VC eval stacks
under-cover product-craft.
Zero published a16z artefacts cover product craft. The most thesis-public fund in venture has 16 startup metrics, 16 more, a Marketplace 100, growth-metric guides, GTM playbooks per vertical, and a 150+ person operating bench — and not one rubric, scorecard, or framework for design quality, brand integrity, build hygiene, or interaction craft. We read eleven more Tier-2 generalists end-to-end. The gap is industry-wide, not a16z-specific.
Every load-bearing claim is tagged.
Discovery work blurs three things together — researched fact, our analysis, and our invented strawmen — and packs go wrong when readers can't tell which is which. So every claim in this pack carries a small pill saying where it came from. Read the colour, then weigh the claim accordingly.
Where your push-back matters most. STRAWMAN is the layer to disagree with hardest — that's where this pack improves fastest.
Five truths the rest of this pack defends.
If you read only this section, you have the thesis.
The wedge is real and structural. a16z, the most thesis-public fund in the industry, publishes no rubric, scorecard, or framework for product-craft. The 16 metrics, 16 more, growth-metrics guide, and Marketplace 100 are explicitly quantitative. Where craft surfaces inside a16z it appears as marketing/creative (crypto design partner) or internal-tooling craft (ASG product designers building deal-flow tools) — never as investment diligence on portcos. RESEARCH
The pattern holds across eleven more Tier-2 generalists. Sequoia (Arc, pitch template), Benchmark (Tavel's Hierarchy of Engagement), USV (network-effects thesis), Bessemer (Cloud Index), Founders Fund, Initialized, Accel, Greylock, Kleiner, NEA, GV, First Round (Levels of PMF) — none publishes a structured product-craft eval. The closest neighbours (BVP Cloud Index, First Round Levels of PMF) are the right shape on the wrong axis. The hypothesis is confirmed. RESEARCH
Segment B is roughly 3x more buyable than Segment A. Operating-partner / portfolio-services funds (Insight Onsite, GA Business Builders, Bessemer Growth, Battery, Stripes, Summit) have a budget owner, a procurement event, and a precedent. Sourcing/competitive funds (First Round, Sequoia, Greylock, Initialized) have partnership-vote or per-deal expense — high-friction, low-stickiness. Lead Segment B; produce Segment-A artefacts as the door-opener. SYNTHESIS
The sweet-spot AUM band is $1B–$10B. Sub-$1B funds lack a procurement line. >$10B funds (a16z, Insight, GA, Sequoia) can buy or build and have long sales cycles. The middle band has an operating function that is resource-constrained — they buy methodology rather than build it. Phase-1 design-partner shortlist filters through this lens. SYNTHESIS
One product, two render modes. The Scan engine should ship a founder-facing collaborative report (Segment A use) and an operating-partner-facing internal scorecard (Segment B use) from a single methodology. The artefact required for Segment A doubles as Segment B sales collateral — write a worked-example Scan on a public portco and run it as the wedge opener for both motions. STRAWMAN
Four exceptions to the pre-Discovery doc.
-
01Was Two segments — sourcing/competitive (A) vs operating-partner (B).Now The sharper cut is buyer-archetype: founder-facing-marketer GP vs internal-tooling Operating Partner. Usually but not always co-varies with A/B.Why it matters: A/B describes use; archetype describes purchasing motion. Sales operating model needs the archetype cut; positioning copy can keep A/B.
-
02Was A and B treated as roughly equal-priority beachheads.Now Lead with B for design-partner conversion. B has real procurement, a precedent (Insight Onsite, a16z operating model, GA Business Builders), and ~3x conversion likelihood. A is the door-opener that builds the references B will demand.Why it matters: collapses GTM ambiguity and re-orders the next-six-weeks sequence.
-
03Was AUM unconstrained, all Tier-1 Seed→Series-B funds in scope.Now Phase-1 sweet spot is $1B–$10B AUM. Sub-$1B is too price-sensitive; >$10B (a16z, Insight, GA, Sequoia) can buy-or-build and burns warm-network credibility too early.Why it matters: Anton-network warm intros land closer to B-archetype procurement than to A-archetype GP-pitching, but only after the credibility-gate passes.
-
04Was Two-product instinct — separate founder-facing and operator-facing SKUs.Now One engine, two render modes. A Scan rendered in operating-partner mode also works as Segment-A sales collateral; the artefact required to land B opens A doors.Why it matters: halves the methodology surface area and unifies the sales artefact. Strawman until tested with a real fund.
What blocks the next phase.
3 hard gates. 2 parallel. Pass 02 (worked-example Scan + first design-partner pilot) does not begin until the hard gates close.
uber-discovery-template-vc-pe-wedge and sku-product-diagnostic. Both exist in the BlackOps and Alpha Suite drawers respectively; both could plausibly be the same engine as Alpha VC Scan. Confirm — fund-side rollup of one engine, or two distinct SKUs — before Pass 02 commits architectural choices.Discovery rigour, quantified.
Listen to the pack. Or ask the research yourself.
The pack rests on a NotebookLM corpus of ~198 sources covering published VC evaluation methodologies, partner essays, podcasts, framework artefacts, and operating-model artefacts across Tier-1 (a16z) and Tier-2 generalists (Sequoia, Benchmark, USV, Bessemer, Founders Fund, Initialized, Accel, Greylock, Kleiner, NEA, GV, First Round). Two surfaces below — a podcast-format walkthrough of the pack, and the live corpus you can query.
Audio overview — the pack as a conversation.
Common questions, already asked.
Run against the corpus, with cited answers. Click to expand.
VC eval is industrialised on the dimensions that don't decide consumer outcomes.
"You quantify everything that's quantifiable. Here is the dimension that decides consumer outcomes and you currently evaluate by vibes. We make it auditable." — the persuasion architecture for the wedge, drawn from the a16z recon. SYNTHESIS
The published surface of Tier-1 and Tier-2 venture evaluation has matured along three vectors over the last decade. Quantitative business metrics — a16z's 16 + 16 More + Growth Metrics Guide, Bessemer's Cloud Index, First Round's Levels of PMF — give partners and LPs a shared scoring language for ARR, LTV/CAC, retention, gross margin, payback. Qualitative founder/market judgement — Andreessen on IQ + courage + drive, Thiel's contrarian conviction, Accel's Prepared Mind — gives partners the cultural permission to back ambition over data. Post-investment platform support — a16z's 150+ operating bench, Insight Onsite, GV's Design Sprint, Greylock Edge + Specialists, Sequoia Arc, First Round's PMF Method — turns the investment into a service relationship. RESEARCH
Product-craft sits inside none of these vectors. It is evaluated qualitatively, by partner taste, with no published rubric, scorecard, or framework anywhere in the eleven Tier-2 funds we read end-to-end. Where craft is named, it is named as talent (KP Fellows, Greylock Specialists) or workshop (GV Design Sprint) — never as measurement. RESEARCH
The closest neighbours, and why they don't close the gap
a16z's design surface is misread on first glance. a16z has a "design partner" role, an ASG product designer team, and a Design Engineer fellowship — but the design partner is a brand/marketing creative role for the firm and its crypto portcos, the ASG designers build a16z's own deal-flow tools, and the fellowship is a talent pipeline. None is investment diligence on portco craft. That defuses the surface signal that a16z already does this. RESEARCH
GV's Design Sprint is the closest neighbour, but it's generative not diagnostic. The Sprint produces a tested prototype; the Scan produces a comparable score. The original Sprint authors — Knapp, Zeratsky, Kowitz, Burka — have all left GV, so the methodology has institutional residue but no living owner inside the firm. The Scan can revive the GV-flavoured craft posture for the broader Tier-2 cohort. RESEARCH
Bessemer's Cloud Index and First Round's Levels of PMF are structurally the right shape on the wrong axis. Both are graded, comparable, portfolio-wide rubrics with public benchmarks. Both score commercial dimensions. The Scan can borrow the shape (level-based, benchmarked, public) and apply it to product/brand/build — and explicitly position itself in the lineage of these two. SYNTHESIS
Tavel's Hierarchy of Engagement is the only published "product as artefact" framework in venture, and it operates one layer above craft. Tavel evaluates whether users perform a Core Action and accrue benefits. The Scan operates one layer below: whether the artefact is well-enough crafted to enable that Core Action in the first place. The two are complementary; the pitch should namecheck Tavel rather than compete with her. SYNTHESIS
The negative space — what no Tier-2 generalist measures
The named gaps that come out of reading the eleven funds end-to-end: design system maturity (token coverage, component reuse, consistency rate); brand integrity across surfaces (visual / voice / positioning consistency); accessibility floor (WCAG conformance); build hygiene (TypeScript strictness, test coverage, deploy cadence, error-rate trend); ship cadence + product velocity; onboarding craft (TTV, activation conversion, drop-off topology); product-craft as moat (Thiel-flavoured argument); ship-quality drift over time (regressions in design system as teams scale). SYNTHESIS
AI is reshaping every Tier-2 fund's positioning in 2024–26. None has updated their craft / product / design content meaningfully. There is an attention vacuum, and the wedge sits inside it. RESEARCH
The auditable craft layer that sits next to the 16 metrics, not inside them.
A structured, comparable, portfolio-wide rubric that reads a startup's product the way a senior product/design lead does — brand coherence, IA logic, interaction quality, accessibility hygiene, motion/perf, write-craft, and the gap between brand promise and product reality. Sold as additive to existing eval stacks, not as replacement. STRAWMAN
The three slots — ranked by conversion likelihood
Slot A — Portfolio-monitoring craft index. A recurring score on portfolio cos that picks up qualitative drift (brand decay, design-system rot, surface fragmentation) before it shows up in churn or NPS. Sells to operating-partner / platform leads. Highest conversion likelihood: budget owner exists, precedent exists (Onsite-style ops functions), the methodology shape (level-based + benchmarked) has working analogues in Cloud Index and Levels of PMF. SYNTHESIS
Slot B — Pre-investment product-craft diligence overlay. A craft read on a Series A/B target, run by the fund as part of a competitive deal — co-branded with the founder, returned as a collaborative diagnostic not a graded report card. Lower conversion: the procurement is per-deal, not per-fund, and the GP's attention is the scarcest resource. But it doubles as Segment-A sales collateral and is the door-opener for Slot A engagements. SYNTHESIS
Slot C — Founder workshop / craft tune-up. Plug-in offer to the platform team, analogous to a16z's recruiting or comms support, but for product-craft. Lower-priority Phase-1 because it competes head-on with internal partners' time. Likely Phase-2 once a portfolio-monitoring footprint is established. SYNTHESIS
Slot D — Sourcing signal. Craft-quality detection as a leading indicator of breakout-consumer potential before quantitative traction shows up. Speedrun explicitly back founder-market-fit pre-traction; craft-signal is a parallel pre-traction lens. Speculative — testable but not yet evidence-backed. SPECULATION
How the wedge plays out — the next-six-weeks sequence
Week 1–2. Pick a single $1B–$10B AUM Segment-B target (Battery, Stripes, Bessemer Growth, Greylock, or a Summit/Insight-tier growth peer). Pick one of their public portcos. Run the worked-example Scan end-to-end in operating-partner render mode. Carry the citations from this pack into the Scan body so the artefact reads as the lineage of Cloud Index / Levels of PMF / Tavel — not a competitor. STRAWMAN
Week 3–4. Send the Scan unsolicited to the named operating-partner / platform-head with the opener: "We ran a Scan on [portco]. Here's what an external craft read picks up that doesn't show in your QBR. The methodology is what we'd embed if you wanted this across the portfolio." Track open / read / reply. STRAWMAN
Week 5–6. If reply, propose a paid pilot at $25k–$60k for a single portco scan with delivery in 10 working days. If no reply, repeat with a second target. The artefact compounds — every Scan run becomes recon for the next pitch. STRAWMAN
Why this works
The persuasion architecture writes itself. Every Tier-1 and Tier-2 generalist publishes quantitative scoring; none publishes qualitative scoring. The fund-internal answer to "how do we evaluate product?" is partner taste, which is unauditable to LPs and unrepeatable across a 40-portco book. The Scan supplies the audit-friendly layer that the existing metric stacks can't reach without changing what they measure. The shape (level-based, benchmarked, public lineage) is borrowed from frameworks the buyer already trusts. The pricing range ($150k–$400k/yr methodology license, or $25k–$60k per portco scan) sits inside the operating-partner advisory comparables band the kickoff brief anchored to. SYNTHESIS
"The Cloud Index for product-craft." A complement, not a competitor.
The cleanest one-line positioning we found in a week of desk recon is the lineage line: the same shape as Bessemer's Cloud Index (publicly benchmarked, comparable across cos), the same axis as the gaps every Tier-2 fund leaves untouched (design quality, brand integrity, build hygiene, ship cadence). Naming neighbours is a credibility move — it lets the buyer place the Scan in a category they already trust rather than evaluate a new one. SYNTHESIS
The narrative architecture — three statements in sequence
1. Acknowledge what works. "You've industrialised financial scoring. The 16 metrics, Cloud Index, Levels of PMF, Hierarchy of Engagement — these are real and we're not replacing them." Disarms the "we already do this" reflex; positions the Scan as additive. STRAWMAN
2. Name the gap precisely. "There is no published rubric anywhere in venture for design quality, brand integrity, build hygiene, or interaction craft. We've read every Tier-1 and Tier-2 generalist artefact. The closest neighbours are GV's Design Sprint (generative, not diagnostic) and Sarah Tavel's Hierarchy (one layer above craft). The space below is empty." Specificity earns the ground. SYNTHESIS
3. Land the offer. "We score the dimension that decides consumer outcomes and is currently evaluated by partner taste. We make it auditable." This is the line that does the heavy lifting in a procurement conversation. STRAWMAN
What we are not
Not a substitute for partner judgement. Not a marketing/brand creative agency. Not a Design Sprint replacement (the Sprint is generative, the Scan is diagnostic — different jobs). Not a financial scorecard. Not a one-trick consultancy on accessibility / performance / brand alone — the Scan reads the artefact whole and surfaces the cross-axis pattern. STRAWMAN
Naming, surface, and brand
The current working name "Alpha VC Scan" reads correctly to the buyer archetype: "Alpha" → Alpha Suite institutional methodology lineage; "VC" → buyer; "Scan" → the diagnostic-not-workshop verb that distinguishes from GV's Sprint. Worth pressure-testing once a real fund is in the conversation. The deployed surface lives at scan.thealphasuite.com, gated by password, indexed-noindex, with a brand-loader that refuses to render without verified tokens. The pack you are reading is the canonical render. RESEARCH
Two archetypes. One is roughly 3x more buyable.
No personas — this is a desk-recon Discovery, not a primary-research one. The two archetypes below are buyer-archetype synthesis from public artefacts and named-fund evidence. Pass-02 will carry these into structured interviews and add PRIMARY data. SYNTHESIS
Archetype A — The Founder-Facing-Marketer GP
Where they live. First Round, Sequoia, Greylock, Initialized, Bessemer (founder-facing surface), Index, Founders Fund (lighter), and the founder-facing surface of a16z. Investment partners or general partners who personally front competitive deals. RESEARCH
What they think about. Win rate on hot rounds. Brand differentiation against the other Tier-1 they're competing with this week. Founder-side value. The artefact they hand a founder in second meeting. SYNTHESIS
How they'd buy. Per-deal expense ($5k–$25k per scan, GP-expensed) or annual platform license ($75k–$150k/yr fund-paid). Procurement is partnership-vote-shaped — high-friction, not centralised, sale lands repeatedly across partners. SYNTHESIS
The phrasing they use. "We need something we can show in second meeting that the founder will actually take with them and use." "Every fund says they have a platform — what's the artefact?" "When we lose a deal it's never on price, it's on the founder feeling someone else gets their product." "Our brand is product-thoughtful but we don't have a thing we hand to founders that proves it." SYNTHESIS
Job-to-be-done. Win the deal by demonstrating product-craft taste before term-sheet, in a way the founder values and remembers. It's a sales-collateral job dressed as a diligence job. SYNTHESIS
What gets in the way. "We already do this ourselves with our partners" — real at funds with strong product-partner DNA. "Founders won't value an external scan from us — it'll feel like homework, not a gift." "This will leak — competing funds will get hold of it and use it against us." Buying is per-deal, not per-fund — no single procurement event. SYNTHESIS
Archetype B — The Internal-Tooling Operating Partner
Where they live. Insight Partners (Onsite), General Atlantic (Business Builders), Bessemer Growth, Battery, Stripes, Summit, Greylock (boundary). The named precedent layer is Insight Onsite specifically — ~100+ operators across functions, publicly described as a structured portfolio-services function. RESEARCH
What they think about. Portfolio coverage at scale. Methodology consistency across portcos. Risk-flagging on portcos before financial signal moves. Defensible, repeatable processes that survive partner turnover. Auditability to LPs. SYNTHESIS
How they'd buy. Methodology license $150k–$400k/yr, or per-portco scan $25k–$60k each scaled across 20 portcos/yr ($500k–$1.2M ARR per fund), or quarterly retainer $50k–$100k/qtr embedded. Procurement is a real procurement event — evaluator/decider/sponsor roles, multi-month cycle. SYNTHESIS
The phrasing they use. "We have 40 portcos. I cannot personally review each one's product quarterly, but I need to know which ones are at risk." "Our QBRs are too financial. The product story drops out, and that's where the surprises come from." "When we add a new operator-partner I want them reading every portco the same way. We don't have that today for product." "I can score GTM, I can score talent. I can't score product without a ten-hour partner deep-dive." SYNTHESIS
Job-to-be-done. Scale a consistent, defensible product-craft read across N portcos with a small team and limited partner attention. Portfolio-monitoring + risk-flagging job, dressed as a methodology adoption. SYNTHESIS
What gets in the way. "We already do this internally" — strong at a16z and Insight specifically (they have the headcount). "Methodology adoption is slow — won't survive a partner change." "You're three people. We work with McKinsey / Bain / Insight Onsite-internal." Long sales cycle (6–12 months realistic at Insight/a16z scale). SYNTHESIS
Why B is ~3x more buyable
1. Procurement is real. B has a budget owner; A has partnership-vote or per-deal expense. 2. Precedent exists. Insight Onsite, a16z's operating model, GA Business Builders all prove the buy is institutional; A's analogous precedent is patchy ("founder platforms" are mostly built in-house, not procured). 3. Methodology fit. Alpha VC Scan's institutional-rigor posture (Triangulation harness, atom IDs, provenance pills — inherited from Alpha Suite) maps to a Head of Portfolio Operations' procurement criteria; overkill for a GP picking a deal-time differentiator. 4. Sales cycle longer but lands harder. A landed B engagement is a 6-figure annual contract with renewal mechanics; A is per-deal and may not recur. 5. Anton-network alignment. Anton's cap table includes a16z and growth-tier funds with named operating partners — warm-intro path lands closer to B procurement than to A GP-pitching, contingent on credibility-gate. SYNTHESIS
Funds that span both
The largest, longest-tenured Tier-1s span both archetypes — a16z (operating-partner machine + deal-winning founder-platform), Sequoia (Foundry/Arc + structured portfolio-engagement), Bessemer (Cloud Index thought-leadership + growth-stage operating playbooks), Greylock (Edge programme + network-as-platform). For dual-segment funds the sale is multi-stakeholder — the opener still goes through B (where procurement lives) but the case for adoption mentions both surfaces. For single-segment funds, pitch the corresponding archetype only — adding the other surface dilutes. RESEARCH
8 measurement gaps. Industry-wide.
No Tier-1 or Tier-2 generalist VC publishes a rubric for any of the dimensions below. The "pain" lands on the operating partner who needs the read across 40 portcos, and (less acutely) on the GP losing a competitive deal because the founder felt the other side "got" their product.
| ID | Gap | Cost / consequence | Type | Atom |
|---|---|---|---|---|
| GAP-01 | Design system maturity. No published rubric for token coverage, component reuse, consistency rate. Funds can't tell which portcos have a system and which have a fragment graveyard. | Drift compounds invisibly until a redesign cycle. Surfaces in churn / NPS only after 6–18 months. | Structural | RESEARCH RECON-A16Z-§7 RECON-T2-SYNTH |
| GAP-02 | Brand integrity across surfaces. No measurement of visual / voice / positioning consistency across product, marketing, docs, sales surfaces. Brand decay is felt, not graded. | Founder + fund both lose deal heat when brand fragments. Hardest to recover post-Series-B. | Sharp | RESEARCH RECON-A16Z-§8 |
| GAP-03 | Accessibility floor. WCAG conformance not evaluated by any fund as portfolio-wide metric, despite consumer + enterprise procurement implications. | Enterprise procurement blockers; legal exposure in regulated markets. | Latent | SYNTHESIS RECON-T2-SYNTH |
| GAP-04 | Build hygiene. TypeScript strictness, test coverage, deploy cadence, error-rate trend — none scored as a comparable craft index across portcos. | Velocity decay; on-call burden; incident rate moves before churn does. | Chronic | SYNTHESIS RECON-T2-§GAPS |
| GAP-05 | Ship cadence + product velocity. Anecdotal in partner essays (Andrew Chen, Linear-style culture) but never operationalised into a portfolio-comparable signal. | Late-stage drag; the "they used to ship fast" pattern. | Slow-drip | SYNTHESIS RECON-T2-§GAPS |
| GAP-06 | Onboarding craft. TTV, activation conversion, drop-off topology — owned by Tavel's Hierarchy at the behavioural level, but not at the craft level (microcopy, interaction quality, friction sourcing). | Direct conversion lift left on the table; founder can't tell craft cost from positioning cost. | Sharp | SYNTHESIS RECON-T2-§BENCHMARK |
| GAP-07 | Product-craft as moat. Thiel names "brand" as one of four monopoly dimensions in Zero to One. No fund evaluates brand-as-moat as a measurable input. | Defensibility argument made on vibes; LP narrative weaker than it could be. | Structural | RESEARCH RECON-T2-§FOUNDERS-FUND |
| GAP-08 | Ship-quality drift over time. Design-system rot as teams scale (Series B → C transition). No fund picks this up before churn or NPS does. | Late detection — by the time the financial signal moves, the craft debt is 12–24 months deep. | Emerging | SYNTHESIS RECON-T2-§GAPS |
How the wedge plays out — channel by channel.
A position is only as strong as the channels carrying it. Channels here are sequenced by how they compound the artefact (worked-example Scan) into compounding inbound.
1. The worked-example Scan as direct outbound
The strongest opener in both archetypes is "Show me a Scan you'd run on [their portco]" — except we don't ask, we just run it and send it. The artefact is the entire pitch. No deck, no intro call, no "happy to chat". Twenty days of recon, ten days of writing, one PDF + a pack URL. Sent unsolicited to the named operating partner with a six-line cover note. Conversion target: 1 in 10 lands a paid pilot. STRAWMAN
2. Thought-leadership artefact — "How VCs evaluate product (and where they don't)"
The audio overview generated against this Discovery corpus is the long-form public artefact. It namechecks Tavel, Burka, Knapp, Cloud Index, Levels of PMF — the closest neighbours — and locates the gap. Doubles as inbound seed for B-archetype operators reading partner essays + listening to podcasts. Lower immediate conversion, compounds. Distribution: drop into Lenny's Newsletter, Acquired LP show, BlackOps-channel posts on LinkedIn from named operators in our network. SYNTHESIS
3. LinkedIn — operator-network adjacent
BlackOps's network of ex-product/design leads at the 8 fintech/institutional brand-led companies (Talos, Fathom, Appital, Access Fintech, Algomi, Pirum, Open Gamma, BlockEx) sits closer to operating-partner conversations than to GP-pitching. Use the Scan + audio overview as content-fuel for those operators' own posts; they tag funds they've worked with. Asymmetric reach. SYNTHESIS
4. Anton-network warm intros — gated
Anton's 28 institutional warm-intro paths include funds across both archetypes. Hard rule: no Anton-network pitching to a16z, Sequoia, or any >$10B fund until the credibility-gate passes (per dont-spend-credibility.md). Phase-1 outreach uses Anton's network only for $1B–$10B B-archetype targets, and only after a worked-example Scan has been published. RESEARCH
5. Inbound surface — scan.thealphasuite.com
This pack, password-gated, brand-loaded against extracted BlackOps tokens, indexed-noindex. The deployed surface is itself part of the wedge architecture: an institutional-aesthetic, evidence-spine-tagged Discovery pack that signals Alpha Suite methodology lineage in seven seconds. The password gate keeps it inside the warm-network corridor while the artefact compounds. RESEARCH
6. Channels deliberately not used in Phase-1
No paid acquisition. No conference circuit (yet). No GP cold outbound — the GP is too senior to receive cold and the artefact is wasted on partnership-vote procurement. No SEO play — the buyer count is small enough that named-account outbound dominates. SYNTHESIS
Twelve questions before the methodology commits.
Answer any of the questions below or drop in any file that helps. Submissions feed directly into Pass-02 and become PRIMARY sources alongside the desk recon.
Ask the notebook.
Type any question about how Tier-1 and Tier-2 VCs do (and don't) evaluate product, and get an answer grounded in the ~198-source corpus. Most answers arrive in 20–40 seconds.