By the Onsites AI team · Last updated · 5-minute read
Solo founders lose deals in exactly two places: the lead that never became a quote, and the quote that expired unwatched between the hundred other things a founder does. The cure is not a 47-stage pipeline; it is a five-column board small enough to fit on one screen and one weekly ritual short enough to survive a founder's week. Stages: Lead (wrote in, needs a reply), Quoted (money sentence sent, clock running), Negotiating (requotes and terms, versions stacking on one thread), Won (accepted — order number born), Delivering (fulfillment, promise log, the invoice trail). The conversation-built records mean the pipeline populates itself — the lead column exists because the channel exists, not because Sunday-you remembered to type — and the weekly review is the only discipline the system actually demands.
The founder's resistance is not laziness; it is experience: CRMs were sold as systems to maintain, and a one-person week has no maintenance slot. The five-board answers each objection by being smaller than it. "I don't type data" — the records build from threads; typing is the thirty seconds a week that capture needs. "Pipelines don't fit my chaotic week" — five columns is one glance, and the glance is Friday's ritual, not a second job. "I lose track during the week" — that is the board's purpose: the deal state (quote clock, negotiation versions, promise log) lives on the card, not in memory, so the Friday review reads state instead of replaying archaeology. "Sales tools cost money I haven't earned" — the desk's whole commercial layer (pipeline, documents, invoice trails) ships in the free tier with the CRM; the founder's expense arrives with their fourth seat or fiftieth GB, which is the honest point in a business's life to pay. The board wins the adoption argument the sales pages never could: by being less work than the sticky notes it replaces, from day one.
The pipeline review is one calendar block with one agenda, and its smallness is the design. Friday, twenty minutes: scan the board left to right — every lead answered (the silence here is where strangers quietly become someone else's client), every quoted deal's expiry checked (the validity discipline: the lapsed quote gets its one-line follow-up or its honest decline), every negotiating thread advanced one turn or honestly stalled (stage-age is the sort order: the oldest card gets the founder's next hour, full stop), every delivering order's promise log current. That is the whole ritual — no forecasting theater, no stage-weighting formulas, no CRM retreats: the founder's edge at this size is not sophistication but nothing dropped, and the twenty minutes exist precisely because everything it checks is a lookup on one screen. The summaries make it honest: each thread's state readable in two lines instead of a re-read, which is what keeps the review at twenty minutes as the week's volume grows.
Run the board a quarter and it starts teaching back, because the stage ages carry the honest diagnostics. A fat Lead column says the desk answers, but never quotes — a pricing or specificity problem (the chat-to-quote flow fixes the second). A fat Quoted column says offers go out but die without a word — the expiry ritual gets added, and the lapsed-quote follow-up starts converting. A fat Negotiating column is the good problem: real demand, bottlenecked on your hours — stage-age sorts it, and the phone call is usually the unlock. A thin Delivering column beside a fat Won one means fulfillment, not sales, is the constraint — the week's honest hour moves. Every one of these readings is a lookup on a screen the founder already keeps, which is the quiet design bet of the whole board: a pipeline this small cannot lie to you, because nothing in it needs a formula — the stage-age sort is just the truth, ordered.
Migration from the founder's existing system (inbox, notebook, memory) is one deliberate afternoon, in order. Hour one, build the backstop: the five columns exist; the channels flow; every new conversation auto-creates its record — nothing new to run, because the desk already answers everyone. Hour two, backfill the living, ignore the dead: current deals (quote out, negotiating, delivering) get cards created from their emails — the migration guide's email-cutover habit, applied commercially; deals older than a quarter stay where they lived, unless a payment or promise is still open on them, because a backfilled pipeline lies about health. The first Friday: the twenty minutes run on the half-real board; the second Friday they run on a full one. That is the entire adoption cost — two hours, once — and the honest comparison that decides it is not board-versus-CRM but board-versus-the-current-system: whatever the sticky notes are catching, they are catching asynchronously, in memories that were never the buyer-facing record. The board is the same information, in the place where the next conversation reads it — which is the difference between a pipeline and a pile.
Automation earns its keep on the structure and refuses the judgment. It does: the auto-built records (contacts from threads), the expiry clocks on quotes, the stage-age sort, the overdue ladder's first touch scheduled, the invoice statuses current, and the summaries that make Friday's twenty minutes possible at all. It refuses the last mile: what to concede on the tough requote, when a stalled deal gets the phone call instead of the third email, which lead deserves the early slot next week — the calls where a founder's judgment is the product. The honest adoption line, from the copilot guide: AI drafts, AI tracks, AI reminds; the human prices, decides, and signs. A pipeline run this way fits in one screen, one ritual, and one mind — which is the entire design, because the alternative is not a better system but a bigger one, and the founder's week does not have a second twenty minutes to spare.
What does a solo founder's pipeline actually need?
Five stages on one screen: Lead (wrote in, needs reply), Quoted (offer sent, validity clock running), Negotiating (versions stacking on one thread), Won (order number born), Delivering (promise log and invoice trail). The conversation-built records populate it automatically; the weekly twenty-minute review is the only ritual it demands.
How should a founder run the weekly pipeline review?
Twenty minutes, one agenda: leads answered, quote expiries checked (lapsed ones get a follow-up or an honest decline), negotiating threads sorted by age (oldest gets the next hour), delivering orders' promise logs current. No forecasting theater — the founder's edge is nothing slipping, not sophistication.
What does automation do in a solo pipeline — and refuse?
It handles the tracking: expiry clocks, stage-age sorts, scheduled first touches of the chase ladder, thread summaries. It refuses judgment: what to concede, when to call instead of email, which lead gets the slot. AI drafts and tracks; the founder prices, decides, signs.
How does a pipeline stay honest without data entry?
Records are born from threads, so the pipeline's left side populates itself; the founder's only maintenance is the thirty-second field capture while a thread is open and a quarterly health pass on the account list. Small enough to be true — that is the design constraint that keeps it alive.
Where do expiring quotes fit into a solo pipeline?
In Quoted, with a visible clock: deliberate validity (two weeks for B2B approval cycles, shorter for spot deals), an announced expiry that re-opens the lane once, and the one-line expiry follow-up that turns lapsed quotes into re-quotes or known declines. Silent expiry teaches buyers to shop elsewhere; announced expiry is the discipline that reads the market.