Turn your evaluated idea into a prompt you can build from.

1Build an app called **Handoff**. Turn customer call transcripts into structured handoff records that help solo SaaS founders preserve context and act on every conversation.
2
3## App summary
4
5Handoff is a calm web app for one or two person SaaS teams. The user pastes or uploads a customer call transcript and gets back one structured record: a short summary, the action items, the blockers and the open questions. Every record can be shared as a read only link and stays searchable, so the next conversation starts where the last one ended.
6
7Build a single column workspace, not a dashboard. The first useful result should land in under a minute from an empty account. The structured record is the product; the summary is a byproduct, and transcription accuracy is never the selling point.
8
9## Target users
10
11The primary user is a solo founder or a two person SaaS team running three to eight customer calls a week, selling something technical enough that the detail of a call matters. They currently keep notes in a document nobody rereads, and they lose context between calls.
12
13Design for one person doing several jobs at once:
14
15- They work in short gaps between calls, so every screen must be readable at a glance.
16- They already have a transcript from a recorder, so paste must always work.
17- They share context with a co founder or a new hire, so output must be presentable without editing.
18
19Do not build for sales teams, managers or enterprise buyers. No seats, no hierarchy, no reporting layer.
20
21## Problem
22
23Founders finish a good customer call and lose most of it within a week. Recordings pile up, notes are written in a different shape every time, and the decisions inside the conversation never land anywhere reusable. When someone else needs the context, the only option is to listen again or rebuild it from memory.
24
25Existing tools produce transcripts and loose summaries. Nothing produces a consistent record a person can act on and hand to someone else.
26
27## Solution
28
29One transcript in, one structured record out, stored and shareable.
30
311. The user pastes text or uploads a transcript file.
322. The app produces a record with four fixed blocks: Summary, Action items, Blockers, Open questions.
333. The user can edit any block inline; edits save immediately.
344. The record gets a shareable read only link with a clean public view.
355. Every record is stored in a searchable history, filterable by account name and date.
36
37Keep the four blocks fixed. The consistency is what makes the record valuable, and it is what makes past records comparable and searchable.
38
39## Key features
40
41- **Paste or upload intake.** A single input that accepts pasted text or a .txt, .md, .vtt or .srt file. Show a character count and an estimated processing time.
42- **Structured record generation.** Produce Summary (three to five sentences), Action items (owner plus task), Blockers (what is stopping progress), Open questions (what to ask next time).
43- **Inline editing.** Every block and every list item is editable in place, with autosave and a small saved indicator.
44- **Account grouping.** Each record belongs to an account name so a history of one customer reads as a timeline.
45- **Search.** Full text search across summaries, items and questions, with account and date filters.
46- **Share links.** One click to create a read only public link; one click to revoke it.
47- **Copy and export.** Copy the record as markdown, or download it as a .md file.
48
49## MVP scope
50
51Ship this and nothing more:
52
531. Email sign in.
542. Paste or upload a transcript.
553. Generate the four block record.
564. Inline edit and autosave.
575. Record list with search and account filter.
586. Read only share link with revoke.
59
60Leave out for now: calendar integrations, recorder integrations, live recording, team permissions, mobile apps, analytics dashboards, CRM sync.
61
62### Phase 2
63
64- Recorder and calendar integrations that create records automatically.
65- Account timeline view that summarises what changed across several calls.
66- A second record type for internal meetings, only after the customer call record is used weekly.
67- Lightweight collaborator access, read and comment only.
68
69## Monetization
70
71One workspace subscription at 29 USD per month, including up to 40 records a month and unlimited stored history. No per seat pricing while the buyer is one or two people.
72
73- Free trial: three records, no card required, full share and export.
74- Overage: additional records in blocks of 20 for 9 USD.
75- Anchor the pricing message on the value of the record, not on minutes transcribed.
76
77Cap single transcript length in the MVP and show the limit clearly rather than truncating silently, so long calls do not erode the margin.
78
79## Differentiation
80
81The alternatives are meeting recorders with summary features, note apps with AI blocks, and a manual document. All of them stop at readable text.
82
83Handoff is different in three ways:
84
85- The output is a fixed, structured record, not free form prose, so it is comparable across calls.
86- It is designed to be handed to another person, so the shared view is a first class screen.
87- It is scoped to customer calls for small SaaS teams, so nothing in the interface is built for managers.
88
89Never position this as a transcription product and never compete on accuracy of audio processing.
90
91## User journey
92
931. **Sign up.** Email and password. No onboarding tour, no questionnaire.
942. **Empty state.** A single large input with the label "Paste a call transcript" and an upload option. It is focused on load.
953. **Processing.** A short progress state with the four block names appearing as each is filled.
964. **Record view.** The structured record, editable inline, with Copy, Download and Share in a compact toolbar.
975. **Share.** Creating a link shows the public preview immediately so the user trusts what the recipient sees.
986. **Return visit.** The home screen becomes the record list, newest first, with search at the top and the intake input one click away.
99
100## Suggested design style
101
102### Mood
103
104Dark, calm and dense but quiet. Editorial rather than app like. Generous line height, restrained borders, no gradients, no decorative illustration. The product should feel like a well set document.
105
106### Brand colors
107
108- Background: #0B0C0A
109- Surface: #14150F
110- Border: rgba(255, 255, 255, 0.08)
111- Primary text: #F5F6F1
112- Muted text: #9A9C93
113- Accent: #DEF947, used only for focus, active state and the primary action
114
115### Typography
116
117Space Grotesk for headings and record block titles. Inter for body text, list items and interface labels. Set body text at 15 to 16 px with 1.6 line height. Use a monospace face only for the copyable markdown view.
118
119### Refs
120
121Linear for interface density and keyboard feel, Notion for editable block structure, Stripe docs for readable long form typography.
122
123## Data structure
124
125- **users**: id, email, created_at
126- **workspaces**: id, name, owner_id, plan, records_used_this_period
127- **accounts**: id, workspace_id, name, created_at
128- **records**: id, workspace_id, account_id, title, call_date, source (paste or upload), status, created_at, updated_at
129- **record_blocks**: id, record_id, type (summary, action_item, blocker, open_question), position, content, owner_label
130- **share_links**: id, record_id, token, created_at, revoked_at
131- **transcripts**: id, record_id, raw_text, word_count, created_at
132
133Rules:
134
135- Every row is owned by a workspace, and every read and write is scoped to the requesting user's workspace.
136- Raw transcripts are stored separately from record blocks so they can be deleted without losing the record.
137- A share link grants read only access to one record and its blocks, never to transcripts or history.
138- Deleting a record deletes its blocks, transcript and share links.
139