InsForge (InsForge, Inc., Delaware corporation)

June 25, 2026

Public disclaimer: This is an independent public teardown based solely on publicly available information retrieved on 2026-06-25. It is research support, not investment, legal, tax, or financial advice. Company-provided statements are labeled as claims, not verified facts. InsForge did not participate in or review this report. Errors and omissions are possible; final decisions remain with the reader.

The backend AI agents want to install is also the backend the incumbents want to own. InsForge has the kind of public developer pull that makes a pre-seed infrastructure deal impossible to ignore: about 12,000 GitHub stars on an open-source agent-native backend, plus a Y Combinator profile and a hosted cloud pitch. The company says agents can spin up Postgres, auth, storage, functions, hosting, and model routing without a human stitching the stack together. That is exactly the dream. It is also exactly where platform gravity, commodity primitives, and founder-controlled numbers can turn a beautiful wedge into somebody else's feature. The public record shows real attention; it does not show whether the value sticks. Read the teardown before the first call turns pull into conviction.

What The Company Does

InsForge markets an agent-native backend platform for AI coding agents: database, authentication, storage, edge functions, hosting, a model gateway, compute, vectors, payments, and messaging, exposed through docs, command-line setup, agent skills, and Model Context Protocol surfaces (InsForge homepage; InsForge docs; InsForge agent doc) company claim. Independent public sources confirm the company is a Y Combinator Spring 2026 startup, that InsForge, Inc. appears on the company's terms as the legal operator, and that the open-source repository exists under an Apache 2.0 license (Y Combinator; InsForge terms; InsForge GitHub) source-backed fact.

The likely user is an AI-forward developer or small engineering team building agent-generated applications. The likely buyer starts as developer self-serve and may need to become an engineering or platform budget owner if the product moves into production. Public pricing shows Free, Pro at $25 per month plus compute credits and overages, and Enterprise custom pricing, while project activity, revenue, and retained production usage remain company-controlled claims (InsForge pricing) source-backed fact for public price structure and company claim for the architecture and compliance assertions on the same page.

Key Takeaways

Investor reaction
InsForge earns attention because the open-source and launch-platform signals are unusually visible for a pre-seed backend company, but the public record does not yet show retained paid production usage or durable value capture.
Current verdict
hold with medium confidence.
Why investors might lean in
The company sits directly in the agentic-coding infrastructure wave, has Y Combinator Spring 2026 status, and shows independent developer interest through its public repository and Product Hunt launches (Y Combinator; InsForge GitHub; Product Hunt) source-backed fact.
Why investors might pull back
Hosted project count, monthly recurring revenue, growth, gateway margin, benchmark superiority, and the $4 million raise claim remain founder- or company-controlled, while Crunchbase and GeekWire point to lower funding figures and Supabase already markets a dedicated agent offering (Crunchbase; GeekWire; Supabase for Agents) source-backed fact for the public records and company claim for the company's higher figures.
Highest-leverage fix
Produce one exportable packet covering active versus paused projects, paid conversion, cohort retention, monthly recurring revenue, gateway take rate, per-project contribution margin, and a reconciled funding history.
Best next move
Take one time-boxed founder call only after the proof packet is shared, with a pass-by date of 2026-09-25 if the packet does not arrive.

What Makes This Potentially Fundable

The fundable version of InsForge is not "Supabase with an agent interface." It is the default backend that AI coding agents install by reflex when they build production applications: a hosted system where agent-built apps keep their database, auth, storage, compute, and model-routing state long enough for switching costs and recurring consumption to compound.

The category ceiling is not imaginary, but it is broader than InsForge's direct wedge. Gartner forecasts worldwide public cloud end-user spending at $723.4 billion in 2025, including $208.6 billion in platform services and $301 billion in integrated cloud infrastructure and platform services (Gartner) source-backed fact. Managed developer-cloud and database comparables already show multi-billion-dollar served pools through MongoDB Atlas, Cloudflare, and DigitalOcean, but no public source retrieved for this research directly sizes the "agent-native backend" category (MongoDB Form 10-K; Cloudflare Form 10-K; DigitalOcean Form 10-K) source-backed fact for comparables and inference when applied to InsForge.

The honest gap is that the company-making mechanism is unproven. Public developer attention is real; retained paid production usage, positive contribution margin, default placement, and a compounding asset are not visible from public sources.

The Four Holes To Close Before Fundraising

Hole	Investor Fear	What To Bring
Usage and revenue are still founder-controlled claims.	Investors will fear the 32,000 to 35,000 project count includes abandoned free sandboxes rather than retained production apps.	Product analytics and billing exports with metric definitions, weekly-active projects, paid conversion, retention, and monthly recurring revenue.
Value capture on the model gateway is undisclosed.	Routing through OpenRouter may create convenience without enough gross margin to matter.	Gateway take rate, token cost of goods sold, blended gross margin by plan, and any second durable margin source.
The dedicated-server architecture may be capital-intensive.	If each mostly free project carries dedicated compute cost, public project growth could worsen unit economics.	Per-project cloud cost, paid conversion, project inactivity data, and contribution margin by cohort.
Funding and founder narrative need reconciliation.	A public $4 million raise claim conflicts with a $2 million Crunchbase trace and a $1.5 million GeekWire figure, and this can taint the rest of the story.	Current cap table, financing documents, securities filings if any, and a one-page explanation reconciling all public figures.

Decision Snapshot

One-sentence company description: InsForge is a Y Combinator Spring 2026 startup building an agent-native backend-as-a-service for AI coding agents to create and run application backends.
Screen: hold.
Confidence: medium.
Suggested next action: One gated founder call with the proof packet supplied before the meeting.
Why this matters now: Agentic coding workflows are forming quickly, but incumbent backend and cloud platforms are already adapting their products to the same wedge.
Investor-readiness diagnosis: Strong public attention, incomplete underwriting evidence.
Best founder use of this report: Convert company-controlled traction and funding claims into auditable evidence before broadening fundraising.

IC Disagreement Map

The public evidence was reviewed independently across five partner lenses: 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 durability.

Power-law partner
pursuelow confidence
The default-agent-backend path is plausible enough to justify buying a small pre-seed option before the proof is fully priced, because the independent developer pull is real and no fatal flaw is confirmed.
Would flip on
Retained paid production usage plus defensible gateway or hosted-data margin would move them to higher conviction; confirmed thin capture or incumbent absorption would move them to pass.
Prepared-mind partner
holdmedium confidence
Timing and pull are real, but the novelty claim is contradicted by incumbent agent offerings and value capture is unproven.
Would flip on
Retained paid cohorts, a reproducible agent-operation benchmark, and non-trivial margin would move them toward pursue; sandbox-dominated usage or benchmark parity would move them toward pass.
Founder-jockey partner
pursuemedium confidence
Public build velocity, security hardening, Product Hunt cadence, and an evidenced CTO builder justify a gated first call despite team-coverage and integrity gaps.
Would flip on
Founder-attributed core commits, an infrastructure/security bench, and a clean funding story would raise conviction; a confirmed misstatement or one-builder-only reality would move them toward pass.
Risk-reduction partner
holdmedium confidence
Gateway spread, per-project cost, unverified traction, and funding consistency are open kill risks that must be tested before deeper diligence.
Would flip on
Positive per-project contribution margin, defensible gateway spread, and funding reconciliation would move them toward pursue; any confirmed negative on those would move them to pass.
Long-horizon partner
holdmedium confidence
The compounding asset is unmeasured, the license is permissive, and follow-on risk is high against better-capitalized incumbents.
Would flip on
Six to twelve months of production retention, rising net revenue retention, durable default placement, and non-trivial margin would move them toward pursue; sandbox retention or vendor-controlled displacement would move them toward pass.

All five partners confirmed their Round 2 votes after the evidence-request round; no partner changed vote or confidence because the public record added no positive closure on usage, margin, benchmark, team coverage, or funding reconciliation. The split remained 2 pursue and 3 hold, so the disagreement is substantive rather than cosmetic.

Scenario Range

Scenario	What The Company Looks Like In 3-5 Years	Falsifiable Trigger To Watch	Earliest Evidence
Strikeout	InsForge becomes a polished ergonomics layer that helped prove the agent-backend need, but incumbents absorb the wedge, free projects dominate, gateway margin stays thin, and the company winds down or sells for talent.	Active paid projects are a small and non-growing share of headline projects, gateway economics are near pass-through, and a primary backend or agent platform ships a displacing native alternative.	Analytics and billing exports, gateway cost data, and incumbent product launches.
Base	InsForge becomes a useful backend for small AI-forward teams, with modest self-serve revenue and some retained projects, but not enough durable production workload gravity to return a fund.	Paid cohorts exist and contribution margin is positive on paid plans, but net revenue retention is roughly flat and usage remains small-team or prototype-heavy.	Retention bridge, pricing and gross-margin data, and reference calls from small production users.
Home run	InsForge becomes the default backend that agent-built production apps live and bill on, with retained workloads, production-data gravity, and default-tool placement compounding into switching costs.	Production-workload share and net revenue retention rise with tenure, removing InsForge breaks agent-built workflows, and default placement persists through vendor changes.	Independent cohort curves, durable integration evidence, and design-partner references treating InsForge as non-removable.

What Investors Will Test

What We Looked For	Current Read	Investor Implication	Founder Prep Priority
Independent proof outside company-controlled surfaces	GitHub, Product Hunt, Hacker News, Y Combinator, Crunchbase, GeekWire, and MindWorks provide real external context; usage and revenue still do not clear that bar.	The company earns a call, not diligence momentum.	Bring raw exports, not screenshots, for usage, revenue, and retention.
A fund-returning path	The default-agent-backend thesis is coherent, but no public evidence shows production workloads compounding.	Investors should keep the upside alive while refusing to underwrite it yet.	Define the home-run metric and show its current cohort trend.
Value capture	Public pricing is visible, but gateway spread and per-project margin are unknown.	Public developer pull could create costs before it creates value.	Bring model-gateway cost and contribution-margin data.
Founder-market fit on the hardest build	Tony Chang has public technical traces, while Hang Huang's engineering footprint and senior infrastructure/security bench remain unresolved.	Team quality is a lean-in factor with a hard verification gate.	Bring commit attribution, role references, and a named reliability/security owner.
Dissent and kill discipline	The committee preserved a genuine pursue-versus-hold split.	The first call must have a fuse and pass triggers.	Agree before the call that negative or withheld proof closes the case.

Claim Reconciliation: Inconsistencies Investors Will Catch

Claim Or Metric	Where It Appears	Conflicting / Unreconciled Versions	Why Investors Flag It	How To Reconcile
Total capital raised	Founder-controlled public profile material, Crunchbase, GeekWire, and MindWorks portfolio context	Founder-controlled material says $4 million; Crunchbase lists about $2 million; GeekWire reported $1.5 million to date; MindWorks confirms portfolio status without amount (Crunchbase; GeekWire; MindWorks).	Inconsistent financing claims create an integrity and runway question before product diligence begins.	Provide cap table, round documents, securities filings if applicable, and explain whether SAFEs, grants, or committed capital are being counted differently.
Team size	Y Combinator profile and LinkedIn company data	Y Combinator lists team size 6, while LinkedIn company data shows a larger employee count (Y Combinator; LinkedIn).	Investors need to know whether the difference is employees, contractors, contributors, advisors, or stale profile data.	Share a current roster with employment type, role, start date, and responsibilities.
"First agent-native backend" positioning	InsForge comparison pages and public descriptions	InsForge markets a primacy claim, while Supabase, Neon, Appwrite, and Firebase show agent or Model Context Protocol surfaces (InsForge vs Supabase; Supabase for Agents; Neon; Appwrite; Firebase).	A false or overstretched superlative makes the company look underprepared for competitive diligence.	Replace the superlative with a narrow, testable differentiation claim.
Founder build coverage	Founder bios, public GitHub traces, and university/public profile evidence	Tony Chang has visible GitHub and UIUC traces; Hang Huang's candidate GitHub account does not publicly connect to InsForge and shows older repos (Tony Chang GitHub; Tony Chang repositories; UIUC CS; Hang Huang GitHub; Hang Huang repositories).	For a hard multi-tenant cloud, investors will test who personally owns reliability, security, and core infrastructure.	Provide commit attribution, primary account confirmation, and references for hard-build ownership.

Diligence Findings: Issues To Fix Before You Raise

3 high3 medium

Usage and revenue traction are not independently evidenced.
high
Commercial and KPI diligence
Fix
Prepare product analytics and billing exports with cohort definitions.
Gateway margin and per-project economics are undisclosed.
high
Financial and technical diligence
Fix
Provide cost model, take rate, gross margin, and contribution margin.
Funding figures conflict across public records.
high
Legal and financial diligence
Fix
Reconcile every figure to cap table and financing documents.
The "first agent-native backend" claim is contradicted by competitor surfaces.
medium
Market and competitive diligence
Fix
Replace the claim with a defensible, evidence-backed wedge.
Enterprise compliance is marketed but not publicly evidenced.
medium
Security and procurement diligence
Fix
Provide dated SOC 2, HIPAA, business associate, and security artifacts or stop implying attainment.
Founder and team coverage for reliability and security is unresolved.
medium
Team and technical diligence
Fix
Show founder-attributed commits, current roster, and named security or site-reliability ownership.

Data Room Readiness Checklist

Procurement-readiness score: partial.
Rationale: InsForge has public legal, pricing, terms, privacy, and security-policy surfaces, but no public audit report, business associate agreement, pen-test summary, subprocessor detail suitable for enterprise procurement, or evidence that SOC2 and HIPAA claims are already attained.

Data Room Area	What Investors Expect	Current Read	Status	Priority
Company overview & corporate	One-pager, incorporation, good standing, structure	Public terms name InsForge, Inc.; current charter, good standing, board consents, and entity history are private.	partial	high
Financials	Monthly profit and loss, three-year model, burn, runway	Revenue and project counts are company-controlled public claims; no financial export exists publicly.	missing	critical
Cap table & funding history	Clean cap table, prior rounds, SAFEs or notes, valuation	Public funding figures conflict across sources.	missing	critical
Legal & IP	Bylaws, board consents, IP assignments, material contracts	Apache 2.0 repository and terms are public; IP assignment and customer contracts are private.	partial	high
Product & technology	Roadmap, architecture overview, security and compliance docs	Docs, pricing, privacy, terms, and security policy are public; architecture review and audit evidence are private.	partial	high
Team	Org chart, key employment/advisor agreements, vesting	Founders are public; team size and hard-build ownership are unresolved.	partial	high
Customers & traction	Retention cohorts, revenue bridge, pipeline, references	No named paying customers or independent retained-usage evidence is public.	missing	critical
Market & competition	Market model, competitive landscape, pricing	Broad market comparables and competitor pages are public; direct category size and differentiated benchmark remain unresolved.	partial	medium

First-Call Agenda For The Startup

Time	Topic	Founder Goal	Evidence To Bring
0-10 min	Clarify the one-sentence wedge	Replace broad "agent-native" language with the specific workflow where InsForge wins.	Short demo and a benchmark outline.
10-25 min	Usage and revenue reality	Show that public pull converts into retained paid production usage.	Analytics and billing exports with cohort definitions.
25-40 min	Unit economics and value capture	Prove that gateway routing and per-project architecture create positive margin.	Gateway cost, take rate, compute cost, and contribution-margin model.
40-50 min	Team, security, and operating ownership	Make hard-build coverage legible.	Org chart, commit attribution, on-call ownership, security artifacts.
50-60 min	Funding reconciliation and next gate	Clear integrity questions and agree on go/no-go milestones.	Cap table, financing documents, and milestone plan through 2026-09-25.

Business Model And Revenue Signals

Free, Pro at $25 per month, and Enterprise custom pricing are public.
source-backed factcompany pricing pagehigh
SourceInsForge pricing
Confirms offer structure, not conversion.
Free projects pause after one week of inactivity.
source-backed factcompany pricing pagehigh
SourceInsForge pricing
This makes active versus paused project data essential.
Hosted project count, weekly growth, and monthly recurring revenue are public self-narrative claims.
company-controlled surfaceslow
SourceInsForge homepage LinkedIn X
No billing or product analytics export was public.
Model Gateway routes via OpenRouter with InsForge-held keys.
technical documentationhigh
SourceInsForge model gateway docs
Margin and cost are undisclosed.
Terms allocate responsibility for AI-agent actions to customers.
source-backed factcompany legal pagehigh
SourceInsForge terms
Relevant for enterprise risk and contract negotiation.

Revenue and profit are not independently available. Any revenue, growth, paid conversion, or margin statement beyond public pricing should be treated as a founder-call item until the company provides billing and cost exports.

Revenue Quality Checklist

Before hold can become pursue, InsForge should be ready to show the raw data investors will rebuild: monthly recurring revenue, annual recurring revenue, paid accounts, paid conversion from total projects, gross revenue retention, net revenue retention, churn, customer acquisition cost, lifetime value, payback, gross margin by workload, gateway take rate, and per-project contribution margin. These figures need to tie across the deck, billing system, product analytics, data-room model, and source documents.

Funding And Ownership Context

Item	Public Read	Evidence Label	Diligence Request
Total raised / rounds	Crunchbase lists about $2 million, GeekWire reported $1.5 million to date in 2025, and MindWorks confirms InsForge as a portfolio company; founder-controlled material says $4 million.	`source-backed fact` / `company claim` conflict	Reconcile all figures to cap table, instruments, and filing status.
Instruments (equity / SAFE / notes)	Not public.	`unknown`	Provide financing documents and instrument summary.
Investors	Crunchbase names Y Combinator, MindWorks, and Baidu Ventures; MindWorks lists InsForge in its portfolio.	`source-backed fact`	Confirm round participants and amounts.
Post-money valuation	Not public.	`unknown`	Provide valuation cap or priced-round terms.
Cap table / ownership	Not public.	`unknown`	Provide fully diluted cap table and option pool.
Burn & runway	Not public.	`unknown`	Provide cash, burn, runway, and hiring plan.
Next-round plan	Not public.	`unknown`	Provide target raise, valuation expectation, and use of funds.

SEC EDGAR was checked and showed no matching Form D indexed under InsForge. This absence does not disprove a private raise or filing under a different legal name, but it increases the need for founder-provided financing reconciliation.

Founder And Team

Hang Huang
Co-founder and CEO. Education is supported by MindWorks; Amazon and esports claims remain founder-controlled.
Y Combinator LinkedIn X MindWorks
Evidencesource-backed fact for identity and education; company claim for role-scope claims.
Confidencemedium
Tony Chang
Co-founder and CTO. Public GitHub and UIUC traces support technical identity.
Y Combinator LinkedIn GitHub UIUC CS X
Evidencesource-backed fact for identity and technical traces; company claim for Databricks scope.
Confidencehigh for identity, medium for scope
Team size
Public sources conflict between Y Combinator team size and LinkedIn company headcount.
Y Combinator LinkedIn
Evidencesource-backed fact conflict
Confidencemedium
Senior site-reliability or security bench
No named senior SRE/security owner was surfaced publicly.
Public profile and hiring checks
Confidencemedium

Founder Competency Coverage

Domain depth
Hang Huang
claimed
Tony Chang
evidenced
What The Evidence Is
Hang's product and GTM role is public but role scope needs references; Tony has GitHub and UIUC traces plus InsForge organization membership.
Technical build capability
Hang Huang
unknown
Tony Chang
evidenced
What The Evidence Is
The candidate Hang GitHub account is not publicly tied to InsForge; Tony's GitHub account is tied to the InsForge organization.
Product
Hang Huang
claimed
Tony Chang
claimed
What The Evidence Is
Company and accelerator materials present both founders as building InsForge; independent product-scope references are limited.
GTM / sales
Hang Huang
claimed
Tony Chang
unknown
What The Evidence Is
Company-controlled materials describe founder resilience and traction, but no sales references were public.
Leadership / hiring
Hang Huang
unknown
Tony Chang
unknown
What The Evidence Is
Team-size conflict remains unresolved and no org chart was public.
Fundraising history
Hang Huang
claimed
Tony Chang
claimed
What The Evidence Is
Public records show pre-seed financing signals, but the $4 million claim is unreconciled.
Prior founding outcomes
Hang Huang
unknown
Tony Chang
unknown
What The Evidence Is
No prior founding outcome was independently verified.

Reliable multi-tenant backend operations
high
Team coverage today
One evidenced technical founder; broader operating bench unknown.
What would close it
Org chart, on-call ownership, incident history, and named reliability/security owner.
Security and compliance for customer production data
high
Team coverage today
Security policy and some hardening commits exist; audit evidence is not public.
What would close it
SOC 2 report, pen-test, architecture review, and security leadership.
Agent-native product differentiation
medium
Team coverage today
Company claims CLI/skills/Model Context Protocol edge; independent benchmark absent.
What would close it
Blind benchmark and customer evidence that agent workflows work better on InsForge.

Public Professional Footprint

These public checks 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)
Hang Huang
unknown
Tony Chang
evidenced
What The Public Record Shows
Hang's candidate GitHub account has older repos and no public InsForge tie; Tony's GitHub account and repository page are public and linked to the InsForge organization (Hang Huang GitHub; Hang Huang repositories; Tony Chang GitHub; Tony Chang repositories).
Professional social content
Hang Huang
claimed
Tony Chang
claimed
What The Public Record Shows
Public profile and company surfaces repeat founder and traction claims; repetition is not validation (LinkedIn; X).
Education / credentials
Hang Huang
evidenced
Tony Chang
evidenced
What The Public Record Shows
MindWorks supports Hang's Yale and Northeastern education; UIUC supports Tony's computer-science affiliation and Meta internship trace (MindWorks Hang Huang; UIUC CS).
Publications / patents / certifications
Hang Huang
unknown
Tony Chang
unknown
What The Public Record Shows
No decision-relevant public patents, certifications, or publications were surfaced.

Net read: the founder footprint is good enough for a first call, but not enough to clear the hard-infrastructure team gate without references and commit attribution.

Founder-Market Fit Read

What is promising: The company shows a visible builder and launch cadence in a category that rewards fast technical iteration, and Tony Chang has public technical traces.
What is missing: Founder-attributed ownership of the reliability and security core, senior operating bench, and a clean funding and credential story.
What to prepare: Role references, commit attribution, current roster, on-call/security ownership, cap table, and a corrected financing narrative.

InsForge, Inc.
Public legal operator named in terms.
InsForge terms
Evidencesource-backed fact
Confidencehigh
Follow-upConfirm incorporation, good standing, IP ownership, and cap table.
InsForge open-source repository
Public Apache 2.0 core tied to the hosted product.
InsForge GitHub
Evidencesource-backed fact
Confidencehigh
Follow-upConfirm contributor ownership, license strategy, and enterprise boundary.

Traction

Customer Status Table

Unnamed hosted project users
Company-controlled project counter and social claims; no named customer confirmation was public.
No proofClaim onlyUnknownUnknownUnknownlow

G2, Trustpilot, Product Hunt reviews, public case-study pages, and customer-controlled references were checked where available and did not surface a named paying customer confirmation. These absences carry moderate weight because production references are a central gate for infrastructure diligence.

Traction Signals

About 12,000 GitHub stars, about 1,000 forks, active commits, and Apache 2.0 licensing.
source-backed facthigh
SourceInsForge GitHub
Distinguish repository interest from hosted product usage.
Product Hunt launch rankings and multiple launches.
source-backed facthigh
SourceProduct Hunt
Determine whether launch attention converts to retained paid usage.
Show HN discussion.
source-backed facthigh
SourceHacker News
Determine whether developer interest becomes production adoption.
Company social audience and headcount signals.
source-backed fact for profile metadata; company claim for posts.medium
SourceLinkedIn X
Validate traction claims through billing and analytics exports.

G2, Trustpilot, app stores, and third-party review platforms were checked and showed no material InsForge product listing. These absences carry limited weight for a developer-infrastructure pre-seed, but they leave production-customer proof unavailable.

Hiring And Org Momentum

LinkedIn company headcount differs from Y Combinator team size.
source-backed fact conflictLinkedIn company data showed a larger employee count than Y Combinator's team-size field.
SourceLinkedIn Y Combinator
The team shape needs reconciliation before investors underwrite hard-build coverage.

Indeed, Glassdoor, LinkedIn Jobs, and the company site were checked and showed no public matching job listing. These absences carry little weight for a small early-stage company, but they leave hiring momentum and senior infrastructure/security coverage unresolved.

Role-mix read: unknown; the public record does not show enough roles to infer a hiring mix.
Momentum read (growth / steady / contraction / unknown): unknown from public hiring data, with visible product and repository activity but no public staffing expansion signal.

Traction Quality Read

Traction Dimension	Current Status	Good Enough For First Call?	Needed For Deep Diligence
Developer attention	Strong for stage through GitHub and Product Hunt.	Yes.	Conversion funnel from attention to active hosted usage.
Hosted product adoption	Company-claimed only.	Only as a question.	Active, paid, retained cohort export.
Revenue	Company-claimed only.	Only as a question.	Billing export and revenue bridge.
Retention	Not public.	No.	Six to twelve months of cohort retention and net revenue retention.
Customer references	Not public.	No.	Named production references willing to speak.

Competitive Landscape

Segment	Examples	Customer Alternative	Pressure On Company
Direct backend-as-a-service and agent-backend platforms	Supabase, Firebase, Nhost, Appwrite, Convex, Neon, PlanetScale	Use a mature managed backend or database platform and wait for agent features to arrive.	InsForge must prove an agent-native surface that incumbents cannot copy quickly.
Multi-vendor substitute stack	Clerk, Neon, Vercel, Cloudflare, AWS primitives	Compose best-of-breed auth, database, hosting, and serverless tools.	InsForge must make bundling clearly better than mature specialized tools.
Platform and cloud incumbents	Google/Firebase, AWS Amplify, Vercel, Databricks/Neon	Stay inside incumbent cloud distribution and procurement.	Incumbents own trust, distribution, and budget channels.

Category Visibility Snapshot

Google / USInsForge vs Supabase

Captured page-one results (2026-06-25): InsForge, Supabase, Reddit, Hacker News, YouTube, and migration content appeared in captured page-one results.
Was the company present?: Yes

Google / USagent backend platform Supabase Firebase coding agents

Captured page-one results (2026-06-25): Supabase appeared, while Firebase was referenced in snippets and InsForge did not appear in the top organic results.
Was the company present?: No

Google / USSupabase vs Convex vs InsForge

Captured page-one results (2026-06-25): InsForge, Supabase, Convex, Reddit, and comparison pages appeared.
Was the company present?: Yes

Google / USbackend as a service for AI coding agents

Captured page-one results (2026-06-25): Reddit discussion mentioning InsForge, Supabase, Base44, Back4app, and Xano appeared.
Was the company present?: Yes

Google / USInsForge vs Firebase

Captured page-one results (2026-06-25): InsForge comparison pages, Firebase-related results, Slashdot comparison content, and LinkedIn appeared.
Was the company present?: Yes

Google Search in the United States was checked for generic agent-backend queries and showed InsForge absent from the top organic results on the main Supabase/Firebase category query. This is a low-confidence visibility signal, not market share.

Visibility read (inference, low confidence): InsForge is visible around branded comparison intent but has not yet clearly won generic category demand against Supabase and other incumbents.

Pricing And Competitive Benchmark

Alternative	What It Offers	Public Price Signal	Price Vs. This Company	Evidence Label
InsForge	Agent-native backend with Free, Pro, and Enterprise tiers.	Pro at $25 per month plus compute credits and overages.	Baseline.	`source-backed fact` for public pricing (InsForge pricing)
Supabase	Postgres platform with auth, storage, edge functions, and agent positioning.	Pro at $25 per month and Team at $599 per month.	Similar entry price, much more mature platform.	`company claim` from Supabase pricing
Firebase	Google-backed app platform with no-cost and pay-as-you-go pricing.	Spark no-cost and Blaze pay-as-you-go.	More flexible and incumbent-backed; direct comparison depends on usage mix.	`company claim` from Firebase pricing
Neon	Serverless Postgres with free and paid tiers.	Launch and Scale tiers plus metered compute and storage.	Cheaper database-first entry, narrower bundle.	`company claim` from Neon pricing
Clerk + Neon + Vercel	Best-of-breed auth, database, and deploy stack.	Multiple vendor prices rather than one bundle.	Potentially more complex, but mature and modular.	`inference` from Clerk, Neon, and Vercel.

Price positioning read: InsForge matches Supabase's $25 self-serve entry point while asking investors to believe its agent-native bundling can offset incumbent maturity.
Price claims to correct or substantiate: Any claim that InsForge is cheaper or more accurate than competitors needs a reproducible task and total-cost benchmark.

Competitive Wedge

The wedge is a backend that coding agents can create and operate end-to-end, not merely a human dashboard with an agent connector. That wedge could matter if agents increasingly create production applications and repeatedly pick the same backend because it reduces failure, cost, or setup time. It remains copyable until InsForge proves an agent-success edge, retained production workloads, and durable default placement that incumbents do not control.

Risks And Open Questions

Risk	Severity	Evidence	What To Ask
Usage is sandbox-heavy rather than production-retained.	high	Project and MRR figures are `company claim`; public pull measures attention.	Ask for active, paid, retained production cohorts.
Gateway margin is thin.	high	Model Gateway routes through OpenRouter; take rate is not public.	Ask for gateway cost and spread by plan.
Per-project economics are negative.	high	Pricing page describes per-project dedicated servers; cost is not public.	Ask for per-project cost and contribution margin.
Incumbents absorb the wedge.	high	Supabase for Agents and other agent/MCP surfaces already exist.	Ask what InsForge can do that incumbents cannot copy.
Funding and credential inconsistencies undermine trust.	high	Public funding figures conflict, and founder role scope is unverified.	Ask for cap table and role references.
Enterprise compliance is aspirational.	medium	SOC2/HIPAA features are listed without public audit evidence.	Ask for current audit and BAA status.

Pre-Mortem: The Most Likely Obituary

Cause of death (one sentence): InsForge proved developers wanted an agent-operated backend, but the company could not retain production workloads or capture enough margin before incumbents absorbed the wedge.
The causal chain (3-5 steps from today to the shutdown): Public developer pull converts slowly into paid production usage; the project counter is revealed as mostly free or inactive; gateway margin remains thin and dedicated compute adds cost; Supabase, Firebase, Neon/Databricks, or a coding-agent platform bundles the same workflow; InsForge raises a hard bridge or sells for the team rather than returning venture capital.
The earliest observable warning sign: Active paid production projects are a small and non-growing share of the headline project count.
The question that defuses this chain today: "Show the retained paid production cohort and the gross margin per project."

Decision-Critical Unknowns

Unknown	Why It Is Decision-Critical	Best Evidence	Decision Effect
Active paid production usage	It determines whether the upside is real or vanity.	Product analytics and billing export with retention.	Strong retained usage would move toward pursue; sandbox-heavy usage would move toward pass.
Gateway take rate and margin	It determines whether the model gateway captures durable value.	COGS and billed-spread data.	Defensible spread would support diligence; near pass-through would trigger pass pressure.
Per-project contribution margin	It determines whether growth creates or destroys value.	Infrastructure cost model by plan and cohort.	Positive paid-plan margin supports continuation; negative margin that worsens with growth supports pass.
Funding reconciliation	It determines whether investor trust survives the first call.	Cap table and financing documents.	Clean reconciliation clears a gate; material misstatement is a pass trigger.
Team coverage for security and reliability	It determines whether the team can operate the hardest system.	Commit attribution, org chart, on-call and security ownership.	Named coverage supports founder-market fit; one-builder fragility weakens the case.

Diligence Questions

First Call

Question	Why It Matters	Good Evidence
How many projects are active, paid, non-paused, and running production workloads, and what is six-to-twelve-month retention?	It resolves the central traction uncertainty.	Product analytics and billing export with definitions.
What is gateway gross margin after OpenRouter costs, and what other durable margin exists?	It resolves value capture.	Cost per token, billed spread, and gross-margin bridge.
What is per-project contribution margin under the dedicated-server model?	It resolves capital intensity.	Project-level COGS and paid conversion by cohort.
How do the public funding figures reconcile?	It resolves a trust and ownership gate.	Cap table, financing documents, and filing status.
Who owns the reliability and security core?	It resolves hard-build team coverage.	Commit attribution, org chart, and security/on-call ownership.

Follow-Up

Question	Why It Matters	Good Evidence
Can a blind benchmark show InsForge beats Supabase for Agents, Neon, Appwrite, or Firebase on agent provisioning?	It tests whether agent-native is a real product insight.	Reproducible benchmark and methodology.
Are agent integrations durable or revocable?	It tests hostile-platform dependence.	Partner terms, registry/default-placement evidence, and fallback architecture.
What is the actual enterprise compliance status?	It tests procurement readiness.	SOC 2 report, business associate agreement, pen-test summary, and subprocessor list.
What round terms and ownership are available?	It tests whether the option is financeable.	Current financing terms, valuation, and allocation.

Kill Criteria

Kill Criterion	Evidence That Would Trigger It
Gateway value capture is near pass-through with no other margin source.	Cost and billing data show negligible spread and no hosted-workload margin.
Project usage is sandbox-dominated and non-retained.	Analytics show low active paid share and weak production retention.
Per-project contribution margin is negative with no path to positive.	Infrastructure cost model shows losses increase with project count.
A major incumbent or agent platform controls or displaces the backend workflow.	Product or partner evidence shows the integration is revocable or superseded.
Funding or credential claims are materially misstated.	Cap table or references cannot reconcile public claims.
The reliability/security core rests on one builder with no credible bench.	Org and commit evidence show no coverage beyond one technical founder.

Double-Down Criteria

Double-Down Criterion	Evidence That Would Justify More Diligence
Retained paid production usage is real and growing.	Six-to-twelve-month cohorts show rising active paid production usage and net revenue retention.
Value capture is defensible.	Gateway spread, hosted-data margin, or per-project contribution margin is positive and durable.
Agent-native edge is measurable.	Blind benchmarks show better task success, lower token use, or faster provisioning versus incumbents.
Team coverage clears the hard-build bar.	Founders and senior hires are tied to core reliability, security, and operations.
Funding history and terms are clean.	Cap table, instruments, and allocation support a meaningful, reserve-backed stake.

Founder Action Plan

Timeframe	Action	Output
Before the first call	Reconcile the public funding story.	One-page cap-table and financing summary explaining $4 million, $2 million, and $1.5 million figures.
Before the first call	Define project status and export usage cohorts.	Analytics package covering total, active, paused, paid, production, and retained projects.
Before the first call	Export billing and gross-margin data.	Monthly recurring revenue, paid conversion, gateway take rate, token costs, and per-project contribution margin.
Early diligence	Make hard-build ownership legible.	Commit attribution, org chart, security ownership, on-call process, and hiring plan.
Early diligence	Replace primacy claims with evidence-backed differentiation.	Benchmark methodology and updated positioning against Supabase, Firebase, Neon, Appwrite, and Convex.
Before broader fundraising	Assemble procurement artifacts.	SOC 2 status, business associate agreement status, subprocessor list, data retention policy, pen-test or architecture review, and incident-response policy.

Decision

Screen: hold.
Confidence: medium.
Rationale: InsForge has rare independent developer pull in a thesis-central category, but the public record leaves every underwriting variable founder-gated: retained paid usage, margin, contribution economics, funding reconciliation, and hard-build team coverage.
What would move this to pursue: Independent, billing-verified cohorts showing retained paid production usage, defensible value capture, verified founder and team coverage, and a reconciled funding history.
What would move this to pass: Any confirmed kill criterion around thin capture, negative per-project economics, sandbox-heavy usage, hostile platform dependence, material funding/credential misstatement, or inadequate hard-build team coverage.
Recommended next step: One time-boxed founder call after the proof packet is supplied, with a pass-by fuse on 2026-09-25.
Founder preparation standard: Bring raw exports and documents, not narrative slides.

How We Would Miss This One

The miss scenario: InsForge could be the fund's antiportfolio entry if public developer pull is the earliest sign of a default backend for agent-built production apps and the fund waits until retained-usage proof reprices the round.
Flip conditions: Retained paid production cohorts, rising net revenue retention, positive contribution margin, durable default placement, and verified hard-build ownership would flip the case toward pursue.
Revisit trigger / date: Revisit immediately if the company supplies the proof packet, or on 2026-09-25 if no packet arrives.

Source Log

InsForge homepage
insforge.dev
Product positioning, feature scope, public self-narrative, and company-controlled traction context.
Retrieved 2026-06-25company-controlled pagehigh for claims appearing, low for underlying traction
InsForge GitHub repository
github.com
Open-source repository, stars, forks, commits, Apache 2.0 license, and public build signal.
Retrieved 2026-06-25public repositoryhigh
Y Combinator company profile: InsForge
ycombinator.com
Batch, founded year, team size, founder roster, and startup profile.
Retrieved 2026-06-25accelerator profilehigh
InsForge documentation hub
docs.insforge.dev
Product documentation and backend scope.
Retrieved 2026-06-25company technical documentationhigh for claims appearing
InsForge pricing page
insforge.dev
Public pricing, free tier, Pro plan, enterprise claims, and dedicated-server architecture claim.
Retrieved 2026-06-25company pricing pagehigh for public pricing; medium for architecture/compliance claims
InsForge blog: Getting into YC
insforge.dev
Founder narrative and historical company-controlled traction claims.
Retrieved 2026-06-25company blogmedium
LinkedIn company: InsForge
linkedin.com
Company footprint, headcount conflict, and company-controlled post claims.
Retrieved 2026-06-25structured professional profilemedium
X profile: InsForge
x.com
Company social footprint and founder-controlled traction context.
Retrieved 2026-06-25social profile datasetmedium
Crunchbase: InsForge
crunchbase.com
Funding baseline, investors, legal/founder context, and competitor profile.
Retrieved 2026-06-25structured startup registrymedium
InsForge terms of service
insforge.dev
Legal entity, terms, AI-agent liability allocation, and commercial risk terms.
Retrieved 2026-06-25company legal pagehigh
InsForge privacy policy
insforge.dev
Subprocessors, data handling, AI partner statements, and privacy posture.
Retrieved 2026-06-25company legal pagehigh for policy text
InsForge agent doc
insforge.dev
Agent workflow, command-line setup, and agent-native operating surface.
Retrieved 2026-06-25company technical documentationhigh for public claim
InsForge changelog
insforge.dev
Release cadence and product update context.
Retrieved 2026-06-25company changelogmedium
InsForge agents page
insforge.dev
Agent integration surface and supported agent claims.
Retrieved 2026-06-25company-controlled pagemedium
InsForge model gateway docs
docs.insforge.dev
OpenRouter routing and OpenAI-compatible model gateway claim.
Retrieved 2026-06-25company technical documentationhigh for routing claim appearing
InsForge SECURITY.md
github.com
Vulnerability disclosure and security-policy posture.
Retrieved 2026-06-25public repository documentationhigh
InsForge vs Supabase comparison
insforge.dev
Company-controlled differentiation and category-novelty claims.
Retrieved 2026-06-25company comparison pagemedium
InsForge vs Firebase comparison
insforge.dev
Company-controlled Firebase comparison and positioning.
Retrieved 2026-06-25company comparison pagemedium
Product Hunt: InsForge
producthunt.com
Launch-platform traction and ranking context.
Retrieved 2026-06-25launch platformhigh
Hacker News Show HN: InsForge
news.ycombinator.com
Developer-community discussion and substitute-stack commentary.
Retrieved 2026-06-25developer forumhigh
Y Combinator Launch: InsForge
ycombinator.com
Launch narrative and accelerator-hosted company description.
Retrieved 2026-06-25accelerator launch pagehigh for claim source
GeekWire Startup Radar InsForge mention
geekwire.com
Independent press context and $1.5 million funding figure.
Retrieved 2026-06-25technology pressmedium
MindWorks Capital: InsForge portfolio
mindworks.vc
Investor relationship and portfolio status.
Retrieved 2026-06-25investor portfolio pagehigh
MindWorks: Hang Huang
mindworks.vc
Hang Huang education trace.
Retrieved 2026-06-25investor profile pagemedium
LinkedIn: Hang Huang
linkedin.com
Founder identity and public profile context.
Retrieved 2026-06-25structured professional profilemedium
LinkedIn: Tony Chang
linkedin.com
Founder identity and public profile context.
Retrieved 2026-06-25structured professional profilemedium
X profile: Hang Huang
x.com
Founder-controlled public profile and avatar provenance.
Retrieved 2026-06-25social profile datasetmedium
X profile: Tony Chang
x.com
Founder-controlled public profile and avatar provenance.
Retrieved 2026-06-25social profile datasetmedium
Hang Huang GitHub profile
github.com
Candidate CEO technical-footprint check.
Retrieved 2026-06-25public repository profilelow
Hang Huang GitHub repositories
github.com
Candidate CEO repository recency check.
Retrieved 2026-06-25public repository profilelow
Tony Chang GitHub profile
github.com
CTO technical identity and organization membership.
Retrieved 2026-06-25public repository profilehigh
Tony Chang GitHub repositories
github.com
CTO repository footprint.
Retrieved 2026-06-25public repository profilehigh
UIUC CS Celebration of Excellence 2022
cs.illinois.edu
Tony Chang education and Meta internship trace.
Retrieved 2026-06-25university pagehigh
Gartner public cloud end-user spending forecast 2025
gartner.com
Public cloud, platform services, and timing envelope.
Retrieved 2026-06-25analyst press releasehigh
Alphabet 2024 Form 10-K
sec.gov
Google Cloud adjacent scale.
Retrieved 2026-06-25public-company filinghigh
MongoDB fiscal 2025 Form 10-K
sec.gov
Atlas revenue mix and database-as-a-service comp.
Retrieved 2026-06-25public-company filinghigh
DigitalOcean 2024 Form 10-K
sec.gov
Developer-cloud revenue, gross margin, and ARPU context.
Retrieved 2026-06-25public-company filinghigh
Cloudflare 2024 Form 10-K
sec.gov
Developer/edge suite revenue, margin, and paying-customer context.
Retrieved 2026-06-25public-company filinghigh
BLS Occupational Outlook Handbook: Software Developers
bls.gov
U.S. developer-seat proxy and growth projection.
Retrieved 2026-06-25government statisticshigh
GitHub Universe 2024 press release
github.com
Developer-activity proxy and ecosystem context.
Retrieved 2026-06-25company-controlled pagemedium
Supabase homepage
supabase.com
Competitor positioning.
Retrieved 2026-06-25competitor pagehigh
Supabase for Agents
supabase.com
Direct competitive response and agent/MCP positioning.
Retrieved 2026-06-25competitor pagehigh
Supabase pricing
supabase.com
Pricing benchmark.
Retrieved 2026-06-25company pricing pagehigh
Firebase homepage
firebase.google.com
Google incumbent positioning and agent messaging.
Retrieved 2026-06-25competitor pagehigh
Firebase pricing
firebase.google.com
Pricing benchmark.
Retrieved 2026-06-25company pricing pagehigh
Nhost homepage
nhost.io
Direct backend-as-a-service competitor positioning.
Retrieved 2026-06-25competitor pagehigh
Appwrite homepage
appwrite.io
Open-source backend-as-a-service and agent/MCP positioning.
Retrieved 2026-06-25competitor pagehigh
Convex homepage
convex.dev
Direct backend competitor positioning.
Retrieved 2026-06-25competitor pagehigh
Neon homepage
neon.tech
Serverless Postgres and agent-adjacent positioning.
Retrieved 2026-06-25competitor pagehigh
Neon pricing
neon.tech
Pricing benchmark.
Retrieved 2026-06-25company pricing pagehigh
PlanetScale homepage
planetscale.com
Database platform competitor positioning.
Retrieved 2026-06-25competitor pagehigh
Clerk homepage
clerk.com
Auth substitute in multi-vendor stack.
Retrieved 2026-06-25competitor pagehigh
Vercel homepage
vercel.com
Deployment and agent-infrastructure incumbent positioning.
Retrieved 2026-06-25competitor pagehigh
AWS Amplify homepage
aws.amazon.com
AWS full-stack incumbent positioning.
Retrieved 2026-06-25competitor pagehigh
Crunchbase: Supabase
crunchbase.com
Competitor scale and funding context.
Retrieved 2026-06-25structured startup registryhigh
Crunchbase: Vercel
crunchbase.com
Competitor scale and funding context.
Retrieved 2026-06-25structured startup registryhigh
Crunchbase: Nhost
crunchbase.com
Competitor scale and funding context.
Retrieved 2026-06-25structured startup registryhigh
Crunchbase: PlanetScale
crunchbase.com
Competitor scale and funding context.
Retrieved 2026-06-25structured startup registryhigh
Crunchbase: Appwrite
crunchbase.com
Competitor funding context.
Retrieved 2026-06-25structured startup registryhigh
Crunchbase: Neon
crunchbase.com
Competitor scale and acquisition-status context.
Retrieved 2026-06-25structured startup registrymedium
Crunchbase: Clerk
crunchbase.com
Substitute scale and funding context.
Retrieved 2026-06-25structured startup registryhigh
Crunchbase: Convex
crunchbase.com
Competitor funding context.
Retrieved 2026-06-25structured startup registryhigh

Access limitations: Public research could not retrieve private billing exports, product analytics, cap table, securities filings under alternate names, customer contracts, partner agreements, audit reports, pen-test summaries, or a trust-center package. LinkedIn Jobs access was unavailable for a structured jobs check, and search-result-only material was used only for absence and visibility footnotes rather than as cited proof. These limits mean usage, revenue, retention, funding reconciliation, compliance attainment, team coverage, and durable integration status remain unresolved until the company supplies documents.

Public disclaimer: This is an independent public teardown based solely on publicly available information retrieved on 2026-06-25. It is research support, not investment, legal, tax, or financial advice. Company-provided statements are labeled as claims, not verified facts. InsForge 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#