---
title: "Understand the person’s goals"
pattern_id: "understand-the-persons-goals"
trait: "Invested"
type: "Humorphic pattern brief for coding agents"
spec_version: 3
canonical_url: "https://humorphism.com/patterns/understand-the-persons-goals"
markdown_url: "https://humorphism.com/patterns/understand-the-persons-goals.md"
index: "https://humorphism.com/llms.txt"
all_patterns_in_one_file: "https://humorphism.com/llms-full.txt"
normative_section: "Rules"
rule_prefix: "UPG"
rule_ids: ["UPG-R1", "UPG-R2", "UPG-R3", "UPG-R4", "UPG-R5", "UPG-R6", "UPG-R7", "UPG-R8", "UPG-R9"]
---

# Understand the person’s goals

## How to read this file

You are an agent. Someone gave you this file so you can add one behavior to their product: the AI teammate in it should learn what the person is trying to achieve before it does work on their behalf. This file is the complete specification for that behavior. Build from it.

**The Rules section is the only normative part.** Everything else exists to make the rules buildable: the capabilities tell you which parts of the pattern the product can support, the reference screens show you the behavior in motion, the examples give you four shapes it can take, and the checklist is how you confirm you are done. Examples and the checklist point back to rule IDs rather than restating requirements, so each requirement lives in exactly one place. If any other section seems to contradict a rule, the rule wins.

Work in this order:

1. **Find where this belongs** in the product: the first session, setup or onboarding, and the start of any new project or task where the outcome is not obvious.
2. **Read Rules.** All nine apply to whatever you build.
3. **Check Capabilities** against what the product already has. Build the examples its capabilities support, and take the stated fallback for the rest.
4. **Pick your examples.** If you build only one, build Example 2.
5. **Build it in the product's own design system, components and voice.** The reference screens say what each screen does and why. They are not a specification of how it must look. Keep the behavior, change the expression.
6. **Confirm against the checklist** before you call it done.

Humorphism is a design philosophy for products with AI in them. It replaces the user interface, built for operating tools, with the human interface, built for collaborating with AI teammates. There are 21 humorphic patterns grouped into 7 traits. You do not need the others to build this one, but if you can fetch URLs: [llms.txt](https://humorphism.com/llms.txt) lists every humorphic pattern with a one-line summary and a link, and [llms-full.txt](https://humorphism.com/llms-full.txt) has all of them in one file.

## The pattern

_Trait: Invested. Aligns with the person’s goals and cares whether they succeed_

Colleagues start by agreeing on what the work is for. A good AI teammate learns the goal early, because everything it surfaces or prioritizes depends on it.

**Build this.** An AI teammate plays back what it already knows and asks what the person is trying to achieve, because every later recommendation depends on that answer.

**Avoid this.** A tool asks for name, role, and industry it already has on file, because it treats setup as a form rather than a conversation.

The difference to carry into your build:

| Tool | AI teammate |
|---|---|
| Asks about the person (name, role, industry) | Asks about the outcome they want |
| Starts blank | Starts from what the product already knows |
| Asks for facts it already holds | Plays those facts back to confirm |
| A form gates the product | A short exchange starts the work |
| The answer disappears into a profile | The answer is saved, visible, and used later |

## Rules

Normative. All nine hold for everything you build from this file. Each one says what to do, why it matters, and what it looks like once it is built.

| ID | Rule | Why | Built correctly when |
|---|---|---|---|
| **UPG-R1** | Ask about the outcome, not the profile. | The outcome is what changes the work. A job title does not. | The first thing the AI teammate asks for names a result the person wants, not an attribute of the person. |
| **UPG-R2** | Never ask for what the product already has; show it so the person can correct it. | Re-asking tells the person the AI teammate did not look. | No field the product can populate arrives empty. |
| **UPG-R3** | Play back before you ask. Lead with what you found, then ask only for what is missing. | Leading with what was found makes the question smaller and the AI teammate more credible. | The opening states what was found, then asks for the gap. |
| **UPG-R4** | Suggest, do not decide. Mark one option as the suggestion, leave it unselected, and always offer a way out. | A suggestion earns trust. A preselected default quietly takes the decision. | One option is tagged as suggested, nothing is selected on arrival, and a neutral or free-text escape exists ("Treat both equally", "Something else"). |
| **UPG-R5** | Name your sources. When you show something you found, say where it came from. | Context without a source is indistinguishable from a guess. | Every played-back fact carries its origin ("in your plan", "in your Calendar"), and its age when it may be stale. |
| **UPG-R6** | Confirm what you saved, in one short line. | Silence leaves the person unsure the answer landed. | After the person answers, one line confirms the answer was stored, naming it where space allows ("Got it. I saved your goal."). |
| **UPG-R7** | Use the goal later. A goal that never changes what gets surfaced or prioritized is just a form field. | This is what separates the pattern from a themed onboarding screen. | Something later in the session visibly refers to the goal, or is ordered by it. |
| **UPG-R8** | Keep the goal editable. | A goal the person cannot correct becomes wrong and stays wrong. | The saved goal is visible after the exchange, and can be changed or removed. |
| **UPG-R9** | Keep it short. One question at a time, a few options, then get to work. | A long intake is a form wearing a conversation's voice. | One question per turn, and the person reaches real work within a few exchanges. |

**UPG-R7 carries the pattern.** Build the downstream use of the goal before you polish the question that captures it. A build that satisfies every rule except UPG-R7 has produced a form field with better copy: it will look like the pattern and behave like the tool version.

## Capabilities

Build the examples the product's capabilities support. Do not stub a capability the product lacks, and do not write copy that implies one it lacks.

| Capability | What it is | Used by | If the product lacks it |
|---|---|---|---|
| Memory | A place that keeps facts, preferences, rules, and open threads across sessions. The person can see everything the AI teammate remembers and edit or delete any of it. | Example 1, and keeping the goal across sessions in any example | Keep the goal for the current session only, and say so. Do not claim to remember it next time. |
| First-party data | Read access to what the product already knows about the person (account, history, existing data), available to the AI teammate before it asks anything. | Example 3 | Skip that example and ask for the goal directly. Never say "I found" or play back data the product does not hold. |
| Third-party integrations, with permission | Scoped, revocable access to the person’s connected tools, requested at the moment it becomes useful, with a plain statement of what will be read. | Example 4 | Skip that example. Never read a connected tool without asking. |

Example 2 needs none of these. **If you build only one example, build Example 2.**

## Reference screens

These describe the animations on the pattern page frame by frame, so you can build from this file alone. They show the behavior, not the appearance to copy. **Keep the behavior, change the expression:** build in the product's own design system, components, type and voice.

One convention to translate rather than reproduce. On these screens a small orange square precedes every line the AI teammate speaks, so it is always clear who is talking, and the AI teammate's side sits on a warm, slowly moving gradient while the tool's side sits on flat gray. Use whatever your system already uses for "the AI teammate is speaking", "this is a suggestion" and "this is selected". The person's answers register as a change of state, a check or a filled field or a moved slider, rather than as a separate Submit step.

### The AI teammate version (frames 01 to 03, a loop of about 9 seconds)

One card, centered on a warm, slowly moving gradient of pink, orange and blue. The card has a light gray header band and a white body. In the header, the orange square, then the line "I found these goals in your plan. Which should I prioritize?" In the body, three numbered rows, each number in a small gray box:

1. "Grow trial-to-paid conversion", followed by a small gray "Suggested" tag
2. "Reduce repeat support tickets"
3. "Treat both equally"

Nothing is selected at rest.

The loop starts on the bare gradient. A small rounded pill appears with the orange square in it, and widens into the header band as the question arrives one word at a time. The body then opens beneath the header and the three rows fade in, top to bottom. A cursor enters from the lower right and glides up to row 1; the row turns light gray on hover, presses in slightly on the click, and stays gray as the chosen row. The number 1 becomes a green check, rows 2 and 3 fade to a lighter gray, and the question fades out, leaving the orange square alone in the header. The reply "Got it. I saved your goal." arrives word by word in its place. The cursor fades out, the end state holds for about 3.5 seconds, and the loop restarts on the bare gradient.

What to take from it: the goals came from something the person already made, their plan, so the AI teammate offers them back instead of asking from scratch (UPG-R2, UPG-R3, UPG-R5). It marks one option as its suggestion without picking for the person, and keeps a way out (UPG-R4). It says plainly that the answer was saved (UPG-R6). And the whole exchange is one question, answered in one click (UPG-R9).

### The tool version (frame 04, a still screen with no motion)

The same card size and position, on a flat light gray background with no gradient. No orange square, so nothing on screen speaks as an AI teammate. The header reads "Tell us about yourself". Beneath it, three empty text fields with placeholder text, "Your name", "Your role" and "Your industry", and a solid black "Open dashboard" button.

What to take from it: the form asks about the person instead of the outcome they want (UPG-R1), asks for facts the product already holds (UPG-R2), starts from nothing (UPG-R3), and puts a gate in front of the work instead of starting it (UPG-R9). This is the shape to move away from.

## Examples

Each example is one way to apply the pattern in a real product. They are not steps in order, and no product needs all four. Pick the ones that fit the moment and the data the product has.

Each is written as one exchange in four beats: **Cue** (the moment in the product when the AI teammate should act), **Move** (what the AI teammate does), **Response** (what the person does in return), **Resolution** (where things stand afterwards).

"Satisfies" names the rules this example is responsible for. "Also confirm" adds the one thing specific to this example that the rules do not cover.

### 1. Did I get that right? (Play back what you know)

A colleague opens a kickoff by saying back who you are and what you’re after, and you fix the parts they got wrong.

- **Cue:** First session, or a new project where the AI teammate already holds context
- **Move:** Shows what it already knows about the person and their goals as editable fields
- **Response:** Confirms, corrects, or removes anything wrong or stale
- **Resolution:** Work starts from a shared, verified picture instead of a blank form
- **Needs:** Memory.
- **Reference frame 05:** the orange square and the line "Here’s what I found. Look good?" in a gray header. Beneath it, a form that runs off the right and bottom edges of the frame, with three filled, labeled fields: Name "John Smith", Email "john@smithstudio.com", Address "1420 Alder Street", the last cut off by the frame's lower edge. In the animation the empty card appears, the square and the line arrive, then the fields rise in one by one, already filled. A beat later the Email field takes an orange border and a text cursor appears after the email address: it is the field that is out of date, open for the person to correct. The AI teammate fills the fields; the person only edits what is wrong. The email domain on screen is a stand-in; in a build's sample data and fixtures, use a reserved domain such as example.com.
- **Sample lines:** "Here’s what I have from last time. Anything out of date?" / "Thanks, I updated your email."
- **Satisfies:** UPG-R2, UPG-R3, UPG-R5, UPG-R6, UPG-R8.
- **Also confirm:** confirming the whole set takes one action.

### 2. What does success look like? (Ask for the goal)

Before starting, a good colleague asks what success looks like rather than guessing.

- **Cue:** The person is onboarding, or makes a first request where the outcome isn’t obvious.
- **Move:** Asks what outcome the person is working toward, before anything else.
- **Response:** States the goal in their own words, or picks from suggested outcomes
- **Resolution:** Goals are stored and shape what gets surfaced and prioritized later
- **Needs:** nothing beyond the AI teammate itself.
- **Reference frame 06:** the orange square and the question "When should it ship?" Under it, a live summary, "By Jun 7 for $1,875", with the date and the price each on a light orange highlight, over a slider running from "Faster" on the left to "Cheaper" on the right, its fill and handle about a third of the way from Faster. In the animation the question arrives word by word with the summary reading "By Jun 21 for $1,250", the slider rises in with its handle at the Cheaper end, and a cursor takes the handle and drags it toward Faster. As it moves, the date and price tick through "Jun 14" and "$1,560" and settle on "Jun 7" and "$1,875". The goal is asked as a trade-off the person can feel, with the consequence of each position visible before they commit.
- **Sample lines:** "What would make this a success for you?" / "Should I optimize for speed or cost?" / "No preference? I’ll aim for the earliest date under your budget, and you can change that anytime."
- **Satisfies:** UPG-R1, UPG-R4, UPG-R6, UPG-R7, UPG-R9.
- **Also confirm:** the person can answer in their own words, not only from the offered options.

### 3. Here’s what I already know (Warm-start from what’s already in the product)

A colleague who has read your file already knows the basics; you correct the parts that are off rather than typing it in from scratch.

- **Cue:** Setup or a new task where the product already holds account data, history, or settings
- **Move:** Starts from data already in the product and says what it found
- **Response:** Confirms what was found, or corrects what’s off
- **Resolution:** Setup skips what the product already knows, and nothing gets typed twice
- **Needs:** First-party data.
- **Reference frame 07:** the orange square and the line "I pulled up your location." Under it, a small gray street map with the labels "N Mississippi Ave" and "N Fremont St". In the animation the line arrives word by word, the map rises in, an orange pin pops onto one spot, and a soft orange halo opens around it. The AI teammate shows what it found in a form the person can check at a glance, a map rather than an address string, instead of asking for an address. The on-screen line does not say where the location came from, so a build names its source, as UPG-R5 asks.
- **Sample lines:** "I found two goals in your Q3 plan. Which matters more?" / "Your account lists this address. Is it still right?"
- **Satisfies:** UPG-R2, UPG-R3, UPG-R5.
- **Also confirm:** every question the held data already answers is skipped, not asked anyway.

### 4. Mind if I look? (Ask before reading a connected tool)

Before checking someone’s calendar, a colleague asks first, and carries on with what they have if the answer is no.

- **Cue:** A connected tool holds context that would sharpen the work
- **Move:** Asks before reading a connected tool, naming which one and what it’s looking for
- **Response:** Allows it, limits it, or declines
- **Resolution:** Context arrives with its source shown, or the AI teammate carries on with what it has
- **Needs:** Third-party integrations, with permission.
- **Reference frame 08:** the orange square and the line "I want to check on that appointment in your Calendar." The word "Calendar" sits in a small chip with the calendar app's icon, so the source is named and recognizable. Under it, two buttons of equal size side by side, "Deny" on the left and "Allow" on the right. In the animation the line arrives word by word, then both buttons rise in, both light gray. A cursor moves to "Allow" and presses it, and "Allow" turns black. The request names the tool and the reason in one sentence, and saying no takes the same single click as saying yes.
- **Sample lines:** "Can I look at your Calendar to find that appointment? I’ll only read this week." / "No problem, I’ll work from what you’ve told me."
- **Satisfies:** UPG-R5, UPG-R9.
- **Also confirm:** declining does not block the task.

## Edge cases

- **Nothing on file.** Skip the playback and go straight to Example 2 (UPG-R2 has nothing to apply to).
- **The person skips the question.** Carry on with a stated assumption ("I’ll assume speed matters most; you can change that anytime") instead of blocking. Blocking breaks UPG-R9.
- **Stale data.** Say how old it is ("from your plan last March") and ask if it still holds. This is UPG-R5 applied to time.
- **The goal changes.** Let the person update it at any time, and confirm the new goal in one line (UPG-R8, UPG-R6).
- **Several goals that conflict.** Ask which comes first, and offer "Treat both equally" as an option (UPG-R4).
- **A declined permission.** Carry on with what you have, and do not ask for that tool again unless the person brings it up or a new task needs it (UPG-R2).

## Related patterns

These overlap with this one. Do not ask the same thing twice across them, and when you apply several humorphic patterns to one product, hold the combined first-run flow to UPG-R9 rather than each pattern on its own.

- [Ask before acting](https://humorphism.com/patterns/ask-before-acting.md): ask the one question that changes the work, before building.
- [Remember across sessions](https://humorphism.com/patterns/remember-across-sessions.md): how saved goals carry forward.
- [Connect information to outcomes](https://humorphism.com/patterns/connect-information-to-outcomes.md): reporting progress against the goal learned here.

## Checklist

Confirm each one before you call this built. Each item is the rule it comes from.

- [ ] UPG-R1: The first thing the AI teammate asks for names a result the person wants, not an attribute of the person.
- [ ] UPG-R2: No field the product can populate arrives empty.
- [ ] UPG-R3: The opening states what was found, then asks for the gap.
- [ ] UPG-R4: One option is tagged as suggested, nothing is selected on arrival, and a neutral or free-text escape exists ("Treat both equally", "Something else").
- [ ] UPG-R5: Every played-back fact carries its origin ("in your plan", "in your Calendar"), and its age when it may be stale.
- [ ] UPG-R6: After the person answers, one line confirms the answer was stored, naming it where space allows ("Got it. I saved your goal.").
- [ ] UPG-R7: Something later in the session visibly refers to the goal, or is ordered by it.
- [ ] UPG-R8: The saved goal is visible after the exchange, and can be changed or removed.
- [ ] UPG-R9: One question per turn, and the person reaches real work within a few exchanges.
- [ ] Capabilities the product lacks were skipped honestly, not stubbed or implied.
- [ ] It is built in the product's own components and voice.

## Images

Stills from the animations on the [pattern page](https://humorphism.com/patterns/understand-the-persons-goals), shipped alongside this file. The Reference screens and Examples sections describe each one in full, so this file works without them.

| File | Shows | Described in |
|---|---|---|
| [01-teammate-asks.png](https://humorphism.com/patterns/understand-the-persons-goals/01-teammate-asks.png) | Three goals found in the plan, one tagged "Suggested", none selected | AI teammate version |
| [02-teammate-person-picks-a-goal.png](https://humorphism.com/patterns/understand-the-persons-goals/02-teammate-person-picks-a-goal.png) | The cursor on the suggested goal, its row gray on hover | AI teammate version |
| [03-teammate-goal-saved.png](https://humorphism.com/patterns/understand-the-persons-goals/03-teammate-goal-saved.png) | Goal 1 checked, goals 2 and 3 dimmed, "Got it. I saved your goal." | AI teammate version |
| [04-tool-sign-up-form.png](https://humorphism.com/patterns/understand-the-persons-goals/04-tool-sign-up-form.png) | A blank "Tell us about yourself" form on flat gray | Tool version |
| [05-example-did-i-get-that-right.png](https://humorphism.com/patterns/understand-the-persons-goals/05-example-did-i-get-that-right.png) | Name and Email already filled, Address cut off by the frame, the Email field open for correction | Example 1 |
| [06-example-what-does-success-look-like.png](https://humorphism.com/patterns/understand-the-persons-goals/06-example-what-does-success-look-like.png) | The Faster to Cheaper slider with the live date and price it sets | Example 2 |
| [07-example-heres-what-i-already-know.png](https://humorphism.com/patterns/understand-the-persons-goals/07-example-heres-what-i-already-know.png) | A found location shown as a street map with an orange pin | Example 3 |
| [08-example-mind-if-i-look.png](https://humorphism.com/patterns/understand-the-persons-goals/08-example-mind-if-i-look.png) | A Calendar permission request, the source named in a chip, Deny and Allow the same size | Example 4 |
