Freshsales

Promptfree, for everyone, and the only version there is

Copy the prompt for Claude CodeCopy the prompt for CodexCopy the prompt for Cursor

Copy

Build me the sales CRM that replaces Freshsales.

One rule decides whether a CRM is used at all: **if a salesperson has to type what they already sent, they will not.** Build the capture first, the reporting second, and everything else only when asked.

STACK
- Node 20+ with Fastify, server-rendered HTML with a little vanilla JS
- SQLite through better-sqlite3, WAL mode, FTS5 across notes and messages
- IMAP or a mail API for the capture
- Caddy in front

THE DATA MODEL
- companies, people, deals: value_cents, currency, stage, expected_close, owner_id, outcome, lost_reason
- stage_changes: id, deal_id, from, to, at, actor — **append-only; every pipeline report is computed from it**
- activities: id, subject_kind, subject_id, kind, body, at, actor
- messages: id, deal_id, person_id, direction, subject, body_text, message_id, at
- tasks: id, subject_kind, subject_id, title, due_at, done_at, owner_id
- Money in integer minor units with the currency stored, never a float

CAPTURE WITHOUT TYPING
- Sync the salesperson's mailbox, match on address to a person and through them to a deal
- A blind-copy address that files anything sent from a phone
- **Only sync messages with people already in the CRM.** Pulling an entire mailbox drags in private correspondence, which is a privacy problem and an employment one
- Ambiguity is resolved by asking. A message filed against the wrong deal is worse than an unfiled one

THE TELEPHONE, HONESTLY
- Do not build telephony. Use a provider and store the call record: who, when, how long, and a note
- The value is that the call appears on the deal, and that is a webhook and a row

THE PIPELINE
- Stages with a probability, and two forecasts side by side: value times probability, and value times the historical conversion rate at that stage. **The gap between them is the most useful number a sales manager has**
- Time in stage, and an ageing list of deals that have not moved — that list is the weekly meeting
- A lost reason required on close. Without it, losing teaches nothing

SCORING, IF AT ALL
- Points for behaviours, decayed over time, and **the contributing events shown on the record**. A score nobody can explain is a score sales ignores
- Check it against outcomes before trusting it: do high scorers actually convert? If not, the weights are guesses

MAKING IT USED
- One screen per deal with everything on it
- A next action on every open deal, and a list of deals without one. That list is the only nagging a CRM should do
- Fast keyboard entry for a note — the action performed twenty times a day

WHAT MATTERS MOST
Mailbox capture scoped to known contacts, and a required lost reason. The first makes it used; the second makes it worth using.