AI Agent Business Ideas: Start With Bounded Authority
Evaluate AI agent business ideas by permissions, error costs, review effort and measurable outcomes before giving software authority to take action.
Read the field noteArmy of One / Field notes
For the ones
figuring it out.
On building something that lasts,
finding your people, and staying human
through all of it.
Start here / Moats & AI
Startup moats in 2026: what makes a small company defensible?
Startup moats in 2026 come from advantages that strengthen with use. Test workflow knowledge, trust, distribution, and customer outcomes before scale.
Read the field note 21 October 2026 · 4 min readThe latest field notes
50 questions worth spending time withAI wrapper or durable business? Ask who owns the outcome
An AI product becomes a business through paid outcomes, reliable delivery, and workable economics. Test what customers need beyond a model response.
Read the field noteBoring B2B Startup Ideas: Find the Admin Work Someone Owns
Find boring B2B startup ideas in recurring admin work, then test ownership, exceptions and budgets before replacing an entire business workflow.
Read the field noteBuild or buy an internal tool? Compare the whole responsibility
Build or buy internal software by comparing fit, ongoing ownership, and opportunity cost. Use a short worksheet instead of comparing subscription and coding time.
Read the field noteBuilding software is one decision. Maintaining it is another.
Building a prototype and operating software are different commitments. Map recurring ownership, dependencies, support, and recovery before promising more.
Read the field noteHow to have the cofounder equity conversation fairly
Prepare a fair cofounder equity discussion by clarifying future work, commitment, decision rights and departure questions before seeking legal and tax advice.
Read the field noteCofounder matching platforms in 2026: choose the right route
Compare YC matching, Entrepreneurs First and Antler India by commitment, location and purpose, then use shared work to evaluate a potential cofounder.
Read the field noteA cofounder trial project: what to test in two weeks
Design a two-week cofounder trial with shared responsibilities, customer learning and a clear ending. Discuss access, payment and ownership before work starts.
Read the field noteCofounder or first hire: which relationship does your startup need?
Choose a cofounder when you need shared direction and company responsibility. Choose a first hire when you can define, fund and manage an ongoing role.
Read the field noteHow a solo startup can compete when a big AI platform adds your feature
Compete with large AI platforms through a specific customer job and accountable delivery. Test complementary value while planning for dependency changes.
Read the field noteCoworking and accountability for solo founders: a weekly format
Try a weekly founder coworking session with a clear task, quiet work and a short review. Accountability should support progress without public pressure.
Read the field noteCustomer Discovery vs Feature Requests: What Should You Build?
Separate customer discovery from feature requests with an evidence ledger that records the job, consequence, workaround and smallest useful response.
Read the field noteCustomer Interview Questions for Founders: A Past-Behaviour Script
Use customer interview questions that explore recent behaviour, existing workarounds, consequences and buying decisions without leading people toward your idea.
Read the field noteWhat if your customer vibe codes your product?
When a customer can build a version of your product, compare ownership costs and outcomes. Find the job they still want a supplier to manage reliably.
Read the field noteA data moat is not a pile of customer data
A data moat needs legitimate rights, reliable inputs, and a feedback process that improves a paid outcome. More stored records alone prove little.
Read the field noteHow solo founders can build distribution that compounds
A distribution moat grows from useful relationships and repeatable discovery. Run focused experiments and measure conversations, activation, and referrals.
Read the field noteHow to find a technical cofounder when you do not code
Attract a technical cofounder with customer evidence, a clear contribution and realistic product ownership, rather than an idea and an unpaid development brief.
Read the field noteHow to Find Your First Ten Customers as a Solo Founder
Find your first ten customers with a narrow segment, manual outreach, a clear first-use offer and a weekly learning ledger that separates interest from sales.
Read the field noteFounder burnout warning signs: notice the pattern, change the work
Understand WHO’s description of burnout, notice changes in your working life and review workload, support and professional help without self-diagnosing.
Read the field noteFounder Community vs Audience: Build Reasons to Return
Understand founder community vs audience through member relationships, useful participation and repeat attendance, with a practical community benefit ledger.
Read the field noteFounder decision fatigue: reduce the decisions you repeat
Reduce repeated founder decisions with clear defaults, decision owners and review dates. Protect attention without relying on unproven brain-energy claims.
Read the field noteFounder Distribution on LinkedIn, Reddit and X Without Spamming
Build founder distribution on LinkedIn, Reddit and X through relevant discussion, honest affiliation and platform-aware participation rather than repeated pitches.
Read the field noteFounder loneliness: build a small support circle that lasts
Build a small founder support circle with consent, practical check-ins and clear boundaries, so honest conversation does not become another performance.
Read the field noteHow to build a founder network in Delhi NCR
Build a useful founder network in Delhi NCR through repeat local conversations, relevant official communities and small practical commitments you can keep.
Read the field noteFounder peer group boundaries: make support safe enough to use
Set clear founder peer group boundaries for consent, confidentiality, advice and promotion. Honest conversations need choice, limits and a respectful exit.
Read the field notePricing Your First B2B Product: Learn Value and Delivery Cost
Price your first B2B product by separating customer value, delivery cost and a clear billing unit, then use small, honest commercial tests to learn.
Read the field noteHow to take time off as a solo founder without leaving loose ends
Plan time away with clear customer expectations, sensible coverage and a return note. Rest does not need to earn its place through a productivity promise.
Read the field noteHow to find a cofounder in 2026 without rushing the decision
Find a cofounder through clear role needs, relevant founder channels and small shared projects. Use this practical search and evaluation process in 2026.
Read the field noteReliable integrations can be a startup advantage. Here is the hard part.
Integration reliability means handling retries, duplicates, stale state, and recovery. Start with one important connection and make its failures understandable.
Read the field noteLocal Service Software in India: Start With One Working Day
Explore local service software in India through one operator, one workflow and one locality, accounting for language, data quality, distribution and support.
Read the field noteHow to Run a Paid Pilot Before Building the Full Product
Design a paid pilot before building with one outcome, clear data access, success measures and an end decision, while keeping manual delivery visible.
Read the field noteQuestions to discuss before choosing a cofounder
Discuss ambition, availability, money, authority and disagreement before choosing a cofounder, then compare the answers with how you actually work together.
Read the field noteRetention Before More Features: Diagnose Why Users Do Not Return
Diagnose retention before adding features by separating acquisition fit, setup, first value and repeat-use failures, then review small cohorts honestly.
Read the field noteGive customers an exit plan before asking them to trust your startup
A clear software exit plan reduces the fear of choosing a young company. Explain exports, access, continuity, deletion, and the limits of your promise.
Read the field noteSoftware Maintenance as a Service: Scope the Promise Before Selling It
Assess software maintenance as a service with a scope and ownership worksheet covering access, support hours, inherited defects and delivery economics.
Read the field noteHow to budget maintenance time as a solo founder
Budget maintenance in hours and cash, then compare it with your actual week. Use a simple capacity worksheet that includes support and interruptions.
Read the field noteSolo founder mental health: design support before a crisis
Support your mental health as a solo founder with manageable work, clear boundaries and trusted people, while recognising when professional care is needed.
Read the field noteSolo founder or cofounder: decide from the work ahead
Decide whether to stay solo or find a cofounder by examining workload, judgment, support and the authority you are willing to share as the business grows.
Read the field noteValidate a Startup Idea in Seven Days Without Pretending It Is Proven
Use a seven-day startup validation plan to investigate a problem, observe real behaviour and request a specific commitment before investing heavily in code.
Read the field noteHow to Find Startup Ideas in Spreadsheets
Find startup ideas in spreadsheets by examining exception handling, routing and change history, then test whether a new product improves the actual job.
Read the field noteStartup Ideas for Solo Founders in 2026: A Practical Hypothesis Map
Compare startup ideas for solo founders in 2026 using buyer access, recurring pain, delivery limits and evidence, with a practical hypothesis map.
Read the field noteAfter startup rejection: a practical debrief without judging yourself
Separate a startup rejection from your identity, capture useful evidence and choose a proportionate next action without forcing a positive lesson immediately.
Read the field noteSystems of action vs systems of record: where should a founder start?
Systems of action execute work; systems of record preserve authoritative state. Choose a narrow starting point with approvals, recovery, and clear ownership.
Read the field noteIs a system of record still a moat in 2026?
A system of record earns durability through trusted history and useful workflows. Learn why storing data alone is weak, and how exports can help.
Read the field noteTechnical debt before product-market fit: what can wait?
Before product-market fit, accept reversible shortcuts that accelerate learning. Keep a debt register and protect access, data integrity, and recovery.
Read the field noteTrust as a software moat: show customers what happens when things fail
Software trust grows through observable reliability, honest limits, and clear recovery. Use practical evidence rather than vague enterprise-ready claims.
Read the field noteHow to Choose a Vertical AI Startup Wedge
Choose a vertical AI startup wedge around one role, one recurring job and a reviewable output, with explicit human handling for costly exceptions.
Read the field noteA production checklist for your vibe-coded MVP
Before real customers rely on a vibe-coded MVP, verify ownership, access, data handling, recovery, and the critical workflow. A demo is only the start.
Read the field noteHow a small startup can build a workflow moat without trapping users
A workflow moat comes from making a recurring job easier to finish. Map handoffs, exceptions, and approvals while giving customers a credible way to leave.
Read the field note