footprint

These checks of public technical artifacts, professional social presence, and education and credential traces carry low decision weight by design; they corroborate or weaken the coverage table above and sharpen founder-call questions, but none alone changes the verdict.

Technical artifacts (repo level)
Loïc Veillard Garoz
evidenced
Clement (prior co-founder)
unknown
What The Public Record Shows
A public profile with published packages and a substantial issue history, though the personal repositories are largely forks rather than original infrastructure work, and the core engine is not public (GitHub — Loïc Veillard).
Professional social content
Loïc Veillard Garoz
evidenced
Clement (prior co-founder)
unknown
What The Public Record Shows
Candid public discussion of what is unfinished, which is a learning-speed and integrity signal; repetition of the founder's own performance and depth claims is not validation (Hacker News).
Education / credentials
Loïc Veillard Garoz
unknown
Clement (prior co-founder)
unknown
What The Public Record Shows
The founder's public profile is thin on verifiable education or credential traces (LinkedIn — Loïc Veillard).
Publications / patents / certifications
Loïc Veillard Garoz
absent
Clement (prior co-founder)
unknown
What The Public Record Shows
No publications, patents, or certifications surfaced.

Net read: the footprint supports a first call because the founder's domain depth is genuine and rare, but the verdict stays gated because the footprint does not prove the engine is staffed beyond one person or that the value can be captured.

Founder-Market Fit Read

What is promising: A multi-year, independently corroborated obsession with graph-database failure modes is one of the most reliable early indicators of a founder who can will infrastructure into a category.
What is missing: Proof the same person, plus a team, can ship, operate, and sell production infrastructure, not only catalog where existing systems fall short.
What to prepare: A commit-author log, a named second core engineer, and reference checks on the prior leadership history.

Blitz Holdings Inc
Legal operator named on BlitzGraph's own legal pages.
BlitzGraph privacy policy
Evidencecompany claim for the named operator.
Confidencehigh
Follow-upConfirm whether BlitzGraph is this entity or a new company.
Blitz (Summer 2022)
The founder's prior accelerator company; the source of the Y Combinator lineage.
Y Combinator — Blitz
Evidencesource-backed fact
Confidencehigh
Follow-upConfirm which entity holds the accelerator relationship and the prior outcome.

Traction

Customer Status Table

No named customer was identified in public evidence.
The company names no customer logos, case studies, or testimonials on its own surfaces.
No.Claim onlyUnknownUnknownUnknownhigh that no public customer proof was found.

The company's own pages and general category searches were checked and showed no public named customer, design win, or case study. This absence is decision-relevant after a public launch, but it does not prove no private evaluations exist.

Traction Signals

A modest launch reception on the founder's public thread.
source-backed facthigh
SourceHacker News
Whether launch interest converts to retained usage.
At most one downstream package dependent on the published client.
source-backed facthigh
Sourcenpm — @blitzgraph/client-core
Growing dependents and a forming tool ecosystem.
Frequent beta releases with repeated rework of the agent-access integration.
high
SourceBlitzGraph changelog
Whether shipping effort converts to adoption rather than only cadence.

Category search engines were checked across agent-memory, agent-native backend, backend-for-agents, and graph-for-agents queries, and BlitzGraph did not appear in the page-one results, while a memory-layer peer ranked first. These checks measure indexed visibility on the retrieval date, not non-existence, and carry low weight for a very early company.

Hiring And Org Momentum

Engineering footprint
unknown for team sizeOnly the founder is publicly visible as an active builder; the engine is closed with no public organization.
SourceGitHub — Loïc Veillard
Consistent with a single-builder or very small team, which is the central execution risk on an infrastructure-grade build.

LinkedIn Jobs, Indeed, Glassdoor, and the company careers page were checked and showed no matching listing, and the careers page was not found. These absences carry little weight since job boards are often incomplete for very early companies, but they offer no evidence of hiring velocity, and employee sentiment remains unavailable.

Role-mix read: No public role mix exists to read; there are no indexed postings.
Momentum read (growth / steady / contraction / unknown): unknown, because shipping cadence is visible but hiring and headcount are not.

Traction Quality Read

Traction Dimension	Current Status	Good Enough For First Call?	Needed For Deep Diligence
Demand pull	No customers; at most one package dependent.	Only if the first call is explicitly gated on resolving it.	Referenceable retained paying or production users.
Value capture	No pricing; payments off; structurally contested.	Only as a question, not as conviction.	Live pricing and a named win with willingness to pay.
Team signal	Strong founder domain depth; engine coverage unverified.	Yes, for the founder.	A named core team and engine commit authorship.
Launch signal	Modest reception and active release cadence.	Yes, as a starting point.	Retention cohorts showing usage tracking releases.

Competitive Landscape

Segment	Examples	Customer Alternative	Pressure On Company
Backend-as-a-service incumbents	Supabase, Convex, Firebase.	Build on an installed backend and add memory or graph features.	A well-distributed base can absorb the agent-state workload as a feature.
Graph and multi-model databases	Neo4j, SurrealDB, TypeDB, Dgraph, MongoDB.	Use an established graph or document database, some of which advertise agent-context features.	Incumbents out-resource BlitzGraph by orders of magnitude and one already markets a "context layer for AI agents."
Agent-memory layers	Mem0, Cognee.	Compose a memory layer onto an installed backend.	The memory mindshare and ranking sit with these layers, which capture the value BlitzGraph creates.

Category Visibility Snapshot

Google Search / United States"agent memory layer database"

Captured page-one results (2026-06-24): A memory-layer peer ranked first, alongside other memory and database domains; BlitzGraph did not appear.
Was the company present?: No.

Google Search / United States"AI-native backend"

Captured page-one results (2026-06-24): Established backend and database domains appeared; BlitzGraph did not appear.
Was the company present?: No.

Google Search / United States"backend-as-a-service for AI agents"

Captured page-one results (2026-06-24): Backend-as-a-service incumbents and agent platforms appeared; BlitzGraph did not appear.
Was the company present?: No.

Google Search / United States"graph database for AI agents"

Captured page-one results (2026-06-24): Graph incumbents appeared; BlitzGraph did not appear.
Was the company present?: No.

Visibility read (inference, low confidence): BlitzGraph has effectively no category discovery presence today, and the relevant queries are owned by incumbents and memory layers. For a very early company this could be the normal early state rather than absence, but it reinforces the distribution and value-capture gap.

Pricing And Competitive Benchmark

Alternative	What It Offers	Public Price Signal	Price Vs. This Company	Evidence Label
Supabase	Postgres-based backend-as-a-service.	Pro tier about $25 per month, with higher team tiers (Supabase pricing).	BlitzGraph has no public price, so relative pricing is `unknown`.	`source-backed fact` for the pricing page contents.
Convex	Application backend aimed at agent workloads.	Pro tier about $25 per developer per month, with higher business tiers (Convex pricing).	BlitzGraph has no public price, so relative pricing is `unknown`.	`source-backed fact` for the pricing page contents.
Mem0	Agent-memory layer, composable onto a backend.	Tiers from about $19 to several hundred dollars per month (Mem0 pricing).	BlitzGraph has no public price, so relative pricing is `unknown`.	`source-backed fact` for the pricing page contents.
Neo4j	Pure-graph database capacity pricing.	Capacity-based pricing per gigabyte per month (Neo4j pricing).	BlitzGraph has no public price, so relative pricing is `unknown`.	`source-backed fact` for the pricing page contents.

Price positioning read: BlitzGraph's price position is unknown because there is no public pricing and payments are off, which is itself a diligence gap.
Price claims to correct or substantiate: Any "faster than a relational database" or "Supabase for graphs" claim should be tied to a named workload and an independent benchmark before it is used in pricing or positioning.

Competitive Wedge

BlitzGraph's possible wedge is a graph-native data model reached on the agent's native tool path, which could become defensible if agents accumulate graph state inside it and switching costs form. Today that is a company claim and an inference, while the pressure against it — composability, incumbent scale, and a graph peer already advertising agent context — is supported by public evidence. Until retained usage, growing dependents, or schema lock-in appear, the wedge looks copyable rather than durable.

Risks And Open Questions

Risk	Severity	Evidence	What To Ask
Value leaks to a composable stack and the platform that owns the access channel.	high	A graph peer markets a "context layer for AI agents," memory layers offer agent integrations, and roughly 80% of agentic spend concentrates with providers (SurrealDB homepage; Mem0 homepage; IDC).	Why does a developer pay BlitzGraph rather than compose alternatives?
No independent demand pull.	high	No customers, at most one package dependent, and absence from category searches (npm — @blitzgraph/client-core).	Can any retained paying or production user be referenced?
Single visible builder on a closed infrastructure-grade engine.	high	Only the founder is publicly visible and the engine is closed (GitHub — Loïc Veillard).	Who else authors and maintains the engine?
Unreadable entity and cap table; accelerator badge maps to a different entity.	high	The badge maps to the prior company and no BlitzGraph raise is public (Y Combinator — Blitz).	Is this a clean new company or an encumbered pivot?
Multi-tenant server-side code execution.	high	Server-side JavaScript runs in the shared backend and a production endpoint advertised a loopback configuration (BlitzGraph docs).	How is execution isolated per tenant?
Thin compliance and production-readiness posture.	medium	United-States-only hosting, no compliance attestation, no service-level commitment, and beta data-durability warnings (BlitzGraph privacy policy; BlitzGraph terms of service).	What is the compliance and data-residency roadmap?
A 2027 agentic project-cancellation air-pocket.	medium	More than 40% of agentic projects are predicted to be canceled by the end of 2027 (Gartner).	What is the financing road through the air-pocket?

Pre-Mortem: The Most Likely Obituary

Cause of death (one sentence): Demand-and-value-capture starvation — a closed, single-visible-builder backend never converted the access wedge into retained paid usage before the composable stack and incumbents absorbed the workload, so revenue never started and the company ran out of money and patience in the agentic air-pocket.
The causal chain (3-5 steps from today to the shutdown): The access wedge never converts because the engine is closed and the only distribution is the access channel and community interest; buyers who want agent memory compose a memory layer onto an installed backend; the single-builder constraint caps execution across the engine, query planner, semantic search, and cloud authentication while payments slip; the 2027 cancellation wave hits before pull forms and the next round, which required paid traction, never closes; and whatever value the data model created accrues to the composable stack while the founder's genuine domain depth is acqui-hired.
The earliest observable warning sign: Adoption stays flat while the founder keeps shipping — release cadence and community activity continue, but downstream dependents stay at a handful, there is still no referenceable paying customer, and payments stay off. That divergence is already the state today.
The question that defuses this chain today: Show one developer who chose BlitzGraph over composing a memory layer onto an installed backend and pays for it, and the exact reason value accrues to BlitzGraph rather than the layers around it.

Decision-Critical Unknowns

Unknown	Why It Is Decision-Critical	Best Evidence	Decision Effect
Any retained paying or production usage.	The entire upside rests on demand reaching BlitzGraph.	A usage or retention export and a referenceable account.	Confirmation opens the path toward pursue; persistent absence after the call moves it toward pass.
Why value accrues to BlitzGraph rather than the composed stack.	It is the possible structural dealbreaker.	A named win and willingness to pay against composing alternatives.	A credible answer relieves the value-capture risk; no answer hardens the pass case.
Core-engine authorship and active headcount.	A custom engine cannot rest on one person.	A commit-author log and named engineers.	Verified coverage strengthens the case; a confirmed solo engine moves it toward pass.
Entity structure, cap table, and badge accuracy.	Ownership underwritability and integrity baseline.	Incorporation documents, a cap table, and accelerator confirmation.	A clean structure makes the option actionable; a misattributed badge or an encumbered cap table is an immediate pass.
Whether agent state is a distinct bought category.	It separates a fund-returner from a capped niche tool.	A named reallocated budget line and a reproducible benchmark.	A distinct category supports the home-run case; a seam read caps the ceiling.

Diligence Questions

First Call

Question	Why It Matters	Good Evidence
How many active beta and paying spaces exist, what is retention, and can any user be referenced independently?	Resolves the highest-impact demand and value-capture gap.	A retention cohort and a referenceable account with permission.
Who, besides the founder, authors and maintains the core engine, and can you walk through commit authorship?	Resolves the single-builder key-person risk.	A commit-author log and named systems engineers.
Is BlitzGraph a new company or a pivot inside the prior entity, and what is the cap table?	Resolves ownership underwritability and badge accuracy.	Incorporation documents, a cap table, and accelerator confirmation.
What workload makes a developer choose BlitzGraph over composing alternatives, and when do payments and pricing go live?	Resolves the value-capture geometry.	A named win with willingness to pay, plus live pricing.

Follow-Up

Question	Why It Matters	Good Evidence
How is server-side code isolated across tenants, and why did the production endpoint advertise a loopback configuration?	Tests the enterprise and security gate.	A tenancy isolation architecture and an independent penetration test.
Can the performance differentiation be reproduced independently?	Tests whether the technical wedge is real or cosmetic.	A third-party-reproducible benchmark on agent-state workloads.
What is the compliance and data-residency roadmap, including a service-level commitment and a beta data-durability guarantee?	Tests the enterprise and regulated-buyer path.	A compliance roadmap, a regional option, and a service-level document.
What are runway, burn, and the financing road through the 2027 air-pocket?	Tests survivability against better-funded peers.	A disclosed runway and a credible financing and hiring plan.

Kill Criteria

Kill Criterion	Evidence That Would Trigger It
Value is structurally uncapturable.	Buyers consistently compose a memory layer onto an installed backend with no willingness to pay for the integrated product.
Zero independent paid or production adoption persists.	At the next milestone there is still no referenceable retained paying or production account.
The core engine is confirmed solo-maintained or generated, with no near-term hiring.	A commit history showing a single maintainer or a generated engine and no hiring pipeline.
The accelerator or ownership claim is hollow.	The "Backed by Y Combinator" badge is materially misattributed, or the cap table carries stale rights from the prior business that block a clean round.
Multi-tenant code execution is unsafe.	Server-side code runs in a shared interpreter without tenant isolation.
Demand is feature-shaped, not category-shaped.	Win-loss shows buyers durably default to composing alternatives, or an incumbent ships an equivalent agent-native graph layer.
No financing path through the 2027 air-pocket.	No disclosed runway or financing road against far better-funded peers.

Double-Down Criteria

Double-Down Criterion	Evidence That Would Justify More Diligence
Independent demand pull appears.	A referenceable retained paying or production customer naming the workload and the displaced alternative.
Value capture is demonstrated.	A named win against composing alternatives, with willingness to pay and live pricing.
Core-engine coverage is verified.	A commit-author log showing founder-authored engine work plus a named systems engineer.
Ownership is clean.	A clean cap table with founder majority and confirmation of the accelerator relationship.
A compounding asset begins to form.	Growing downstream dependents, schema lock-in, or retention cohorts tracking releases.

Founder Action Plan

Timeframe	Action	Output
Before the next investor conversation	Line up one or two referenceable beta or production users and export a retention or usage cohort.	A live, independently verifiable demand-pull proof point.
Before the next investor conversation	Assemble a one-page entity and cap-table fact sheet and confirm which entity holds the accelerator relationship.	A cap-table document and accelerator confirmation.
Before the next investor conversation	Name the core-engine team and prepare a commit-author log, including the co-founder's status.	An engineering roster and commit authorship.
Near-term	Enable payments, publish pricing and a value metric, and capture first conversions and one named win against composing alternatives.	A live pricing page, checkout, and a win-loss note.
Near-term	Document the tenancy isolation architecture, remediate the production configuration, and scope a penetration test.	An isolation document and a corrected production endpoint.
Near-term	Commission or publish an independent benchmark on agent-state workloads.	A reproducible third-party benchmark.
Next one to two milestones	Publish a compliance roadmap and a credible senior-infrastructure hiring plan.	A compliance roadmap and a hiring plan.
Next one to two milestones	Ship one disclosed-immature component to general availability and show a forming compounding asset.	A general-availability release and ecosystem or retention metrics.

Decision

Screen: hold.
Confidence: medium.
Rationale: BlitzGraph pairs a genuinely domain-deep, accelerator-lineage technical founder with a non-obvious distribution idea on a very large agentic tailwind, but every load-bearing commercial fact is company-stated or unknown, a dedicated second pass over the public record upgraded none of them, and the binding constraints — demand and value capture — are exactly where the evidence is weakest.
What would move this to pursue: One independently referenceable, retained paying or production customer, plus a named workload where a developer chose BlitzGraph over composing alternatives with willingness to pay, plus evidenced core-engine authorship with a named second engineer, plus a clean cap table with the accelerator relationship clarified.
What would move this to pass: Any one confirmed kill criterion — in particular, if after the call there is still no referenceable paying customer and no credible value-capture answer, the disciplined outcome is pass rather than a third hold; a materially misattributed accelerator badge or an encumbered cap table is an immediate pass.
Recommended next step: One decisive, time-boxed founder call with pre-committed pursue-or-pass triggers and a pass-by date if no call with the proof packet occurs.
Founder preparation standard: Bring evidence a skeptical infrastructure investor can route, verify, and model — referenceable users, a value-capture win, engine authorship, and ownership documents — not additional narrative.

How We Would Miss This One

The miss scenario: BlitzGraph could be the rare case where the founder's genuine domain depth, the non-obvious access wedge, and the real agentic tailwind compound into a default agent backend, and the first referenceable paying customer appears just after a cautious committee steps away, making today's invisibility the normal early state of a new category rather than its absence.
Flip conditions: A referenceable retained paying or production customer, a named value-capture win against composing alternatives, evidenced core-engine authorship with a named second engineer, and a clean cap table with the accelerator relationship clarified would flip the case toward pursue.
Revisit trigger / date: Revisit immediately after a time-boxed founder call, or by 2026-09-24 if no call with the proof packet occurs.

Source Log

BlitzGraph homepage
blitzgraph.com
Positioning as an AI-native graph backend, the access wedge, public-beta status, and the accelerator badge.
Retrieved 2026-06-24company-controlled pagehigh
BlitzGraph docs
blitzgraph.com
Product architecture, packaging, and the server-side code-execution surface.
Retrieved 2026-06-24technical documentationhigh
BlitzGraph changelog
blitzgraph.com
Beta release cadence and repeated rework of the agent-access integration.
Retrieved 2026-06-24company-controlled pagehigh
BlitzGraph privacy policy
blitzgraph.com
Legal operator, United-States-only hosting, subprocessors, and self-asserted security.
Retrieved 2026-06-24company-controlled pagehigh
BlitzGraph terms of service
blitzgraph.com
Usage-based billing model, no service-level commitment, and beta data-durability terms.
Retrieved 2026-06-24company-controlled pagehigh
Hacker News Show HN thread
news.ycombinator.com
Founder narrative (claims only) and the launch reception.
Retrieved 2026-06-24founder forum post (Hacker News)medium
npm @blitzgraph/client-core
npmjs.com
Published client package with at most one downstream dependent.
Retrieved 2026-06-24public database (npm)high
GitHub — Loïc Veillard profile
github.com
Founder public profile, published packages, and the fork-versus-original repository split.
Retrieved 2026-06-24public profile (GitHub)high
GitHub — SurrealDB issues by lveillard
github.com
Founder graph-database domain depth (56 issues, independently visible).
Retrieved 2026-06-24technical artifact (GitHub)high
LinkedIn — Loïc Veillard
linkedin.com
Founder identity, location, and a thin public profile.
Retrieved 2026-06-24public profile (LinkedIn)medium
Y Combinator — Blitz (S22)
ycombinator.com
Accelerator lineage for the prior company, the founder bio, and the co-founder name.
Retrieved 2026-06-24startup database (Y Combinator)high
SEC EDGAR — company search
sec.gov
No securities issuer surfaced for the operating company.
Retrieved 2026-06-24public registry (SEC EDGAR)medium
IDC agentic AI forecast
my.idc.com
Agentic spend reaching $1.3 trillion by 2029 and provider concentration.
Retrieved 2026-06-24analyst estimate (IDC)high
Gartner agentic AI cancellation prediction
gartner.com
The 2027 agentic project-cancellation air-pocket and the adoption tailwind.
Retrieved 2026-06-24analyst estimate (Gartner)high
Gartner public cloud forecast 2025
gartner.com
Adjacent cloud and platform-as-a-service ceilings (context, not direct market).
Retrieved 2026-06-24analyst estimate (Gartner)high
MongoDB FY2025 Form 10-K
investors.mongodb.com
Data-management software market outer boundary.
Retrieved 2026-06-24public-company filing (SEC)high
MongoDB Q3 FY2026 results
investors.mongodb.com
Public database-as-a-service comparable with software-like margins at scale.
Retrieved 2026-06-24filing / investor relations (MongoDB)high
Neo4j $200M ARR press release
neo4j.com
Pure-graph value pool capping below general database-as-a-service (company claim).
Retrieved 2026-06-24company-controlled pagemedium
Supabase pricing
supabase.com
Backend-as-a-service pricing benchmark.
Retrieved 2026-06-24company-controlled pagehigh
Convex pricing
convex.dev
Agent-backend pricing benchmark.
Retrieved 2026-06-24company-controlled pagehigh
Mem0 pricing
mem0.ai
Agent-memory pricing benchmark; composable-substitute pricing.
Retrieved 2026-06-24company-controlled pagehigh
Neo4j pricing
neo4j.com
Graph-database capacity pricing benchmark.
Retrieved 2026-06-24company-controlled pagehigh
Supabase homepage
supabase.com
The installed backend a developer could compose alternatives onto.
Retrieved 2026-06-24company-controlled pagemedium
Convex homepage
convex.dev
A direct agent-backend peer's positioning.
Retrieved 2026-06-24company-controlled pagemedium
SurrealDB homepage
surrealdb.com
A graph peer marketing a "context layer for AI agents."
Retrieved 2026-06-24company-controlled pagemedium
Mem0 homepage
mem0.ai
A composable agent-memory substitute.
Retrieved 2026-06-24company-controlled pagemedium
Cognee homepage
cognee.ai
A composable agent-memory substitute.
Retrieved 2026-06-24company-controlled pagemedium

Access limitations: Public research could not retrieve BlitzGraph's customer list, usage or retention data, pricing, billing status, incorporation documents, cap table, or any BlitzGraph-specific financing record, and the engine is closed so its authorship and team size could not be verified. Job-board datasets could not be retrieved, so hiring conclusions rest on search checks and the absent careers page. Category search results and unavailable profiles were used only to mark gaps, not as client-safe proof, and self-run performance claims could not be independently benchmarked. These limitations mean this report can support a gated first call, but it cannot validate demand, value capture, team coverage, ownership, or enterprise readiness without founder-provided documents and reference checks.

Public disclaimer: This is an independent public teardown based solely on publicly available information retrieved on 2026-06-24. It is research support, not investment, legal, tax, or financial advice. Company-provided statements are labeled as claims, not verified facts. BlitzGraph 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#