Answer 1
In terms of when to stop and pivot: Framework-as-Diagnostic Overlay
Greg's 7-tactic framework overlaid on Yuri's current infra: | Tactic | Greg's framework says | We have | Verdict | |---|---|---|---| | 1. MCP server as sales team | Publish to Smithery/MCPT/OpenTools for $0 CAC discovery | Zero — neither BizBuilder/KPDD nor Solacian have an MCP server | **REAL GAP** — but applicability depends on whether the product answers a queryable question; Solacian (Maze-dissolving AI) plausibly does, KPDD (PMF discovery) plausibly does | | 2. Programmatic SEO at 10K-page scale | Next.js + Firecrawl + AI content for "best X for Y" patterns | Zero pro…
Source: src/lib/bricks/sources/greg-isenberg-bootstrap-distribution.md
Answer 2
In terms of when to stop and pivot: Quick-reference — named tactics and when they work/fail
| Tactic | Works when | Fails when | |---|---|---| | **Atomic network** | Pick the tiniest specific group at a specific time; build density | "Peanut-buttering" across a whole geography/industry | | **Solve a Hard Problem** | Product nails the hard side's unaddressed need (Tinder for women) | Hard side churns → degrades for everyone | | **Come for the Tool, Stay for the Network** | Tool + network tightly integrated (Dropbox folders) | Tool/network divergent → low conversion | | **Invite-Only** | Curated connected users invite connected users | Used purely for hype; or kills…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 3
In terms of when to stop and pivot: 4.0 Forecast-First Creative Loop (Karpathy autoresearch pattern transfer)
Source: [[creative-autoresearch-loop-pattern-transfer.md]] | Raw: [[../knowledge-env/raw/2026-04-09-karpathy-autoresearch-source.md]] **When to use**: you want to generate hundreds of creative variations and SCORE them before spending real budget. Pairs with §1.5 (Apify mining) — mining gives the input data, this gives the experimentation engine. **Core pattern (transferred from Karpathy/autoresearch GitHub repo, 69K stars)**: - 3 files: `forecast.py` (READ-ONLY invariants — the metric the agent can't touch), `creative.py` (the only file the agent edits — generator config…
Source: src/lib/bricks/sources/perf-marketing-playbook.md
Answer 4
In terms of when to stop and pivot: Cross-cutting takeaways most relevant to BizBuilder (solo founders / vibe-coders seeking first traction)
1. Start with one small, dense atomic network — **never a Big Bang launch**. The first network always looks like a tiny market. Embrace unscalable, hustle-driven early tactics. 2. **Density beats raw size** ("all supply isn't created equal"). Build the Expectations Gap → exceed low expectations → high NPS → organic word of mouth. 3. **Ignore vanity metrics** — top-line counts mean nothing if users churn; quality is visible only from inside the network. 4. Pick the right entry point — **high economic value per transaction** and/or **high frequency/stickiness**. 5. **Acquisit…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 5
In terms of when to stop and pivot: 1.2 Creative Kill Threshold (Kirill / Poisson method)
Source: [[knowledge/references/2026-03-20-data-driven-marketing-conf.md]] **When to use**: managing 10+ creatives with significant spend, need statistical kill rules **Process**: 1. Pull data: creative ID, date, spend, CPA (CSV) 2. Feed to Claude: "Apply Poisson distribution to identify optimal kill thresholds. Scatter plot spend vs CPA. Flag creatives I should kill." 3. Red line = 90% confidence kill threshold 4. View each creative as a trajectory (path over time), not a point **Benchmarks**: - Dataset reference: $2.5M spend, 1700 creatives, 8 months - Real waste exampl…
Source: src/lib/bricks/sources/perf-marketing-playbook.md
Answer 6
In terms of when to stop and pivot: Cross-cutting takeaways most relevant to social / community growth
1. **Engagement Effect levers** — layer on new use cases, reinforce the core loop, reactivate churned users. 2. **Master the hard side (creators) first** — give them tools, monetization (even small — Twitch tipping, "$50/month was a big deal"), distribution. Homegrown native creators beat imported ones. 3. **Context collapse** is the central late-stage social problem — manage with networks-of-networks (channels, groups, finstas, Stories). 4. **Moderation must be built as software** — upvote/downvote/flag/block; Reddit's governance-as-city-planning model. 5. **Overcrowding &…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 7
In terms of when to stop and pivot: PART II — THE COLD START PROBLEM
**Ch. 4 — Tiny Speck / Slack.** Tiny Speck spent 4 yrs 10 mo, raised $17M, hired 45 people on the multiplayer game **Glitch** — Butterfield: "97% who signed up would be out of there within five minutes" (leaky bucket). Relaunched as **Slack** → 20M DAU, ~1M businesses, exited to Salesforce for ~$26B, $800M+ revenue. Slack grew from an internal IRC-based "frankentool" (Slack = Searchable Log of All Conversation and Knowledge). **[BIZBUILDER]** Butterfield personally signed up 45 companies in private beta — "I just had friends at other companies" — and personally handled the …
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md