A companion with character that helps you keep what matters and return to what you want to do.
A physical Buddy, an app and a personal AI service. Connecting conversation, context and useful action, with the person in control.
Investment summary
The video illustrates planned scenarios. It does not demonstrate a working physical device.
Verification applies to a specific result and its date. These statuses do not mean the entire product is ready.
The physical format is a hypothesis for a more natural presence: a visible character on your desk, a conversation without searching for an app, and clear status signals.
A character and a face, a camera in the badge, two mechanical paws, Wi-Fi and a mandatory charging base are the current desktop Buddy requirements. Components and a working implementation have not yet been selected.
Concept / hardware not selectedSetup, permissions, reviewing results and managing memory. In the phone-bridge option, the phone also connects Buddy to the cloud.
App and components existDialogue, personal context and tools. The large model and main memory are in the cloud; offline operation needs separate validation.
Software foundation existsWeb / Telegram, a profile and account-scoped memory, voice conversation components, draft → confirmation → action, and financial calculations with Excel. A specific automatic audio-routing workflow was accepted on iPhone on 29 September.
Board and firmware, pairing, microphones and speaker on the sample, secure updates, offline recording, and the complete device → phone → cloud chain. A browser test does not verify this hardware workflow.
Microphone access and AI processing require consent. An explicit start, visible indicator and a way to stop recording are planned for the device. Continuous hidden listening is not a promised feature.
A camera in the badge is mandatory in the current desktop Buddy concept dated 10 October; its module has not been selected. Setup permission, an explicit capture mode, an indicator and a stop control are requirements. “Record me” starts an authorised recording; publishing requires a separate command.
The software foundation links memory to an account and confirmed notes. Complete multi-turn memory on a physical Buddy still needs acceptance testing. Connecting other accounts and importing a personal archive require separate permissions.
The user must be able to revoke access, correct and delete memory. Before the hardware pilot, source recordings, transcripts, local buffers, the cloud and the search index need to be checked together. Retention periods and backup behaviour still need approval.
Current security page ↗We start with adults and specific tasks. Each story needs validation with people.
The conversation is over. What did we decide, and who takes the next step?
Buddy: From a recording made with permission, proposes a summary with sources and draft tasks. The person reviews and confirms.
For the person: One clear agreement instead of searching through messages.
In the pilot: summary accuracy, corrections and returning to tasks.
People at the table speak different languages.
Buddy: Translates in both directions in the selected mode. Shows when it is listening and lets the person stop the conversation.
For the person: Less switching between people and a screen.
In the pilot: latency, meaning errors and performance in noise.
The day dissolves into incoming messages and unfinished tasks.
Buddy: Helps choose one step and return to a saved goal. Reminders stay within the permissions given.
For the person: The person sees the next step and marks it complete.
In the pilot: step completion and voluntary repeat use.
After a busy day, it is time to switch to rest.
Buddy: Offers a chosen sound and a timer at an agreed volume.
For the person: A clear evening ritual controlled by the person.
In the pilot: returning to the ritual. No medical claims.
AI dialogue, emotional devices and recording tools are evolving alongside each other. Buddy’s hypothesis is to connect character, context and action in one experience.
Buddy’s hypothesis: a familiar desktop character that connects conversation, controlled personal memory and useful actions. A phone or app is also a substitute; the pilot must show why a physical companion earns a place on the desk.
Six references selected from 17 profiles: desktop companions, mobile home robots and a separate child-learning product. Features are manufacturer claims. Open a card for package, payment and primary-source details.
Romi Lacatan · Japan: ¥98,780 hardware + a conversation plan. Kept in JPY; no unverified exchange-rate conversion.
A desktop character; AI dialogue through an extra station.
Robot $139.99 + station $99Compare +A walking desktop companion with reactions and games.
$279 · saved offerCompare +A desktop conversational character; a close functional reference.
¥98,780 + conversation planCompare +A mobile robot pet for play and interaction at home.
$499.90 · 21 Apr 2026 referenceCompare +A mobile home camera and communication robot; an adjacent use case.
$399 · saved international offerCompare +Child learning, stories and games; a separate audience.
$299 · US snapshotCompare +These reference segments overlap. Their offers and audiences do not establish Buddy demand.
93 offer records across 44 geographies. Switch between the broad research, 17 selected devices and company locations. Offers and shipping claims are distinct from unit sales.
Full 10 Oct research · offers · 93 records · 44 geographies. Archive dated 10 October; current stock and checkout were not reverified.
Grey means missing evidence. The map shows recorded offers or company locations, not sales or market share.
Choose a geography and a segment. A statistical audience proxy passes four unmeasured filters. The result is hypothetical purchases in one year, not measured demand, orders or a Buddy sales forecast.
One device per purchasing household is assumed. These are model units, not customers, orders or revenue.
Actual annual competitor sales and actual occupied/free shares remain unknown. This optional calculation applies an explicit hypothetical sales count to the same geography, segment and year, divided by the model’s unrounded annual units. It is not a measured free market.
Enter an integer from zero to the unrounded model volume. A blank input remains unknown.
Hardware value is a conditional gross device value, not revenue, profit or market size. Shipping, taxes, channel margin, discounts, returns, subscriptions and costs are excluded. The saved S1 benchmark converts the 29,900 RUB planning price using 85.4173 RUB/USD on 9 October 2026; a global Buddy price was not approved.
Lower / base / upper are three sets of assumptions. The nine records with an audience base have numerical scenarios. Missing bases stay unknown. These rows are not summed. The table retains the saved display rounding; the CSV preserves the original unrounded formulas and values.
| Geography / segment | Statistical proxy / period | Annual units · lower / base / upper | Device value USD · lower / base / upper |
|---|---|---|---|
| RussiaS1 · Personal conversational companions | UnknownNo validated denominatorUnvalidated | Not calculated · missing base | Unknown |
| United StatesS1 · Personal conversational companions | 117,597,021Households × rounded broadband shareACS2020–2024Source ↗ | 1,180 / 12,700 / 58,800 | $412,000 / $4,450,000 / $20,600,000 |
| Dubai onlyS1 · Personal conversational companions | 795,200Private households in Dubai only2024 revisedSource ↗ | 8 / 86 / 398 | $2,780 / $30,100 / $139,000 |
| Saudi ArabiaS1 · Personal conversational companions | 8,174,674All householdsCensus2022Source ↗ | 82 / 883 / 4,090 | $28,600 / $309,000 / $1,430,000 |
| BrazilS1 · Personal conversational companions | 15,400,000Households with any smart devicePNAD2025Source ↗ | 154 / 1,660 / 7,700 | $53,900 / $582,000 / $2,700,000 |
| RussiaS2 · Mobile home robots | UnknownNo validated denominatorUnvalidated | Not calculated · missing base | Unknown |
| United StatesS2 · Mobile home robots | 117,597,021Households × rounded broadband shareACS2020–2024Source ↗ | 1,180 / 12,700 / 70,600 | $469,000 / $5,070,000 / $28,200,000 |
| Dubai onlyS2 · Mobile home robots | 795,200Private households in Dubai only2024 revisedSource ↗ | 8 / 86 / 477 | $3,170 / $34,300 / $190,000 |
| Saudi ArabiaS2 · Mobile home robots | 8,174,674All householdsCensus2022Source ↗ | 82 / 883 / 4,900 | $32,600 / $352,000 / $1,960,000 |
| BrazilS2 · Mobile home robots | 15,400,000Households with any smart devicePNAD2025Source ↗ | 154 / 1,660 / 9,240 | $61,400 / $664,000 / $3,690,000 |
| RussiaS3 · Child education, separate segment | UnknownNo validated denominatorUnvalidated | Not calculated · missing base | Unknown |
| United StatesS3 · Child education, separate segment | 30,400,000Households with a child aged 0–132024Source ↗ | 608 / 5,470 / 30,400 | $182,000 / $1,640,000 / $9,090,000 |
| Dubai onlyS3 · Child education, separate segment | UnknownNo validated denominatorUnvalidated | Not calculated · missing base | Unknown |
| Saudi ArabiaS3 · Child education, separate segment | UnknownNo validated denominatorUnvalidated | Not calculated · missing base | Unknown |
| BrazilS3 · Child education, separate segment | UnknownNo validated denominatorUnvalidated | Not calculated · missing base | Unknown |
| RussiaS4 · Excluded reference | UnknownNo validated denominatorUnvalidated | Excluded · unidentified product | Unknown |
| United StatesS4 · Excluded reference | UnknownNo validated denominatorUnvalidated | Excluded · unidentified product | Unknown |
| Dubai onlyS4 · Excluded reference | UnknownNo validated denominatorUnvalidated | Excluded · unidentified product | Unknown |
| Saudi ArabiaS4 · Excluded reference | UnknownNo validated denominatorUnvalidated | Excluded · unidentified product | Unknown |
| BrazilS4 · Excluded reference | UnknownNo validated denominatorUnvalidated | Excluded · unidentified product | Unknown |
Saved offers: RED Loona 89,999 RUB; the claimed 20% coupon implies 71,999.20 RUB before checkout. Eilik.ru: 31,500 RUB. These are 10 Oct offers, not verified sales. RED claims Russian voice/app and a 12-month warranty; eilik.ru claims three months. Authorization and product quality were not independently tested.
Measure paid orders, returns, voice/memory usefulness, cloud costs and repair. A small Russia pilot does not prove international demand. The national household table was located but its numeric denominator was not validated; Moscow’s 97% broadband statistic is city infrastructure only.
Rosstat household table · denominator gap ↗129,227,496 households × a rounded 91% broadband share gives about 117.6m as an infrastructure proxy. ACS 2020–2024 median annual household income is $80,734 in 2024 dollars. Neither number measures device need or price acceptance. The separate child reference is 30.4m households with children aged 0–13 in 2024, not Miko’s exact target ages.
Test paid conversion, privacy expectations, returns and differentiation from phones and apps in a competitive market.
795,200 private households in revised 2024 Dubai statistics; collective residential compounds excluded. No UAE total or income/Internet filter is inferred. Saved Virgin UAE Loona offer: AED 2,099, about $571.55 at the broad report’s 10 Oct FX rate. The national offer evidence and Dubai audience are different geographic scopes.
Test mixed-language English/Arabic use, gift-to-daily-use retention, SKU costs, channel margin and local service.
8,174,674 total households in the 2022 census; 4,153,497 Saudi households. Median monthly disposable household income in 2023: SAR 7,362 overall, 13,655 Saudi and 3,657 non-Saudi. These are household incomes, not device purchase rates. The 99.6% individual Internet-use measure is not a household filter.
Test Arabic dialect/content acceptance, willingness to pay, channel delivery and repair. Ropet’s saved Saudi record requires an enquiry; no delivery is inferred. UAE expatriate demand is not transferred to Saudi households.
PNAD 2025: 15.4m households with any smart device; 76m connected households. The narrower proxy is not AI-robot owners or the whole potential audience. Mean monthly income per household member is BRL 2,316 in 2025; it is neither a median nor total family income. The saved BR Metaverso catalogue covers Eilik, Eiliko, EMO variants, AIBI and Loona; authorization, stock and sales are unconfirmed.
Test Brazilian Portuguese, import/tax/service costs, local payments and instalments, and recurring usefulness. Premium import buyers do not represent the whole country.
CIS countries need separate language, import, payment and service checks. Canada / UK / Australia: reuse English UX while comparing country-level commerce and repair. Mexico / Spain: Spanish is already offered by competitors; test a local user need and payment funnel. Germany / France and other EU countries need separate pricing, language and service checks. Oman / Bahrain / Qatar are research candidates only, without validated channels or budgets.
No country ranking, entry order, market-share estimate or funding budget was approved in the source.
Income statistics use different periods, denominators and definitions and are not compared as a common affordability score. Original sources and access limitations are preserved in the 33-source CSV. Several primary documents were read through indexed text; direct downloads returned 403/502 or redirects. No new independently downloaded source is claimed here.
Each percentage applies to the households that passed the previous filter. None is measured. Changing one updates the result; it does not validate demand.
annual units = audience × need × price acceptance × channel/service reach × annual purchase117,597,021 × 3% × 40% × 30% × 3% = 12,700.48 model units before display rounding
US household proxy: 129,227,496 households × a rounded 91% broadband share ≈ 117,597,021. It is a derived proxy, not a validated buyer count.
Saved inputs are dated 10 October 2026; statistical periods differ. The 20 saved records cover five geographies × four segments. The five S4 records remain explicitly excluded. Segments overlap and must not be summed. Comparable competitor unit sales are missing, so occupied and free-market shares remain unknown.
Three proposed sources of revenue. Prices, package contents and paid conversion have not been approved.
The first commercial format is a desktop Buddy. Its price must cover the package, logistics, fees, warranty and acquisition.
Price scenariosA cloud assistant with clear quotas. Core conversation, work integrations and specialised packages are proposals.
Hypothesis / billing not enabledExtra recording and translation minutes, music generation and video renders. Partner commission is a separate hypothesis.
Hypothesis / package not approvedThe customer journey is a hypothesis. CAC, purchase conversion, retention and referrals have not been measured.
First, the contribution of one device and one user. Then, cash needs by stage. The pilot tests demand.
One desktop Buddy configuration. The price must cover the device, charging base, delivery, fees, warranty and acquisition. Factory quotes and paid orders determine whether this range works.
Dashed line: Miko Max monthly content-plan reference, $14.99, official FAQ reviewed 11 Oct 2026. It is a different service and quota, not Buddy revenue or an equivalent cost benchmark. Primary source ↗
30 days; Live $0.05/min; 2 LLM requests/min, 2,500 input and 60 output tokens/request, $0.4/$1.6 per million tokens; infrastructure $1; reserve 20%. Live minutes include pauses. Infrastructure is unmeasured. Rates are dated assumptions, not a current quote.
(30 × m × 0.05 + 30 × m × 2 × (2500 × 0.4 + 60 × 1.6) / 1,000,000 + 1) × 1.2| Minutes / day | USD / user / month |
|---|---|
| 5 | $10.59 |
| 10 | $19.99 |
| 15 | $29.38 |
| 20 | $38.78 |
| 25 | $48.17 |
| 30 | $57.57 |
Recording, translation, payments and support are excluded. STT/TTS is not added twice on top of Live. A cheaper architecture needs its own model.
At 10 voice minutes a day, this dated model costs about $19.99 per user per month before payments, support and other processing. An approved package needs a price, an explicit voice quota, separate recording/translation allowances and paid overages or a verified cheaper architecture.
Blank means unknown. Enter an explicit zero only when a cost is excluded. These inputs are a local scenario, not a published tariff.
contribution = net subscription revenue × (1 − payment fee / 100) − cloud cost − other variable costsNet revenue is after applicable indirect taxes/refunds but before the payment fee entered above. Other variable costs must include support, storage and any recording, translation or tools excluded from the voice model. Fixed development and overhead are funded from contribution; this is not net profit. Test typical and heavy use, repeat payment and retention. Package approval requires positive contribution at the promised quota and a measured plan for excess use.
development + fixed costs × duration + manufacturing advances + reserve − available cash − confirmed receipts
Amounts and timing are unknown until estimates, factory terms and a pilot plan exist. Unknown is not zero. This is a decision diagram, not an invented cash-flow curve or an announced funding round.
Monthly device contribution + monthly subscription contribution must cover fixed monthly costs. A numeric break-even point requires positive contributions and known fixed costs.
The 29 September workbook separates development, per-unit work, fixed support and variable AI. The 9 October sales model uses different configurations and currencies and predates the current camera, mechanical paws, Wi-Fi and charging-base requirements. Its alternative price tiers and reference valuations remain in the detailed materials; they are not the international Buddy offer. Neither model is an approved launch budget.
We show each result alongside its verification boundary. Supplier contacts are not partnerships.
Editable Blender / GLB files and NAVY Buddy in a NY cap. Web models and scenarios have been checked in the browser; these are designs, not manufacturing samples.
Scenarios and models ↗Web / Telegram, a voice prototype, confirmation of notes and actions, and a financial test harness. Native and hardware acceptance of specific workflows remains separate.
Financial formulas, 121 scenario stories and a sourced competitor registry are available. The next result is a measured user pilot.
Founder of Grand House, Russia’s largest TikTok house by the combined follower count of its creators. The project’s creators reached a combined 150 million followers. His experience includes creator development, marketing and more than 100 advertising campaigns with major brands.
For Buddy: experience building a community, running content production and testing a product through creator-led campaigns. The pilot will measure paid conversion, acquisition cost and repeat use.
The Grand House figures above were supplied and approved by the founder on 11 October 2026. They describe accumulated professional experience and combined subscriptions, rather than unique people or the Buddy customer base. Public profiles also describe Grandee Production, work across media, AI and digital products, Elysium Gate venture architecture and authorship of the “5 Scene” method and workbook.
The engineering sample needs embedded electronics, acoustics, mechanical design and physical testing alongside mobile and backend integration. Specific roles and supplier commitments follow the agreed sample scope.
AXIOMA Media professional profile ↗Define acceptance thresholds before the experiment. They determine the next investment decision.
Test: Pilot one complete workflow; observe results, corrections and repeat use.
Decision criterion: Proceed when a repeat-use threshold agreed in advance is met and the error rate is acceptable.
Test: Demonstrate the prototype and test willingness to pay; measure the channel, conversion and customer acquisition cost (CAC) separately.
Decision criterion: Order a batch after evidence of paid demand and a positive contribution per sale.
Test: Log minutes, tokens, tools and storage for typical and heavy-use cohorts.
Decision criterion: Approve a package only with a positive contribution and quotas that withstand heavy use.
Test: Engineering sample: audio, camera, mechanical paws, Wi-Fi, heat, current draw, charging base and recovery; obtain quotes for the bill of materials (BOM), minimum order quantity (MOQ), flashing and testing.
Decision criterion: A batch follows an accepted specification, tests and a payment schedule.
Test: Test two accounts, permission revocation, stopping recording, and deleting the source and memory across all connected stores.
Decision criterion: Start the pilot after isolation checks and the complete deletion workflow pass.
Test: Test the SDK, access rights, network failure, repeated actions and switching AI providers.
Decision criterion: Promise a feature after its result is verified on the selected device and account.
The plan is based on decision gates. Dates and batch size depend on the chosen specification, quotes and acceptance.
One device → phone → cloud → result workflow.
A working sample with measured latency, audio, current draw and recovery. Dependencies: selected board, SDK, firmware bridge and package contents. Proceed after physical acceptance of the workflow and permissions.
Usefulness, repeat use, willingness to pay and actual AI cost.
Cohort data, service cost, errors and corrections. Dependencies: consent, deletion, support and an experiment plan. Proceed after preselected usefulness and economic thresholds are met.
Specification, quotes, tooling, testing, quality and a cash schedule.
An accepted BOM, MOQ, production firmware, test bench and compliance plan for the selected markets. Proceed after sample validation and funding for the batch is secured.
Delivery, support, returns and paid recurring use.
Observed margin and retention, with tested returns and warranty handling. Dependencies: manufacturing availability, payments, logistics and service. Scale after demand and a positive contribution are verified.
The next investment should produce a measurable result: an engineering sample and a pilot that informs a manufacturing decision.
Request investor materials ↗[email protected] · for investors and partnersDevelopment, samples, tests and a pilot. The amount follows estimates and stages; no approved round size is announced.
Embedded engineering, electronics, acoustics, mobile integration and physical QA to deliver a verified prototype.
Design for manufacture (DFM), BOM, quotes, MOQ and production testing to support a justified batch decision.
Request the source models, technical materials and a discussion of the sample and pilot scope by email.