Instinct: No new app. Just text it.
Instinct is a personal assistant you text or call, with no app to learn. Explore when meeting people inside the tools they already use beats a new interface, and what you must earn before an agent acts for them.
See the idea in one picture ↓- new apps to install
- 0
- the only way in
- Text or call
- public launch, invite-only
- Aug 2026
- reported valuation, Sept 2026
- $10B
Funding and valuation as reported by the press.
Independent educational analysis. Scenarios, people, numbers and illustrations on this page are original PMcademy practice material, not company research results or an endorsement.
Think about the person with no spare minutes.
I will not learn another app. I would send one more text.
What would help?
Errands done without set-up. And a say before money is spent.
What gets in the way?
One more app to learn. No time to check every step.
Fictional personRead the analysis
Our fictional learner, Rhea, is a nurse on night shifts. She already texts her family, her manager and her pharmacy. She will not learn a new dashboard to order groceries, but she would send one more message. Ask what your person already does every day without thinking, and whether your product can arrive inside that habit instead of beside it.
Every new product asks for one more app.
Help that lives in one more app
Help that arrives in a text you already send
Read the analysis
Imagine your team has built something genuinely helpful. The usual next step is an app: an icon, a sign-up, a home screen, a tour. Each of those is a small cost the person pays before getting any value. For a busy person, the cost can be the whole reason they never start. The product question is where the help should live, and what the person has to learn before it arrives.
An assistant with no new screen
- 1You text it
You send a message the way you would to a friend.
- 2It does the work
It completes the task with its own phone and computer, across apps that already exist.
- 3It reports back
The result arrives as a message in the same thread, with nothing to install.
Animated illustration. Names, numbers and people are fictional teaching material.
Move the work, not the person.
Illustrative data and simplified behaviour. This model does not represent a live company system or measured company results.
Read the analysis
Use the model below to compare three ways to deliver the same help. A new app asks people to come to you. A text or a call goes to where they already are. A text that checks in before anything costs money adds a moment of control. Watch how many of ten everyday tasks get finished, and how many need the person to step in. The numbers in the model are illustrative.
Funding is not the same as finished tasks.
Instinct says it is reached by text or call and acts across your apps and devices; press reported $1.25B raised in two rounds.
How often tasks finish correctly, or how mistakes are caught and undone.
- Tasks finished correctly
- Approvals asked before spending
- People who come back a month later
Read the analysis
The figures around Instinct describe investor confidence and early interest, not how often tasks are completed correctly. The privacy discussion at launch shows that people judge an agent by what it can see as much as by what it can do. In your project, decide what you will count as a finished task, how you will catch a wrong one, and what you will tell people about their data.
No interface also means nothing to look at.
- Nothing to install or learn
- Works inside existing habits
- Reaches people apps miss
- Nothing to inspect
- Depends on apps you do not control
- Broad access raises privacy stakes
Read the analysis
Without screens, there is nothing to learn, but also nothing to inspect. People cannot see what the assistant is doing, what it can reach, or why it chose one option. Acting inside other people’s apps also means depending on screens you do not control. Decide how the person will see progress, approve risky steps and undo mistakes, before deciding how little interface you can get away with.
New app, existing channel or a call?
Your fictional team is building a helper that books doctor appointments for older adults in a small town. You can ship a new app with large buttons, let people send a text message to the helper, or offer a phone number they can call and talk to. Choose a channel and explain how the person approves a booking and fixes a mistake.
Choose a next move
Leave with a channel decision.
Your artifact
Write a channel decision for one job: the channels your users already use, where the help will arrive, the approval and undo moment, and the access you need and will disclose.
Critique my project artifact using only the evidence I provide. Write a channel decision for one job: the channels your users already use, where the help will arrive, the approval and undo moment, and the access you need and will disclose. Separate observed facts, assumptions, proposed tests and missing evidence. Identify a failure mode and a cheap check. Do not invent company results or user research. Ask for my artifact before starting.
Read the analysis
Pick one job your product does. List the channels your users already use every day. Choose where the help should arrive, and write the one moment where the person must approve or can undo. Name what the product will need access to, what you will tell people about it, and the signal that would make you add a screen after all.