Linh Truong · Solo AI company · 2026

The One‑Person AI Company Architect

How I structure a one-person company when agents do most of the recurring work—model, stack, workflows, money, legal, risk, and a 180-day path from idea to revenue.

Author: Linh Truong, MA (Harvard), MBA Source: LinhTruong.com Email: Linh@Alumni.Harvard.edu For: Founders · indie hackers · operators · engineers Version: 2026.05

1 · Executive Summary

A One‑Person AI Company (1P‑AIC) is a legal entity owned and operated by a single human in which the majority of recurring work — engineering, marketing, sales, support, operations, finance, and research — is performed by orchestrated AI agents and software automations. The human's role shifts from doer to architect, editor, and capital allocator.

1Headcount
$1–10MRealistic ARR ceiling
80–95%Gross margin
180dLaunch‑to‑revenue target

KPI ranges are directional—verify against §18 before citing.

My take. In 2026, capable agents plus cheap APIs and infra make it realistic for one operator to run something that behaves like a small SaaS or services team—if you design orchestration, evals, and governance on purpose.

What's in here

2 · Why Now — The 2026 Inflection

Cost curve collapse

Inference cost for top‑tier reasoning dropped ~95% in 24 months. Background tasks (drafting, summarising, coding) now cost cents per workflow rather than dollars.

Agentic tool use is reliable

Tool‑use loops, long‑horizon planning, and computer‑use agents cleared the 80%+ success threshold for repeatable office work. That's the line at which automation economics flip positive.

API‑first SaaS & usage pricing

Stripe, Vercel, Supabase, Cloudflare, HubSpot, Notion, Linear, and every major channel expose programmable surfaces. Per‑seat pricing is being out‑competed by per‑task pricing.

Distribution is unbundled

YouTube, TikTok, X, LinkedIn, Reddit, and Substack each can produce $100K+ in inbound revenue solo. AI generation lowers the cost of quality not just quantity.

Implication. The constraint on solo entrepreneurship is no longer skills or capital — it is taste, focus, and orchestration ability. This paper is structured around those constraints.

3 · Operating Principles

  1. Founder as Architect. Your job is to design systems, not to do the work the systems are designed to do. If you spend >20% of your week inside a single function (e.g. writing copy), you have failed to architect that function.
  2. Default to automate; fallback to delegate; last resort, do it yourself. A task should pass three gates: (1) Can an agent do it? (2) Can a contractor do it for <$X? (3) Only if both fail, do it.
  3. Compounding loops over linear effort. Every workflow must produce a reusable asset: content, a customer record, a code module, or a learned prompt.
  4. One product, one ICP, one channel (at a time). The solo founder's #1 enemy is surface area. Concentrate until the first $30K MRR; then expand.
  5. Pay for leverage, not for status. $500/mo of API credits is cheaper than a $5K/mo VA. Cancel any tool not used weekly.
  6. Document everything to disk. Memory, prompts, SOPs, customer notes — all in plain text, versioned, agent‑readable. The company is its repo.
  7. Resilience > optimization. Two agents from two providers beat one perfectly tuned pipeline. Single points of failure are forbidden.

4 · The Architecture Overview

The reference architecture has six layers. Each layer can be swapped without rewriting the others. This is the diagram you should pin above your desk.

Figure 1 · One‑Person AI Company — Reference Architecture
L1 · FOUNDER (Architect / Editor / Capital Allocator) Strategy · Taste · Brand · Final approvals · Customer relationships · Capital deployment L2 · ORCHESTRATION (the company's "operating system") Claude Code / SDK n8n / Temporal Cron / Schedulers MCP Tool Registry Eval & Observability L3 · AGENT WORKFORCE (specialised AI roles, each a prompt + tools + memory) Engineering code · review · deploy Marketing content · SEO · social Sales outbound · CRM · proposals Support tickets · docs · onboarding Ops & Finance billing · books · reports Research market · product · brief L4 · MODEL LAYER (frontier + small + specialist; multi‑vendor by default) Claude 4.7 Opus/Sonnet GPT‑5 / o‑series Gemini 3 Pro / Flash Llama / open weights Voice · Image · Video Embeddings L5 · DATA & MEMORY Vector DB (pgvector / Pinecone) Postgres / Supabase Object storage (S3 / R2) Knowledge repo (Git / Notion) Event log / analytics (PostHog · BigQuery) L6 · CHANNELS (the company's interface with the world) Product / SaaS Stripe · Vercel Website / SEO Next · Astro · MDX Email / Newsletter Beehiiv · Resend Social X · LinkedIn · YT Community Discord · Circle CRM / Sales surface HubSpot · Attio · Cal.com

Reading the architecture

5 · Business Function Map

A 20‑person SaaS company has roughly the following functions. The right‑hand columns show how a 1P‑AIC handles each.

FunctionTraditional FTEsAutomation ApproachFounder Time
Product strategy1 PMFounder + research agent + customer interview synthHigh
Engineering3–5 engineersClaude Code as primary IC; founder reviews PRsMedium
Design1 designerComponent library + Figma‑to‑code agent + V0/LovableLow
DevOps / SRE1Vercel/Fly + Cloudflare + agent on‑call for incidentsLow
Marketing / content2Content pipeline agent + research + scheduler; founder gives anglesMedium
SEO0.5Programmatic SEO + topical map agentLow
Sales / BD2 AEsOutbound agent + Cal.com; founder runs demosHigh
Customer support2Tier‑1 AI agent on docs + Slack/Discord; founder for escalationsLow
Customer success1Health‑score agent + drip emails + quarterly callsMedium
Finance / accounting1Stripe + Mercury + bookkeeping agent + fractional CPALow
Legal0.25 fractionalTemplate library + agent redlines + lawyer on retainerLow
HR / recruiting1N/A — contractors only, via agent‑drafted briefsNone
Data / analytics1PostHog + SQL agent + weekly digestLow
Founder time discipline. Sum the "High" rows. If they exceed ~25 hours/week, you have designed a job, not a company. Re‑architect.

6 · The AI Agent Stack

Each agent is defined by a one‑page charter: Mission InputsToolsOutputs KPIsEscalation rule. Below is the canonical roster for a 1P‑AIC.

Figure 2 · Agent workforce — roles, models, & primary tools
Orchestrator Claude Code / SDK + MCP registry Strategy & Research Opus · web · readwise → weekly briefs Marketing & Content Sonnet · CMS · social API → posts, SEO pages Sales / Outbound Sonnet · Apollo · HubSpot → qualified meetings Customer Support Sonnet · docs · Intercom → ticket resolution Engineering Opus · GitHub · Vercel → PRs, deploys Finance & Ops Sonnet · Stripe · Mercury → books, alerts, reports Knowledge / Memory Embeddings · pgvector → context for all agents Customer Success Sonnet · PostHog · email → retention, NPS, QBR

Agent charter (template)

# Agent: Marketing & Content Producer
mission:  Ship 5 high‑quality posts/week that rank or convert.
inputs:   - product changelog
          - keyword map (Ahrefs export)
          - customer call transcripts
tools:    web.search, cms.publish, image.generate, social.post, analytics.read
outputs:  - draft post (Markdown) for human edit
          - SEO meta, OG image, social variants
KPIs:     organic traffic, top‑10 keywords, email signups attributed
escalate_if: claim is unverifiable | tone deviates from brand | legal risk

Roster & recommended models

AgentRecommended modelWhyBackup
Strategy / ResearchClaude Opus 4.7Best long‑context reasoningGPT‑5
EngineeringClaude Opus 4.7 (Code)SOTA on real‑world code tasksGPT‑5 / Gemini 3 Pro
Marketing / ContentClaude Sonnet 4.6Voice quality, costGemini 3 Pro
Sales / OutboundSonnet 4.6 + HaikuPersonalization at low costGPT‑5 mini
Support tier‑1Haiku 4.5 + RAGLatency & costSonnet for hard tickets
Finance / OpsSonnet 4.6Numeric reliabilityGPT‑5
Knowledge / MemoryVoyage / OpenAI embedBest retrieval qualityCohere

7 · Tool & Platform Matrix

Pick exactly one tool per row. Resist accumulating. Total stack cost target: $300–800/mo at pre‑revenue, $1.5–4K/mo at $1M ARR.

LayerNeedDefault pickAltMonthly cost (est.)
ModelsFrontier reasoningAnthropic API (Claude 4.7)OpenAI · Google$50–1500 usage
Agent harnessBuild/run agentsClaude Agent SDK / Claude CodeLangGraph · Mastraincl.
WorkflowCross‑tool automationn8n (self‑host) / ZapierMake · Temporal$0–50
ComputeApp hostingVercel + Fly.ioCloudflare Workers$20–200
DBPrimary storeSupabase (Postgres + pgvector + auth)Neon + Clerk$25–100
StorageFiles/mediaCloudflare R2S3$5–30
PaymentsBillingStripeLemon Squeezy (MoR)2.9% + 30¢
Email infraTransactionalResendPostmark$20
NewsletterList + sendBeehiivGhost · Kit$0–100
CRMPipelineAttioHubSpot Starter$30–60
OutboundSequencingSmartlead / InstantlyApollo$60–150
SchedulingDemo bookingCal.comCalendly$15
SupportTickets/chatPlain · Intercom FinCrisp$30–100
DocsPublic docsMintlifyDocusaurus$0–80
AnalyticsProductPostHogPlausible · Mixpanel$0–100
ObservabilityAgent tracesLangfuseHelicone · Phoenix$0–50
KnowledgeSOPs / wikiGit repo (Markdown) + ObsidianNotion$0–20
BankingOperating acctMercuryBrex · Wise$0
AccountingBooksPilot / BenchQuickBooks + CPA$200–500
LegalTemplates + lawyerStripe Atlas / Clerky + fractional$50–300 avg
Identity / SSOAuthClerk / WorkOSSupabase Auth$25
VoiceCalls / agentsVapi · RetellElevenLabsusage
Selection rule. Prefer tools with (1) public API, (2) MCP server or SDK, (3) usage‑based or flat pricing — never per‑seat. Avoid any tool whose growth path requires hiring a human admin.

8 · Revenue Models & Unit Economics

The 1P‑AIC has five viable revenue archetypes. Pick one for years 1–2.

A · Vertical SaaS

Software for a narrow ICP (e.g. dental clinics, marine surveyors). $50–500 ACV × thousands. Best for compounding moat.

Margin 85–92%CAC payback 3–9 mo

B · Productised Service

Fixed‑scope deliverables (audits, migrations, designs) priced flat. AI does 80% of work; you sell taste.

Margin 70–85%Cash up front

C · AI Agent / Copilot

An autonomous worker rented per task or per month. Usage pricing aligns with value.

Margin 60–80%Variable COGS

D · Media / Audience

Newsletter, YouTube, podcast → sponsors, courses, community, affiliates. Long ramp, high LTV.

Margin 90%+12–24 mo ramp

E · Marketplace / Aggregator

Match supply (data, talent, listings) with demand. Hard to start; defensible if it works.

Margin 60–80%Chicken‑egg risk

F · Hybrid (Year 2+)

Combine: e.g. media → SaaS, or service → product. Don't attempt before $30K MRR from one model.

Only after PMF

Reference unit economics — Vertical SaaS path

MetricTarget Y1Target Y2Notes
ACV$1,200$2,000$100–170/mo plans
Gross margin85%88%Inference + infra
CAC$200$300Mostly content + outbound
CAC payback3 mo3 moCash‑in‑cash‑out
Logo churn4%/mo2%/moVertical helps lower
Net revenue retention95%110%Add usage tier
ARR target$120K$600K–$1.2MSolo realistic
Owner take‑home$60–90K$300–600KAfter tax & reinvest

Pricing principles

9 · Operational Workflows

Below are the eight workflows that run a 1P‑AIC. Each is implemented as an n8n or Claude‑SDK flow with explicit human checkpoints (★).

Figure 3 · Daily / weekly workflow loops
INBOUND ENGINE (daily) Keyword scrape Draft post Image + SEO meta ★ Founder edit Publish + cross‑post + index ping OUTBOUND ENGINE (daily) ICP list pull Personalize w/ web research Send sequence ★ Reply triage Book demo → CRM → calendar PRODUCT DEV (continuous) Issue intake Branch + Claude Code PR + tests + eval ★ Founder review Auto‑deploy + changelog post SUPPORT (24/7) Ticket in RAG over docs Draft reply ★ Auto‑send if confident Log + tag for docs / product gap FINANCE (weekly) Pull bank + Stripe Categorize Cash runway calc ★ Approve anomalies Send weekly dashboard email CUSTOMER SUCCESS (daily) Compute health Detect churn risk Trigger play ★ Founder call if >$5K ACV Update CRM · log outcome STRATEGY & RESEARCH (weekly) Scrape competitors Synth customer calls Rank ideas by ICE Weekly brief draft ★ Founder picks 1 bet Spawn product issue Update bets log → quarterly review BRAND & COMMUNITY (daily) Listen (mentions, RSS) Draft replies / threads ★ Founder voice pass Schedule posts Community DMs Weekly digest Pipe insights → research workflow
★ Human checkpoint (founder) Automated step

Workflow design rules

10 · Data, Memory & Knowledge

Your data layer is the only thing that compounds. Tools change, models change, channels change. The corpus of customer conversations, code, prompts, post‑mortems, and SOPs is the moat.

Working memory

Per‑agent scratch + summarised conversation history. Lives in Postgres jsonb.

Semantic memory

Embeddings of all docs, calls, tickets, posts. pgvector or Pinecone.

Episodic memory

Append‑only event log of every agent action with inputs, outputs, cost, latency.

Source‑of‑truth hierarchy

  1. Git repo — code, SOPs (Markdown), prompts, agent charters. The repo IS the company.
  2. Postgres — customers, orders, events, tickets, agent traces.
  3. Object storage — call recordings, exports, generated assets.
  4. External SaaS — treated as caches, not sources. You can rebuild them.
Backup discipline. Nightly DB snapshot to a second region + a third destination outside your primary cloud (e.g. Backblaze B2). Test restores quarterly. A 1P‑AIC has no IT department to catch ransomware or vendor outages.

Knowledge engineering for agents

11 · Finance, Tax & Legal

Entity structure (US‑centric defaults)

StageEntityWhy
Pre‑revenue / sideSole prop or single‑member LLCCheap, simple, pass‑through
$50K+ profitLLC taxed as S‑CorpSave self‑employment tax via salary + distributions
Raising / partners / exitDelaware C‑CorpStandard for investors and acquirers
Non‑US founderWY/DE LLC or UK LtdStripe Atlas / Clerky / Tide; consult tax advisor

Accounting set‑up (week 1)

  1. Separate operating bank (Mercury / Brex). Never commingle.
  2. Stripe (revenue) → bookkeeping tool → CPA monthly.
  3. Set up a tax savings sub‑account: auto‑move 25–35% of every Stripe payout.
  4. Track 3 numbers weekly: cash on hand, MRR, runway months. Nothing else matters.

Contracts you need on day 1

AI‑specific legal considerations (2026)

Solo‑founder failure mode: a single missed sales‑tax filing in a US state where you have nexus can wipe a year of profit. Use TaxJar/Anrok from day 1 if selling B2B SaaS in the US.

12 · Risk, Security & Compliance

Risk register

RiskLikelihoodImpactMitigation
Founder illness / burnoutHighExistentialRunbook + 2 trusted contractors on retainer + 3 mo cash buffer
Model provider outageMediumHighMulti‑provider routing; degrade gracefully
Account suspension (Stripe / Google / X)MediumHighBackup processor, own DNS, export everything weekly
Hallucinated output to customerMediumMediumRAG grounding + confidence threshold + human review queue
Prompt injection / data exfiltrationMediumHighSanitise tool inputs; restrict agent file/network scopes; secret scanner in CI
Data breachLowExistentialEncryption at rest, MFA everywhere, password manager, SOC2 readiness from day 1
Competitor builds same thingHighMediumSpeed + niche depth + brand; defensible data, not features
Cost runaway from agentsMediumHighPer‑user/per‑workflow spend caps; alerts at 80% of monthly budget
SEO/algorithm shockMediumMediumEmail list as owned channel; diversify 3+ acquisition channels by Y2
Regulatory change (AI Act, US state laws)MediumMediumQuarterly legal review; subscribe to Stanford HAI / IAPP updates

Security baseline (do this in week 1)

Agent safety constraints

13 · Scaling & Exit Strategy

The scaling staircase

StageARRYou add…What breaks first
0 → $10K MRR$0–120KNothing. Find PMF.Focus
$10K → $30K MRR$120–360KVA (10h/wk) for inbox + opsSupport volume
$30K → $80K MRR$360K–1MContract engineer (20h/wk) + fractional CPAProduct velocity
$80K → $200K MRR$1–2.5MDecide: stay solo or hire first FTFounder bandwidth
$200K+ MRR$2.5M+Either FT team OR a partner‑operatorStrategic ambition

Decision: do you ever hire?

Stay solo (Calm Company path)

$1–3M ARR, 80% margin, 25 hr/week, total optionality. The default recommendation for most readers.

Hire (Scale path)

First hire = senior generalist who can own one function end‑to‑end. Not a junior. Equity 1–5%. Only if you have a $10M+ outcome thesis.

Exit options

Prepare to exit even if you never sell. Clean books, documented SOPs, transferable contracts, no founder‑locked credentials — these increase optionality and reduce stress now.

14 · 180‑Day Implementation Roadmap

Figure 4 · 180‑day roadmap, phase by phase
Day 0 Day 30 Day 60 Day 120 Day 180 Phase 1 · FOUND Phase 2 · BUILD Phase 3 · LAUNCH Phase 4 · GROW & SYSTEMISE • Pick ICP & problem • Validate with 15 calls • Choose revenue archetype • Form entity + banking • Lock domain + brand • Stand up repo + DB • Pre‑sell 3 LOIs • Define agent roster • Founder learns Claude Code Exit gate: signed LOIs • Build MVP (4 wk sprint) • Wire orchestrator + 2 agents • Stripe + auth + billing • Docs + onboarding • Inbound + outbound workflow • Internal eval suite • Closed beta with LOIs • Iterate weekly • Set up risk register Exit gate: 5 paying customers • Public launch (PH / X) • Activate content pipeline • Outbound 100/day cadence • Add support agent • Pricing v2 • Customer success plays • Weekly metric review • 1st external integration • First retro / kill list Exit gate: $10K MRR • Programmatic SEO at scale • 2nd acquisition channel • Add finance + research agent • Quarterly bets board • First VA / contractor • Annual plan launch • Compliance baseline (SOC2 Lite) • Backup & DR drill • Personal sustainability audit Exit gate: $30K MRR, <30 hr/wk

Week‑by‑week (first 30 days)

WeekOutcomeConcrete deliverables
1Problem locked15 customer calls booked; ICP doc; 3 problem hypotheses
2Calls doneCall notes embedded; top‑2 problems chosen; 3 LOIs requested
3Entity + infraLLC formed; Mercury + Stripe live; repo + DB + Vercel skeleton
4Architecture setAgent roster signed off; tool matrix locked; first agent (research) shipped

15 · KPI Dashboard & Decision Frameworks

The 12 numbers a 1P‑AIC tracks

LayerMetricCadenceHealthy range
CashCash on handWeekly≥6 mo runway
CashBurn / monthWeekly< 30% of revenue
RevenueMRR + growth %Weekly>10%/mo early
RevenueNRRMonthly>100%
FunnelTrials → paid %Weekly>20% B2B SaaS
FunnelDemo show rateWeekly>70%
ProductWAU / MAUWeekly>0.5
ProductTime‑to‑value (TTV)Monthly<1 day
SupportAI deflection %Weekly>60%
Cost$ per active user (AI)Weekly<15% of ARPU
FounderHours workedWeekly<40, ideally <30
Founder% time on "Doing"Weekly<30%

The "Should I automate this?" decision tree

  1. Does the task repeat ≥10×/week or block growth? → continue. Else skip.
  2. Can it be defined in <1 page (inputs, outputs, success)? → continue. Else clarify first.
  3. Is failure recoverable in <1 hour? → automate. Else add human checkpoint.
  4. Will an agent be ≥80% as good as you? → automate. Else delegate to contractor.
  5. Build cost < 1 month of doing it manually? → build. Else buy.

The ICE prioritization (weekly bets)

Score every bet on Impact (1–10), Confidence (1–10), Ease (1–10). Multiply, sort, do the top 1–2. Anything below 200 is shelved.

16 · Common Failure Modes

1 · Tool collector syndrome

Subscribing to 30 SaaS, integrating none. Cure: monthly stack audit; cancel anything unused 14 days.

2 · Demo‑ware loop

Endless polishing of agents without ever asking for money. Cure: pre‑sell before you build.

3 · Founder‑as‑bottleneck

Every workflow ends "ping founder." Cure: raise the auto‑approval threshold every week.

4 · Hallucination tolerance

Letting an outbound or support agent ship low‑quality output to brand customers. Cure: eval suite + human review queue with SLA.

5 · Surface‑area explosion

Three products, five channels, two ICPs. Cure: one of each until $30K MRR.

6 · Burnout from "freedom"

No team = no forcing function. Cure: weekly review with a peer; calendar block your off‑days.

7 · Single point of failure (you)

Cure: write a 1‑page "if I'm hit by a bus" doc; share access with spouse/lawyer.

8 · Vendor lock‑in

Building everything around one platform's quirks. Cure: keep prompts & data portable; abstract LLM calls.

17 · Appendix — Templates & Checklists

A · One‑page agent charter

# Agent: <name>
mission:        <1 sentence>
owner:          founder
model:          primary | fallback
tools:          [tool1, tool2, ...]
inputs:         <what triggers it / what it reads>
outputs:        <artifacts produced + where stored>
KPIs:           [metric1 target, metric2 target]
spend_cap:      $/day, $/run
escalation:    <when to ping founder>
eval_set:      ./evals/<name>.jsonl
last_reviewed: YYYY‑MM‑DD

B · Launch‑week checklist

C · Weekly review (30 min ritual)

  1. Read the auto‑generated KPI digest.
  2. Write 3 bullets: what worked, what didn't, what I'm doing about it.
  3. Update the bets board (ICE).
  4. Pick 1 thing to automate further; pick 1 thing to kill.
  5. Send a public build‑in‑public update (compounds distribution).

D · Founder operating rhythm

Time blockActivity
Mon AMStrategy + bets review
Mon PMCustomer calls / demos
Tue–Thu AMDeep work (product, content angles)
Tue–Thu PMReview agent outputs, approvals, customer touch
Fri AMFinance, ops, KPI digest, weekly review
Fri PMLearn / read / build a small experiment
Sat / SunOFF. Non‑negotiable. The agents work; you don't.

E · Repo / folder skeleton

company/
├── README.md              # mission, ICP, current bets
├── agents/                # one folder per agent
│   ├── research/
│   │   ├── charter.md
│   │   ├── prompt.md
│   │   ├── tools.json
│   │   └── evals/
│   └── marketing/...
├── workflows/             # n8n / sdk flows as code
├── prompts/               # shared prompt library
├── knowledge/             # SOPs, brand voice, do‑not‑say
├── data/                  # schemas, seed data
├── infra/                 # Terraform / scripts
├── ops/
│   ├── runbook.md
│   ├── on‑call.md
│   └── bus‑factor.md
└── product/               # the actual app

Closing principle

"The one‑person AI company is not the absence of a team. It is the presence of a different kind of team — one that never sleeps, scales with capital not headcount, and forgets nothing. Your job is what agents can't replace: judgment, taste, and trust."

— Linh Truong

18 · References & sources

Reading list behind the architecture layers, agent stack, unit economics, legal/risk notes, and roadmap. ARR ceilings, margin bands, and tool costs are illustrative—calibrate to your market and re-check vendor pricing before you budget.

Synthesis of public research and operator practice; diagrams are original unless noted. Not legal, tax, or investment advice—confirm statutes and filings with qualified professionals.

Solo AI companies & the 2026 inflection (§1–2)

  1. Stanford HAI, AI Index Report. Capabilities, economics, and deployment trends. aiindex.stanford.edu
  2. McKinsey & Company, “The state of AI.” Enterprise adoption and productivity context. mckinsey.com/state-of-ai
  3. Epoch AI and major provider pricing pages—for inference-cost curves in §2 (re-benchmark quarterly).
  4. Yao et al., “ReAct: Synergizing Reasoning and Acting in Language Models.” ICLR 2023—agent tool-use baseline. arxiv.org/abs/2210.03629

Operating model, architecture & agents (§3–7)

  1. Zaharia et al., “The Shift from Models to Compound AI Systems.” Berkeley BAIR Blog, 2024. BAIR blog
  2. Lewis et al., “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.” NeurIPS 2020. arxiv.org/abs/2005.11401
  3. Es et al., “Ragas: Automated Evaluation of Retrieval Augmented Generation.” 2023. arxiv.org/abs/2309.15217
  4. Huyen, Designing Machine Learning Systems. O’Reilly—data, deployment, observability (§10).
  5. LangChain / LangGraph documentation—orchestration patterns cited in §6–7. langgraph
  6. Anthropic, Claude documentation & Agent SDK materials—harness and tool-use references in §6–7.

Indie SaaS, pricing & unit economics (§8, §13)

  1. Walling, The SaaS Playbook / Start Small, Stay Small. Bootstrap SaaS economics and lifestyle-business framing.
  2. Shapiro & Varian, Information Rules. Harvard Business Press—versioning, bundling, metering (usage/token pricing).
  3. MicroAcquire / Acquire.com marketplace materials—indie M&A and SDE multiples for §13 exit paths.
  4. OpenView, SaaS benchmarks (annual)—ARR, churn, and margin bands; verify the current edition.

Discovery, workflows & prioritization (§5, §9, §15)

  1. Ries, The Lean Startup. Validated learning and build-measure-learn loops.
  2. Torres, Continuous Discovery Habits. Interview cadence and opportunity mapping.
  3. Intercom, RICE prioritization—scoring framework related to ICE in §15. Intercom blog
  4. Amplitude, North Star Playbook—North Star and KPI design in §15. amplitude.com/north-star

Finance, legal, tax & compliance (§11–12)

  1. IRS publications (U.S.)—entity classification, quarterly estimated tax, R&D credit eligibility; consult a CPA.
  2. Stripe Atlas guides—incorporation and payments basics for solo founders.
  3. European Union, Artificial Intelligence Act (Regulation (EU) 2024/1689). EUR-Lex
  4. NIST, AI Risk Management Framework (AI RMF 1.0). nist.gov/ai-rmf
  5. OWASP Top 10 for Large Language Model Applications. OWASP LLM Top 10
  6. GDPR (EU) and applicable U.S. state privacy laws—for customer data handling in §11–12.

Evals, safety & failure modes (§12, §16)

  1. Zheng et al., “Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena.” NeurIPS 2023. arxiv.org/abs/2306.05685
  2. OWASP LLM Top 10—prompt injection and exfiltration mitigations in §12.
  3. Google, Site Reliability Engineering—SLOs, error budgets, on-call patterns for solo operators with agents. sre.google
KPI strip (§1). Headcount, ARR ceiling ($1–10M), gross margin (80–95%), and 180-day launch targets are planning heuristics—not industry averages. Ground your model in your ICP, price, and cost telemetry; see §18 sources above for external benchmarks.