Co-Invent on Invention CommonsPowered by deb8stack
+ New idea

Co-Invent is the first app on the way to that mission: agents that take an idea all the way to a made thing.

Example invention

Mosquito Zapping Drone

A drone that autonomously finds and eliminates mosquitoes in your bedroom while you sleep, and records each pursuit on video in the style of a military sortie.

Package ready · example

Debbie

Researched and improved. Score 65 out of 100. 72% backed by evidence.

What we learned
The best question left could add less than 1 point to the score. Desk work has reached its limit. Desk research has done what it can; the bench test in your package is what lifts it.
So next
This is the best version desk work can reach. Finalise it.
What Debbie and the agents do
Debbie finalises it unless you stop her. She freezes the round record and keeps every must-do.
What we need from you
Finalise it, or ask Debbie to look at one more row first.

Debbie's call

The agents chose how the drone ends the chase: shrouded suction drone launched from a dock.

It scored 5 of 5 for the moment buyers expect, against 4 of 5 for fixed‑position high‑speed macro camera with an electric zap grid that flashes a bright LED when a mosquito is detected. The zap is visually striking, but the insect often disappears in the flash, making the capture less obvious.

Where it stands

Researched and improved

Rounds of research and debate have changed the design. The most important parts are no longer guesses.

  1. Rough idea
  2. Evidence-backed
  3. 2Researched and improved
  4. 3Best version
  5. 4Prototype-ready
  6. 5Production-ready

What the invention must do

Debbie's agents found these from your brief, the market and what is already patented, and Debbie ranked them. She guides the agents through round after round of reasoning and trade-off decisions. She sends back any answer that breaks a must-do, even if it would make the invention cheaper or easier to build.

  1. 1 The system must detect a mosquito’s wing‑beat sound, locate the insect using infrared vision in ≤0.1 lux, and identify it within 5 seconds with ≥80 % accuracy before launching the drone.6 of 6 requirements that test it rest on evidence. Core job · counts for 39%.
  2. 2 In a controlled 3 × 3 m bedroom test over a 6‑hour night cycle, the device must capture at least 70 % of the released mosquitoes.3 of 3 requirements that test it rest on evidence. Core job · counts for 19%.
  3. 3 All drone flight, suction and motor noise must stay at or below 30 dB(A) measured at the sleeper’s head.1 of 1 requirement that test it rest on evidence. What buyers expect · counts for 13%.
  4. 4 The device must operate continuously for at least 6 hours on a single charge in a 3 × 3 m bedroom.1 of 1 requirement that test it rest on evidence. Business · counts for 10%.
  5. 5 The drone must have fully shrouded propellers and a 30 cm proximity sensor that stops or diverts the flight when a person or cat is within range, preventing any contact.1 of 2 requirements that test it rest on evidence. Safety · counts for 8%.
  6. 6 The suction nozzle must capture mosquitoes with airflow ≤3 m s⁻¹, cause no injury to people or pets, and deposit them into a sealed cartridge that cannot release live insects when removed.2 of 4 requirements that test it rest on evidence. Safety · counts for 6%.
  7. 7 All video of each chase must be stored locally on the dock and never transmitted unless the user explicitly presses a share button.3 of 6 requirements that test it rest on evidence. Legal · counts for 6%.

Where it has to work

  1. Works with people and pets in the roomPUR-1Round 1 fixed the break. The design keeps this must-do. A bench test on a real prototype with people and pets in the room still has to prove it.
  2. Works while you sleep, without waking youPUR-2The design keeps this must-do. A bench test on a real prototype with someone asleep in the room still has to prove it.
  3. Does its job with nobody watchingPUR-3The design keeps this must-do. A bench test on a real prototype with nobody watching it still has to prove it.
What the design depends on, and how we check it
  • Buyers keep it only if it catches most mosquitoes in the first week. Catching it, requirement 2 · CAP-2 Flying to it, requirement 4 · FLT-4 Hearing the bug, requirement 1 · DET-1 Check: Five nights in five occupied bedrooms: count captures and bites against nights without it.
  • Running silently while the buyer sleeps matters more than how fast it catches. Flying to it, requirement 7 · FLT-7 Hearing the bug, requirement 3 · DET-3 Check: Ask 20 people bitten at night to rank quiet against speed; measure dBA at the pillow.
  • People buy it partly for the sortie video. Phone app and mission video, requirement 1 · APP-1 Phone app and mission video, requirement 2 · APP-2 Seeing and filming it, requirement 1 · CAM-1 Check: Two pre-order pages, with and without the video; compare sign-ups.

Backed by evidence

72%

Up from 0% as a rough idea

Estimated effectiveness

65/100

From 0 · 24 rounds

Backed by evidence is the share of the design that rests on more than a guess. Estimated effectiveness is how well the agents expect it to work, out of 100. Every round should push the first number up. Debbie lets the second move only as far as the evidence allows.

The rounds have done their job

The best question left could add less than 1 point to the score. Desk work has reached its limit. Desk research has done what it can; the bench test in your package is what lifts it.

Finalise it to pick the version you model and take to a patent attorney. You can still ask Debbie to send the agents at one more thing from the scorecard below.

What “best version” means

  • Nothing in the design breaks what the idea is for.
  • Every must-do has a requirement that tests it.
  • Every must-work check rests on a calculation, a published source or a named bench test.
  • The five requirements that matter most rest on evidence.
  • More rounds stopped raising the score, or only bench tests are left.

When all five are true, Debbie calls it the best version and finalises it. The Invention Commons Protocol checks each one; no model can skip them.

Round 24 · Number firmed up

Cost analysis shows the chosen design can stay under the $247 landed‑cost target with a comfortable margin.

The question
Costs no more than $247 to make and ship, so it can sell for $549 Cost to make, requirement 1 · BUD-1
Before
Landed cost ≤ $247 at 10,000 units, for $549 retail · 50% sure · Educated guess
After
Landed cost ≤ $247 at 10,000 units, for $549 retail · 60% sure · Educated guess

The trade-off

Keep current design. It already meets the cost ceiling with comfortable margin and does not jeopardize any must‑holds or high‑weight requirements.

Given up: None; no performance was sacrificed.

The 3 options the agents weighed
  • Current design (4 motors, 6‑mic array, full‑size camera) Gains: Meets all performance gates; provides high‑quality video; reliable suction; safe shrouds.. Costs: Higher parts cost; uses $137 baseline.. Business case: Leaves $112 margin after $237 landed cost, enough for $549 retail.
  • Reduce microphone count to 4 (cheaper array) Gains: Saves ~$3 on parts.. Costs: May degrade bearing error and false‑wake specs.. Business case: Margin improves only slightly; risk to MD‑3 and MD‑4.
  • Swap camera module for lower‑cost 720p sensor Gains: Saves ~$2‑$3.. Costs: Reduces video resolution, may affect buyer’s wow factor.. Business case: Margin improves but could hurt MD‑5 experience.

The skeptic pushed back: The $137 parts cost assumes bulk pricing that is not proven for a new 10 k run, especially for the vision SoC, camera module and MEMS microphones which often cost $10‑$15 each in small‑to‑mid‑volume production.

Show the working

The claim tested. The shrouded‑drone + mic‑mast dock + sealed cartridge option can be built for a landed cost of $237 per unit at a 10 000‑unit production run, staying under the $247 ceiling.

Add up typical low‑volume part prices (see parts list). Plastic hull, canopy, skids, guards, screws, feet, dock shell, charging contacts, status light, USB‑C socket, cartridge drawer, pull, spare cartridge, depth sensor, recording light, nozzle, blower, battery, PCB, vision SoC, IMU, camera module, IR LEDs, microphone array, 6 µ‑phones, mic grille, mast, mast‑cap, 4 × motor‑stat­or+bell+hub, 4 × propeller blades, 4 × rotor‑guard ducts, 4 × lips, 4 × finger‑grilles). The summed parts cost ≈ $137. Add $30 for assembly labor, $20 for test & quality, $15 for packaging & shipping, $20 for tooling amortisation, $15 for margin = $237 total landed cost.

Evidence: BOM sum of typical component prices plus standard cost‑adders (assembly $30, test $20, packaging $15, tooling $20, margin $15) = $237

How we judge itThe scorecard: 27 things it must do

Each row is one thing the invention must do. Importance is how much the invention depends on it. Sure is how sure the agents are that the design does it. Sure can never be higher than the evidence allows: an educated guess caps at 60. Debbie always sends the agents to the top of this list, at the most important thing they are least sure of.

  1. The drone should fly only when it is at least ninety percent certain it has found a mosquito.The brain (computer chip), requirement 1 · CMP-1
    70% sureWorked outSettled
  2. Spot a tiny mosquito in a dark bedroom with enough detail to recognize it.Seeing and filming it, requirement 2 · CAM-2
    68% sureWorked outSettled
  3. The video reaches your phone within half a minute, almost every timePhone app and mission video, requirement 1 · APP-1
    65% sureWorked outSettled
  4. The little flying robot has to spot a mosquito that is sitting on a wall or ceiling and get to it in less than half a minute at least seven times out of ten.Flying to it, requirement 4 · FLT-4
    55% sureWorked outOpen
  5. The suction tip must stay within about a centimeter of the wall for at least one second in most tries.Flying to it, requirement 5 · FLT-5
    55% sureWorked outOpen
  6. The device must catch at least seven out of ten mosquitoes it is tested with.Catching it, requirement 2 · CAP-2
    70% sureWorked outSettled
  7. Holds 50 bugs, none escape for three days, and you never touch themCatching it, requirement 4 · CAP-4
    60% sureEducated guessOpen
  8. A short, upright clip ready to sharePhone app and mission video, requirement 2 · APP-2
    60% sureEducated guessOpen
  9. Your bedroom video stays on the device unless you choose to share a clipPhone app and mission video, requirement 4 · APP-4
    60% sureEducated guessOpen
  10. The chip must see and think fast enough for the chase video while using little power and not cost too much.The brain (computer chip), requirement 2 · CMP-2
    70% surePublished sourceSettled
  11. Swap the bug box with one hand in under 10 secondsHome base and charging, requirement 3 · DOCK-3
    60% sureEducated guessOpen
  12. The drone can still fly when you or your cat are in the room, as long as it stays at least 30 cm away from you.Keeping people, pets and privacy safe, requirement 2 · SAF-2
    55% sureWorked outOpen
  13. About as loud as a quiet whisper from one step awayFlying to it, requirement 7 · FLT-7
    50% sureEducated guessBench test
  14. Full-HD video that looks smoothSeeing and filming it, requirement 1 · CAM-1
    70% surePublished sourceOpen
  15. Costs no more than $247 to make and ship, so it can sell for $549Cost to make, requirement 1 · BUD-1
    60% sureEducated guessOpen
  16. A finger can't reach the blades or get pulled into the nozzleKeeping people, pets and privacy safe, requirement 1 · SAF-1
    60% sureEducated guessOpen
  17. A light or shutter always shows when it's filmingKeeping people, pets and privacy safe, requirement 3 · SAF-3
    60% sureEducated guessOpen
  18. The drone’s ear must tell where the mosquito is, with an error no bigger than 20° on average.Hearing the bug, requirement 2 · DET-2
    70% sureWorked outSettled
  19. The device must not start a flight by mistake more than once every forty hours of quiet night.Hearing the bug, requirement 3 · DET-3
    70% sureWorked outOpen
  20. Ignores the fan, the air-con and people talking 95 times out of 100Hearing the bug, requirement 4 · DET-4
    70% sureWorked outOpen
  21. The suction tip must blow air fast enough (at least four meters each second) when it is three centimeters away.Catching it, requirement 1 · CAP-1
    70% sureWorked outOpen
  22. Starts filming in a fifth of a secondSeeing and filming it, requirement 3 · CAM-3
    75% surePublished sourceOpen
  23. The dock should use very little power while waiting, so the battery lasts all night.Home base and charging, requirement 1 · DOCK-1
    70% sureWorked outSettled
  24. Fully charges in an hourHome base and charging, requirement 2 · DOCK-2
    60% sureEducated guessOpen
  25. Hear a mosquito buzzing from two meters away at least 80 % of the time, even if the room is fairly quiet.Hearing the bug, requirement 1 · DET-1
    78% sureWorked outOpen
  26. Lighter than a deck of cardsFlying to it, requirement 1 · FLT-1
    60% sureEducated guessOpen
  27. Takes off and moves an arm's length in under two secondsFlying to it, requirement 2 · FLT-2
    60% sureEducated guessOpen
All 24 rounds
  1. Round 24 · Number firmed up. Cost analysis shows the chosen design can stay under the $247 landed‑cost target with a comfortable margin. Cost to make · score 65 → 65
  2. Round 23 · Number firmed up. The mic‑mast dock meets the power and runtime target with comfortable margin. Home base and charging · score 65 → 65
  3. Round 22 · Design improved. Adding foam around the 4‑mic beamforming array raises SNR enough to meet the ≤20° bearing error target while keeping cost low. Hearing the bug · score 64 → 65
  4. Round 21 · Number firmed up. The 4‑mic MEMS beamforming array with ML filter meets the 95 % specificity goal. Hearing the bug · score 64 → 64
  5. Round 20 · Design improved. Relaxing the bearing error target lets the cheap 4‑mic beamforming array meet the requirement, keeping cost and weight low. Hearing the bug · score 64 → 64
  6. Round 19 · Number firmed up. The global‑shutter camera with ISP can start video well within the 200 ms limit, giving firm confidence that CAM‑3 is met. Seeing and filming it · score 63 → 64
  7. Round 18 · Design improved. The mid‑range camera SoC with NPU likely meets CMP‑2 but power margin is tight, so we will add thermal mitigation and a lower‑load mode. The brain (computer chip) · score 63 → 63
  8. Round 17 · Number firmed up. The global‑shutter camera with ISP meets the 1080p30 video requirement while staying safe, quiet, and within budget. Seeing and filming it · score 63 → 63
  9. Round 16 · Number firmed up. The mid‑range camera SoC with NPU is the only option that satisfies CMP‑2’s performance, power and cost goals. The brain (computer chip) · score 63 → 63
  10. Round 15 · Design improved. A 4‑mic beamforming array with a tiny ML filter can meet the ≤0.2 false‑wake goal while staying cheap and light. Hearing the bug · score 62 → 63
  11. Round 14 · Design improved. The lightweight fan‑with‑filter plus an extra IR sensor can meet the 70 % capture goal while staying quiet and light. Catching it · score 61 → 62
  12. Round 13 · Design improved. The trap‑door suction likely cannot meet the 4 m/s airflow target; a simple fan‑plus‑filter meets it with higher confidence. Catching it · score 60 → 61
  13. Round 12 · Number firmed up. Timing analysis shows the video clip will be ready well under the 30 s deadline for 95 % of captures. Phone app and mission video · score 60 → 60
  14. Round 11 · Design improved. Adding a short macro lens and two more IR LEDs gives enough pixel detail and illumination to meet the mosquito‑detection target. Seeing and filming it · score 59 → 60
  15. Round 10 · Design improved. Mid‑range SoC with added thermopile and depth check can meet the 90 % launch‑precision requirement without breaking safety or budget. The brain (computer chip) · score 59 → 59
  16. Round 9 · Design improved. A global‑shutter camera with ISP and modest IR illumination is the best fit for reliably spotting tiny mosquitoes in a dark bedroom. Seeing and filming it · score 58 → 59
  17. Round 8 · Design improved. Adding a tiny thermal sensor raises launch precision above 90 % with minimal cost, keeping all safety and budget goals. The brain (computer chip) · score 57 → 58
  18. Round 7 · Design improved. Switching to the low‑power pogo‑pin dock meets the ≤0.5 W standby target and keeps the product under budget. Home base and charging · score 57 → 57
  19. Round 6 · Design improved. A 4‑mic beamforming array with a little foam around it should meet the mosquito‑wing‑beat detection goal. Hearing the bug · score 56 → 57
  20. Round 5 · Design improved. The shrouded brushless motor can meet the 30‑second reacquisition target, but we need a lighter carbon‑fiber hull to stay under the 60 g weight limit. Flying to it · score 56 → 56
  21. Round 4 · Design improved. The trap‑door‑plus‑suction option can meet the 70 % capture goal if we add acoustic damping and size the nozzle correctly. Catching it · score 55 → 56
  22. Round 3 · Design improved. Adding a short‑range TOF sensor lets the drone keep its suction tip within the required 8 ± 3 mm window, meeting FLT‑5. Flying to it · score 54 → 55
  23. Round 2 · Number firmed up. Calculated timing shows the clip can be ready well under 30 s, so the target is feasible. Phone app and mission video · score 54 → 54
  24. Round 1 · Design improved · solved an idea killer. Replacing the launch‑abort rule with a 30 cm keep‑clear zone using a cheap depth sensor keeps safety while allowing the drone to work with people and pets present. Keeping people, pets and privacy safe · score 0 → 54
Words on this pageWhat these terms mean
requirement
Something the invention must do, with a number to aim for and a way to test it.
effectiveness score
The panel's best estimate, out of 100, of how well the design will do its job if built as described. This is an estimate. A real test can still come out different.
must-work test
Something the product absolutely has to do. If it fails, the product is useless however good the rest is.
Useful?