Impulse — Master Product & Marketing Overview¶

A single reference describing what Impulse is, what every part of it does, how the pieces fit together, and how to talk about it. This is the high-level companion to the detailed firmware specification; where the spec tells an engineer how to build it, this document explains what it is and why it matters.

Revised 2026-09-14. Supersedes the 2026-08-08 revision. See the revision note at the end for what changed and why.

1. What Impulse Is¶

Impulse is a habit-enforcement system for people who already know what they want to do and can't reliably make themselves do it.

The core idea is self-binding: you decide, in a calm and clear-headed moment, what kind of day you want — when to wake, when to move, when to put the phone down — and the system holds you to those decisions later, when a tireder, more impulsive version of you would rather renegotiate. Impulse is not a coach, a tracker, or a source of motivation. It is the mechanism that makes your own earlier decisions stick.

The system has three parts: a watch you wear, anchors you place around your home, and a mobile app where you design everything. They work together to enforce commitments tied to time, location, and proximity.

2. The Three Components¶

2.1 The Watch¶

The watch is the wearable core of the system. It is the device that knows the schedule, tracks where you are relative to your anchors and WiFi networks, and delivers the enforcement when a commitment isn't being met.

What it does:

Holds your daily schedule and wakes itself at the start and end of each commitment window, sleeping in between to conserve battery.
During a commitment window, checks whether you're meeting the criteria — near or away from a specific room, on or off a specific WiFi network — and escalates through vibration and sound if you're not.
Knows when it's being worn, using a reflective infrared skin sensor on the underside (it pulses an invisible IR light and reads the reflection off your skin, with ambient light subtracted out). Taking the watch off during an active commitment doesn't end enforcement; it hands enforcement over to the anchors instead.
It is the root of trust, not the app. The watch itself refuses schedule changes that would loosen an active or imminent commitment (see the day-delay principle in 2.3), refuses clock or timezone changes that would end a window you're currently inside, and keeps its internal timers on elapsed time rather than wall-clock time — so winding the clock forward ages nothing. This is implemented in firmware, not policy.
Shows an always-on 12-LED analog clock face; the ring doubles as a status indicator (what the watch is doing and why it last woke).
Charges overnight and powers the system's signature wake-up experience (see Sunrise Lock, Section 3).
Runs at least 8 hours on a charge — more, depending on how many enforcement windows the day carries — and refills in 2–3 hours, making overnight charging the natural rhythm.

What to call it in plain language: the watch keeps you to the line you drew. Avoid describing it as "enforcing rules" — it enforces your own earlier decisions.

Hardware-revision note (internal): on the current board revision the vibration motor and the LED ring share a pin, so a given unit runs one or the other. The prototype in hand has a working motor; the next board revision moves the ring to its own pin and supports both. Any demo footage showing vibration should be shot on a motor-enabled unit, and no shipped-product claim should depend on both features until the new revision is validated. This revision is also a gate on tooling — see Section 9.

2.2 The Anchors¶

Anchors are small devices you place at the locations tied to your habits: the nightstand, the desk, the front door. They are what lets Impulse work in the physical world rather than just on a phone screen.

What they do:

Broadcast a consistent identity so the watch can tell which room it's in and how close it is.
Sound an alarm when the watch is removed during a commitment window, in the rooms that matter — so walking away from the watch is the loud choice, not the easy one.
Lock themselves in place during an active window: a small servo clamps the anchor's own strap so the anchor can't be quietly moved, flipped over, or carried out of the room on impulse. This locks the anchor down, not the watch — never claim it restrains the watch or the wearer.
Run from wall power with an internal backup battery that currently bridges about an hour if unplugged (target: 2–3 hours after power optimization). Pulling the plug mid-commitment doesn't buy quiet — but do not claim the battery is indefinite. The honest consumer framing is "unplugging it just gets you a battery-powered version of the same noise," with the runtime figure stated plainly on any spec sheet.
Hold each event schedule locally and recompute it at local midnight on their own, so an anchor keeps working even if the phone and watch are unreachable. If an anchor has no valid time or schedule, it stays silent (fail-open) rather than alarming wrongly.
Serve as a docking point for the phone during phone-distance commitments (see Section 3). The anchor directly senses whether the phone is docked on it.
Self-heal their network connection: an anchor that loses WiFi reports its distress over Bluetooth, and a charging watch will quietly re-share known credentials overnight with anchors in its own schedule. (Credential sharing is restricted to the watch's own scheduled anchors, so a stranger's device can't phish the WiFi password.)

What to call them in plain language: the part of Impulse that makes your commitments physical, and the reason you can't quietly opt out. Be honest that they are hard to bypass on impulse — not "impossible to cheat." Overclaiming invites people to disprove you.

2.3 The Mobile App¶

The app is the calm side of the system — the place where clear-headed you sets everything up and then hands the day off to the hardware.

What it does:

A visual weekly schedule builder where you lay out commitments: a time window, a place or network, and a firmness level.
Per-commitment firmness, from a gentle nudge you can shrug off to a wake-up you can't.
Anchor naming and placement guidance, including identifying which physical anchor is which (tap an anchor in the app and it beeps).
The ability to disable a window in advance, before it starts, so a sick day or an early wake-up is never a fight.
An emergency-pass system for the days you can't plan for, on a frequency you set. Passes are tracked by the watch itself: spending one grants a one-day exception immediately, and raising your own allowance is treated as a loosening (day-delayed), while lowering it takes effect immediately.
Live status: which anchors are online, where the watch is, whether it's being worn, whether the current commitment is being met, battery level.

A critical design principle runs through the app: changes that loosen enforcement take roughly a day to take effect. You can always make the system harder on yourself instantly, but you cannot grant yourself an escape in the heat of the moment. That delay is not a limitation — it is the feature. And critically, this rule is enforced by the watch, not by the app: the watch inspects every incoming schedule change, applies tightenings immediately, and quarantines loosenings for the delay period. Deleting or bypassing the app doesn't bypass the delay.

What to call it in plain language: where clear-headed you runs the show.

3. The Signature Experiences¶

These are the two features that best define the product and should anchor most marketing.

3.1 Sunrise Lock¶

The flagship demo, and the centrepiece of all top-of-funnel marketing. The watch charges overnight on your nightstand, next to its anchor. At your wake-up time, the anchor begins to sound while the watch sits charged and ready. Putting the watch on quiets the room for a short grace period you set yourself (a per-commitment "donning grace," up to 30 minutes, though a few minutes is typical). If you're still in bed when the grace period ends, the watch and the anchor sound together, escalating, until you're up and out of the room.

Why it matters: it is the alarm you can't snooze, because the alarm isn't on your nightstand — it's the whole room. It directly solves the single most relatable problem in the category with a mechanism that obviously works, and every parameter is something you set the night before. Past-you runs the morning.

Why it leads marketing: it is instantly legible without explanation, it demos in under twenty seconds on video, and it requires no comparison to any competitor to make its point. Shoot it properly once and cut it many ways.

3.2 Phone-Distance Commitments¶

This is how Impulse addresses screen time, and it is deliberately not a copy of app-blockers.

Instead of blocking individual apps, Impulse enforces physical distance from the phone. You set windows in the app during which your phone must be docked near a specific anchor (or, secondarily, far from you). If you spend too long near your phone during one of these windows — past a configurable tolerance (60 seconds by default), so a quick check is fine — the watch and anchors sound until you put it back down. A brief undock to glance at a text is tolerated exactly like a brief approach; the timer resets when you put it back.

This is the one feature in the system that fails closed, and that is deliberate. Every other Impulse feature treats a degraded link as a reason to stay quiet. Phone-distance cannot, because the thing being enforced is the link. If a broken connection meant silence, the bypass would be walking away — or deleting the app — and the feature would enforce nothing.

So phone-distance is presented to the user as an explicit agreement. When you enable it, you are agreeing that:

the app must remain installed,
the app must remain open, and
the Bluetooth link between the docking anchor and the phone must stay strong enough to show the phone is near the anchor.

If any of those three stops being true during an active window, that is treated as non-compliance and enforcement sounds.

Capture this consent explicitly. The three conditions must be shown as a distinct in-app acknowledgment at feature-enable time, with a timestamped record. This is the difference between a documented user agreement and a surprise alarm, and it is the single most important support and refund mitigation in the product.

The framing that makes this a differentiator: Impulse does not detect uninstalls, and no material should claim it does. What it does is require continuous proof of compliance rather than trying to prevent non-compliance. Blockers put an obstacle in your path and you defeat the obstacle; Impulse asks you to keep demonstrating you're doing what you said, and the absence of that demonstration is itself the answer. That distinction is true, architectural, and doesn't depend on anyone else's implementation.

Implementation status: the anchor-docked variant is implemented and verified end-to-end on hardware (August 2026), including tolerance behaviour and protection against false alarms from transient radio glitches. The phone-far-from-watch variant remains inherently noisier and should stay a secondary, expectation-managed option.

Known cost of failing closed: false alarms are now a product category rather than an edge case — low power mode, iOS backgrounding the app under memory pressure, congested 2.4GHz environments. Mitigations: require several consecutive bad readings before believing the phone has moved, steer setup guidance hard toward anchor-docking, and instrument link-quality telemetry so that a "it alarmed for no reason" report can be diagnosed rather than argued about.

Honesty note: Impulse does not block apps. Earlier concepts explored it; it did not survive contact with what the platforms allow. Do not market app-blocking, and do not leave stale app-blocking claims in any comparison material.

4. How It All Fits Together¶

A single day shows the system working as a whole:

The night before, you charge the watch on your nightstand. At 6am, Sunrise Lock gets you up and out of the bedroom. From 6:30 to 9, a phone-distance window keeps your phone docked at the kitchen anchor so your morning isn't lost to scrolling. At 9, a commitment to be at your desk begins, verified by the office anchor. At 5:30, a window enforces that you've left home and arrived on your gym's WiFi. Each of these is something you decided once, in the app, and then stopped having to think about.

The throughline: you design the shape of the day, and then you walk into the shape you built.

5. Positioning & Competitive Frame¶

Impulse sits near three existing categories but doesn't belong to any of them:

Habit apps track what you do and trust you to keep going. The trust is exactly where they fail — nothing happens when you ignore them.
Phone blockers block apps but live on the phone they're policing.
Shock-based wearables deliver aversive feedback to break habits, but don't follow you into the physical world or tie to real locations.

Impulse's distinct ground: it meets you in the physical world, at the moment of the decision, tied to real places and real proximity. The sharpest true differentiator is that the enforcer isn't the thing being bypassed — take off the watch and the anchors take over; try to edit the schedule and the watch itself, not the app, decides whether the change is a tightening (instant) or a loosening (delayed a day). Time and location commitments run entirely on watch and anchors, which hold their own schedules locally. Lose the phone, break the app, change the router: Sunrise Lock still fires at 6am.

5.1 What is actually true about iOS blockers¶

Documented, and worth stating precisely because precision is what makes the claim survive scrutiny:

On iOS, third-party blockers work through Apple's Screen Time APIs (FamilyControls, ManagedSettings, DeviceActivity). They cannot see which apps you launch, cannot run continuously in the background, and cannot draw over another app — the OS does the blocking on their behalf.
A blocker can prevent its own deletion and add friction (delays, math problems, strict modes).
A blocker cannot prevent the user from revoking Family Controls authorisation in Settings, which voids its tokens and drops every shield at once. This holds even when Screen Time itself is passcode-locked.
The only genuinely hard case is when a different person holds the Screen Time passcode — which is parental control, not self-restriction.

The correct mechanism to name is "revoke access in Settings," never "uninstall the app." Uninstalling is the one path a well-built blocker may actually be able to stop. Getting this wrong in public turns a true claim into something a single informed reviewer can dismiss.

5.2 How to use this — and how not to¶

Do not build advertising around demonstrating a competitor being defeated, named or "obviously implied." Reasons, in order of severity:

It invites the reciprocal video and we lose that exchange. Impulse is beatable — muffle an anchor, unplug it and wait out the backup battery, leave the watch in a room with no coverage. Publishing an attack ad commissions the "I beat Impulse in 45 seconds" response, and it will outperform anything we make.
It is comparative advertising regardless of naming. The standard is whether consumers understand the reference. An unnamed-but-obvious competitor is the definition of actionable, not a shield — and a true-but-incomplete claim (omitting that blockers do resist deletion) is exactly what turns defensible into misleading.
It sells doubt about an adjacent category. Most viewers will absorb "these things don't work," not the finer distinction.
It violates our own voice rules — mechanism-worship and adversarial framing.
The precedent is bad. Controversy-forward positioning in this category has historically produced abundant awareness and poor conversion.

The version that works keeps the footage and changes the subject. Don't film a competitor being defeated; film a person defeating their own blocker at 1am. Their thumb, the Settings toggle they already know how to find, their face afterwards. No brand, no comparison, no claim about anyone else.

You know where the toggle is. You found it at 1am, three weeks ago. It took eleven seconds.

That is problem-statement advertising, not comparative advertising: legally inert, emotionally stronger, and it lets the viewer reach the conclusion themselves.

Comparison belongs on our own site, not in ads — a page someone reads when already considering us, where every row is accurate and checkable. The honest Impulse column includes: real-world location enforcement, phone-proximity enforcement, removal detection, anchor hand-off, anchor physical lockdown (of the anchor itself), and watch-held schedule integrity — and excludes app-blocking and uninstall detection.

6. Audience¶

The gap between intention and action is the whole market: people fixing their sleep, people trying to get to the gym, students, remote workers, parents trying to be present, creatives trying to show up to the work. You don't need a diagnosis to be tired of breaking your own promises.

Adults with ADHD are disproportionately likely to buy, and they should feel spoken to. But there is a hard line here, and it is a regulatory one rather than a matter of taste:

Market to the audience. Never market about the condition.

Speak in the language of the experience — the exhausting, all-day project of managing yourself into doing what you already wanted to do — without ever naming a diagnosis or implying an effect on one. "Helps you get out of bed when you said you would" sits comfortably inside FDA's general-wellness policy for low-risk products. "Helps manage ADHD symptoms" or "improves executive function" points at a condition and risks the product being treated as a medical device, which is a regulatory pathway this company cannot afford.

No medical-efficacy claims of any kind, anywhere, including in founder social posts and interview answers. In any material that might reach people in recovery or in crisis, tread gently; those audiences are best reached through professional partnership, not direct marketing.

7. Voice & Language Guide¶

The single most important rule: sell the freedom on the other side of the friction, not the friction itself. The buzzing and beeping are the mechanism, not the brand. The brand is the calm, the headspace, and the time you get back when you stop negotiating with yourself.

Words and frames to use¶

Self-binding language. The system honours your own earlier decision. "You decided this, this morning." "Past-you runs the day." This frame turns every "isn't this harsh?" reaction into "oh, I'm doing this to myself, on purpose."
Freedom and relief. "Stop negotiating with yourself." "Get your mornings back."
Agency. The user is the expert on their own life. They design the day; the hardware carries it out.
Plain, concrete specifics. "Up and out of the bedroom by 6:30." "Your gym's WiFi confirms you made it."
Honest confidence. "Hard to beat on impulse," not "impossible to cheat."

Words and frames to avoid¶

The word "enforcement" in consumer-facing copy. Internally it's fine; to a buyer it sounds like punishment. Use accountability, follow-through, commitment, holds the line — and even those, sparingly.
Parole-officer language. Flip to the user's own authorship wherever possible.
Mechanism-worship. Don't make relentlessness the personality.
Coercive flourishes. Lines like "it doesn't care that it's cold outside" are memorable and sell exactly the wrong thing.
Attacking competitors. See 5.2.
Overclaiming. No "impossible to cheat," no medical-efficacy claims, no app-blocking, no uninstall detection, no claiming the anchor locks the watch, no claiming indefinite anchor battery.
AI cadence. Watch for the relentless triple-beat rhythm (short fragment, elaboration, punchy reversal) repeated paragraph after paragraph. Read copy aloud — if you hear the same drumbeat twice in a row, rewrite one.

Taglines in play¶

Stop negotiating with yourself. (primary)
Keep the promises you make to yourself.
Outsource your willpower.

8. Pre-Launch Marketing Principles¶

The goal of the pre-Kickstarter period is not awareness. It is a warm list of people who will back on day one. Every decision is measured against that.

Sunrise Lock is the entire top of funnel. One properly shot hero video, cut many ways.
Two channels, not six. Short-form video and a genuine community presence. No paid spend until organic creative has told us what converts.
Build in public. Board revisions, failed prints, the pin-share problem. This audience rewards visible iteration, and it buys forgiveness if fulfilment slips.
Publish our own weaknesses. A "here's how to beat Impulse" video — the muffled anchor, the unplugged anchor, the uncovered room — then explain why people don't do these things: the friction isn't physical, it's that deliberately defeating a system you built for yourself requires admitting you're doing it. This is unbeatable on trust, makes any competitor's disproof video redundant, and is the purest expression of "hard to beat on impulse, not impossible."
Seed real units early. Twenty to forty beta systems with real users and creators. UGC from a trusted voice outperforms any paid creative for a considered purchase.
Validate intent with money, not emails. Convert part of the waitlist to a small refundable reservation before sizing production. Email signups overstate real intent substantially.

9. Manufacturing & Hardware Status¶

Enclosures. Printed in ABS/ASA, acetone vapour-smoothed, then batch-tumbled. Both finishing steps are batch-parallel and largely unattended, so finishing is not a per-unit labour cost — the remaining per-system labour is assembly, firmware flash, QC and packing, roughly 25 minutes. The result reads as better than raw FDM without claiming to be injection moulded; never describe printed units as injection moulded.

Capacity. Print time binds before labour does. Roughly 360 systems/month per two printers, scaling linearly; at about six printers labour binds again and needs a part-time assembler. Capacity is therefore a capital question (printers are cheap), not a structural ceiling.

Workspace safety. Acetone vapour at production scale is a flammable-vapour hazard, and ABS printing requires real ventilation. This belongs in the insurance conversation and possibly the lease, and it is a live risk to the business independent of any product risk.

Radio certification. Pre-certified FCC modules for both devices. Two conditions must hold: the module's integration requirements (antenna type, trace layout, keep-outs) must be followed exactly or modular approval is void, and product labelling must carry "Contains FCC ID: [module ID]." Part 15B verification is still required on each finished product, as is UN38.3 on the cells for shipping.

Injection moulding — decision gates. Tool only when all four hold: (1) enclosure geometry unchanged across two consecutive production batches and the motor/LED board revision validated in shipped units; (2) 1,500+ systems shipped to real customers, not merely backed; (3) tooling payback under six months at current run rate; (4) tooling ≤25% of cash on hand with four months' runway remaining after. Accelerate if sustained run rate exceeds ~600/month, cosmetic RMA exceeds 5%, a wholesale opportunity appears, or print-attributable COGS exceeds ~$15/system.

Use aluminium soft tooling first (10k–50k shots, 3–5 week lead, cheap to modify); steel only past 30–50k units. Tool the anchor before the watch — as the attach SKU it ships at higher volume, so it pays back sooner and de-risks the moulder relationship on the simpler part.

Do not tool before fulfilling the campaign. Campaign overfunding measures demand for a promise, not maturity of a design. The sequence is: fund → ship printed → learn from 500+ users → freeze → tool for the retail phase.

10. Safety, Liability & Design Constraints¶

Not legal advice; retain counsel before the campaign. Ranked by what actually generates claims.

Lithium cells. Two per system, and by a wide margin the largest real risk. Protected cells with integrated protection circuitry from a documented supplier. UN38.3 regardless. Do not value-engineer this line.
Acoustic output. A device designed to be loud, on a nightstand, near a sleeping person's head, that escalates. Measure and publish maximum SPL at 30cm and 1m, and cap it. "It's supposed to be loud" is not a defence if the level was never characterised.
Failure to warn. Highest-probability exposure, cheapest to fix. Explicit and prominent: not an alarm clock of record, not a medical device, not for safety-critical timing.
Mechanical. Design out servo pinch points (children, pets). Consider strap and IR-window skin contact.
Single-user pairing is a safety constraint, not a feature decision. A device that detects removal, alarms in rooms, and physically locks hardware becomes a coercive-control tool if configured by one person for another. One watch pairs to one account, and only the wearer may tighten their own commitments. If accountability-partner features are ever added, every tightening requires the wearer's on-device confirmation. Build this constraint in now, while it costs nothing.
Corporate and contractual. Entity formed before any money is taken — never run a campaign as an individual. General liability with products coverage in force before launch. Counsel-drafted ToS, EULA, limitation of liability, warranty terms, arbitration clause. Age-gate to 18+.
Crowdfunding delivery. Non-delivery has drawn regulatory action, not just angry backers. Under-promise the timeline substantially.

11. Honest Caveats to Keep Front of Mind¶

Phone-distance reliability. Fails closed by design, so false alarms are a real and expected failure mode. The three-condition user agreement must be explicit and logged; setup guidance must steer users to anchor-docking and state the phone conditions plainly.
The anchor-lock and muffling. The servo locks the anchor's own strap against being moved on impulse. It does not restrain the watch or the user, and a determined person can defeat an anchor. Hard-to-beat-on-impulse, never impossible.
Anchor backup battery runtime. About one hour today, 2–3 hour target. Never imply indefinite.
Watch battery. Does not last multiple days; the overnight-charging rhythm must be communicated clearly.
Vibration motor vs LED ring. Shared pin on the current board; the next revision supports both. Don't ship marketing depending on both until validated.
Capabilities that do not exist. No app-blocking. No uninstall detection. Keep all materials consistent.

Revision Note — 2026-09-14¶

Phone-distance corrected to fail-closed (§3.2). The prior document described it as failing open, which was wrong and would have made the feature unenforceable. The three-condition user agreement is now documented, along with the requirement to capture consent explicitly and the accepted false-alarm cost.
Differentiator reframed from uninstall detection (which doesn't exist) to continuous proof of compliance — architectural, true, and independent of competitors' implementations.
New §5.1 documenting what is actually true about iOS blockers, including the correction that the exit path is revoking Family Controls authorisation in Settings, not uninstalling.
New §5.2 ruling out competitor-attack advertising, with reasoning, and specifying the problem-statement alternative.
Audience rewritten (§6): the constraint is on claims, not on who we speak to. "Market to the audience, never about the condition," with the general-wellness rationale.
New §8, pre-launch marketing principles, including publishing our own weaknesses and validating intent with reservations rather than emails.
New §9, manufacturing status: batch finishing (acetone + tumble) corrects the earlier per-unit-labour assumption, pre-certified FCC modules, capacity figures, and injection-moulding decision gates.
New §10, safety and liability, including single-user pairing as a hard design constraint.

This document is a living overview and should be updated as the product evolves. For implementation-level detail, refer to the firmware specification.