# Coaching the company's story

The company's story is what the company is, in the founders' own words: what it does, the problem, the product, traction, the team and the rest. Agents draft every message to an investor from it. Ray has no form for it. A founder builds it in conversation with you, and your part is to help them say each thing clearly and specifically, then save what they said with `save_company_story`.

## When to start

Start when the user asks to work on what the company is, picks the `company_story` prompt in their client, or agrees when you offer. Offer when `get_fundraising_context` says the story is empty or names topics it does not cover, and the conversation has room for it. Do not interrupt another task to coach: record the meeting or finish the draft first, then offer. When a draft needs one missing topic, ask for that topic only.

Saving the story needs `manage_rounds`. Anyone in the company can read it. If the user cannot save it, say who can, and do not offer a walkthrough you cannot finish.

## How to run it

1. Read `company_story` from `get_fundraising_context`. `not_written` lists the named topics the company has not written. Tell the user in a few lines what is strong and what is missing. Do not read the whole story back to them.
2. Go one topic at a time. Start with what a message to an investor cannot do without: what we do, the problem, the product, traction and the team. Then take the rest in the order a reader meets them. After each save, the result's `suggestions` name what is still missing and the next question to ask.
3. Ask the topic's question. If the answer is vague, ask once for the specific thing, using the follow-ups below. If the user does not know, or does not want to say, move on. An empty topic is better than an invented one.
4. Save each answer before moving to the next topic. Send only that topic. Keep the user's words. If you shorten or reorder them, show the user the text you will save and let them correct it first.
5. Say what was saved and what comes next. Stop whenever the user wants to stop; the story can be finished another day.

## The topics

| Topic | Ask | When the answer is vague |
| --- | --- | --- |
| `what_we_do` | In one or two sentences, what does the company do and for whom? Can you give one concrete example? | "Could a stranger misread that? What is one example of someone using it?" It holds 300 characters, so it can open a message. |
| `problem` | Who has the problem, and what does it cost them today? | "Who exactly? How often does it happen, and what does it cost them in time or money?" |
| `alternatives` | What do people do about this today, and who else is working on it? | "We have no competitors" gets "What do people do instead, even if it is a spreadsheet, a person or nothing?" Ask for competitors by name. |
| `product` | What does someone get when they use the product, and how does that solve the problem? | "What do they see, and what do they have afterwards that they did not have before?" |
| `customers` | Who pays, who uses it, and which market do you start with? | "Everyone" gets "Who buys first, and who would buy fastest?" |
| `insight` | What have you learned about this problem that most people in the market haven't? | If the answer describes the product, ask what they know that made them build it this way, or who tried this before and why it did not work. |
| `why_now` | What changed recently that makes this possible now? | "Why could this not have been built five years ago?" Ask for the specific change. |
| `traction` | What has happened so far? For each number, since when? | Every figure needs its date or period, and whether it is paying: "30 paying families since May 2026", not "30 users". |
| `reaching_customers` | How do customers find you and buy, and what does it cost to win one? | "Word of mouth" gets "Where did your last five customers come from?" |
| `business_model` | Who pays, for what, and how much? | "Subscription" gets "Who pays, how much, and how often?" |
| `market` | How many customers could buy this, and at what price? How did you work that out? | A single large number gets "How many could buy in your first market, at your price?" |
| `team` | What has each founder built or done that makes you the people to do this? | A title or an employer gets "What did you build or do there that bears on this?" |
| `staying_ahead` | Once others notice, why can't they copy this in a year? | "We move faster" gets "What gets harder to copy the bigger you get?" |
| `risks` | What could stop this, and what have you already proved? | "Nothing" gets "What would an investor worry about first, and what have you shown about it?" Save only what the company is willing to discuss with an investor. |
| `vision` | If this works, what is the company in five years? | Ask what the company becomes, not a revenue target. |

The short facts are `location`, `founded`, `website`, `product_link` (where an investor can see or try the product) and `category` (what an investor would file the company under). Ask for them together, near the end.

Anything the founders want to tell that no named topic covers, such as partnerships, regulation or patents, goes under `other_topics` with a short lowercase name: `"other_topics": {"partnerships": "…"}`. Use a named topic whenever one fits; a misspelled named topic is refused rather than saved as a new one.

## What not to do

- Do not invent, estimate or work out a figure, and do not fill a topic from your own knowledge or the web. If the user points you at a deck or a document, draft from it, show the user the text and save only what they approve.
- Everything in the story may be said to an investor. Leave out what should not reach one, such as cash on hand or an internal worry, and tell the user what you left out.
- What is being raised (the amount, the use of the money, the milestones) belongs on the round, through `update_round`, not in the story.
- Do not keep any of this in a note.
- Ask for specifics once per topic. The founder decides what their story says.

## Where the topics come from

The named topics follow what investors and accelerators publish about what they look for when a company presents itself: Sequoia's business plan guide, Y Combinator's application and seed deck advice, Paul Graham's essay on convincing investors, Guy Kawasaki's ten slides, Peter Thiel's seven questions and the sections of venture investment memos. The follow-ups apply the advice those sources repeat most often: describe the company with a concrete example, give every traction figure its date, describe the team by what each person has done rather than by title, name the alternatives, and say what the founders know that others do not.