---
title: "Personal AI Agents"
description: "A landscape of delegated work, approval boundaries and evidence of completion."
---

> Documentation Index
> Fetch the complete documentation index at: https://latentlit.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Personal AI Agents

import FocusCanvas from '../../../../components/FocusCanvas.astro';
import LandscapeOverview from '../../../../components/landscapes/LandscapeOverview.astro';
import ResearchTable from '../../../../components/landscapes/ResearchTable.astro';

“Agent” can mean a shopping recommendation, a recurring alert, a prepared reply
or a completed booking. The useful comparison is **what responsibility the
person delegates, what authority the service receives and how completion is
verified**.

This study includes **23 company groups** and a **separate five-case watchlist**.
Large ecosystem platforms sit alongside specialist services. Some combine personal
and professional use. Useful
boundary comparisons remain explicit rather than being relabeled autonomous.

My working conclusion: a narrow service that reliably prepares the right next
step may be more useful than a broad assistant that needs constant supervision.
That is a product hypothesis, not a market-size forecast or tested ranking.

[Browse companies](#company-explorer) · [Check the watchlist](#watchlist-and-status) · [Compare delegation patterns](#delegation-patterns) · [Evaluate an agent](#how-to-evaluate-an-agent)

## Landscape at a glance

<a id="figure-1"></a>

![Personal AI overview: discover, watch, prepare and act within a scope. Broad platforms span several modes; AI-only versus human-assisted execution is a separate dimension. Permissions, confirmation, receipts and recovery surround the task.](/figures/landscapes/personal-agents-overview-v1.webp)

**Figure 1.** Delegation modes and the controls around them. Author’s synthesis; AI-generated illustration.[^daydream][^scouts][^jo][^ohai][^mindtrip][^duckbill][^broad-chatgpt][^broad-gemini][^broad-spark][^broad-copilot][^broad-apple][^broad-alexa][^broad-qwen]

The image gives the territory at a glance; the explorer and comparisons below
preserve the detail. Placement shows a product role, not market share, measured
quality or a partnership. Categories overlap.

## Company explorer

Funding was reviewed **October 11, 2026** and has its own column, separate from
access and approval limits. Each event retains its date. Company-level rounds,
cumulative capital and valuation are not interchangeable. Historical disclosures
may not be the latest; unknown funding does not mean unfunded. USD unless marked CAD
or RMB. Product evidence retains its original inspection dates.

Search a company, region, task or phrase such as “approval.” Filter by category
to compare similar responsibilities. Related products are grouped once.

The directory retains dated September 2026 research, supplemented by selected
October 10 product checks. References preserve their dates. Product documentation
establishes advertised capabilities, not independently verified task success.
Funding was reviewed **October 11, 2026** and has a separate column from
product status. Historical financing is not a guarantee of today’s stage;
reported and unresolved amounts stay labeled. Round size, cumulative capital,
valuation and secondary share sales are distinct. No accounts, purchases or
live agent tasks were used in this study.

<a id="company-directory"></a>

| Company | Category | Delegated experience / business boundary | Control and availability limits | Funding / ownership (dated) | Evidence boundary / status |
| --- | --- | --- | --- | --- | --- |
| **[OpenAI / ChatGPT agent](https://openai.com/index/introducing-chatgpt-agent/)** | Broad platforms | Web research, browser actions and editable outputs using a virtual computer; personal/professional workflows.[^broad-chatgpt] | Consequential-action confirmation, interruption and browser takeover described. Some critical tasks need active supervision; launch plan access is dated. | **Company round** — \$122B committed capital, Mar. 31, 2026. Committed capital is not a received-proceeds measure or a ChatGPT-only raise.[^finance-openai] | Launch access and controls retain the announcement’s date. |
| **[Google / Gemini, Spark](https://gemini.google/overview/agent/spark/)** | Broad platforms | Connected-app tasks, recurring schedules and background work; Google service context.[^broad-spark][^broad-gemini] | Opt-in; app connections off by default. Designed to check before major actions; select countries, age/plan restrictions and rollout apply. | **Listed parent: Alphabet.** Gemini is a Google product, not a separately funded startup.[^finance-alphabet] | Spark and connected-app access need product-specific checks. |
| **[Microsoft / Copilot](https://www.microsoft.com/en-us/copilot/solutions/individuals)** | Broad platforms | Assistant and document/productivity bundle; plans advertise agentic research and analytics.[^broad-copilot] | Prepare / artifact comparator. This page does not establish autonomous checkout, universal browser control or one approval policy for all Copilot products. | **Listed parent: Microsoft.** No standalone Copilot venture round.[^finance-microsoft] | Copilot branding does not imply one common permission model. |
| **[Apple / Apple Intelligence](https://www.apple.com/apple-intelligence/)** | Broad platforms | Contextual action suggestions and browsing assistance within the device/OS experience.[^broad-apple] | Bounded platform comparator. Suggested actions are not unrestricted autonomous execution; device, language and region limits apply. | **Listed parent: Apple.** Platform capability, not a separately financed startup.[^finance-apple] | Suggested actions are not unrestricted autonomous execution. |
| **[Amazon / Alexa+](https://www.aboutamazon.com/news/devices/new-alexa-generative-artificial-intelligence)** | Broad platforms | Home coordination, supported reservations, shopping and connected-device actions.[^broad-alexa] | Supported service actions and forward-looking web-agent examples coexist in the announcement. Local rollout and per-service confirmations are not tested. | **Listed parent: Amazon.** No standalone Alexa+ venture round.[^finance-amazon] | Supported service actions and future plans remain distinct. |
| **[Alibaba / Qwen App 千问](https://www.alibabagroup.com/en-US/document-1948497434959151104)** | Broad platforms | Service coordination across commerce, payments and travel; China-facing public testing announced Jan. 2026.[^broad-qwen] | Explicit user confirmation for in-chat Alipay payments. Shopping can hand off to Taobao; separate Task Assistant was invite-only beta in that release. | **Listed parent: Alibaba.** Qwen App is an ecosystem product, not a separately funded company.[^finance-alibaba] | The announcement’s beta scope is not a current access guarantee. |
| **[Ohai](https://www.ohai.ai/)** | Household | Turn school emails, PDFs and photos into schedules, reminders and tasks. Consumer household offering.[^dir-ohai] | **AI + human support.** Complex documents can receive human review. Public free/premium offering; local integrations and actual task reliability not tested.[^dir-ohai] | **Strategic investment**, Aug. 19, 2025; Muse-led. Amount and series label undisclosed. Earlier seed is not the label of this new round.[^dir-ohai-funding] Historical disclosure; later round unverified. | Human review is part of the household service. |
| **[Duckbill](https://getduckbill.com/)** | Household | Personal calls, bookings, follow-ups and administration. Consumer memberships plus team/developer channels.[^dir-duckbill][^dir-duckbill-funding] | **Hybrid concierge:** every task gets AI preparation and human handling. Sells task capacity, not unlimited autonomous execution.[^dir-duckbill] | **~\$37M cumulative**, undated cofounder disclosure checked Oct. 11, 2026. Not a newly identified round.[^finance-duckbill-founder] Historical seed + A: \$33M combined (Sept. 2023).[^dir-duckbill-funding] | A hybrid concierge, not an AI-only completion system. |
| **[jo](https://askjo.ai/)** | Personal assistant | Connected-app context, overnight briefings, reminders and follow-up drafts. Individuals/families with work-life overlap.[^dir-jo][^dir-jo-yc] | **Approval-led.** Current profile says consequential actions require approval; Mac/web generally available, Apple Silicon for Mac. Native mobile is still planned.[^dir-jo-yc] | **YC-backed; round and amount not verified.** Do not infer a financing amount from YC's standard terms.[^dir-jo-yc] | Native mobile remains planned in the cited product evidence. |
| **[Yutori / Scouts, Delegate](https://yutori.com/)** | Web delegation | Recurring monitoring, research and connected-app/browser tasks. Mixed consumer, professional and API channels; Delegate leans professional.[^dir-yutori-product][^dir-yutori-delegate] | **Not monitoring-only:** Apr. 2026 updates describe actions. Delegate says it asks before sensitive actions; this is a policy claim, not a complete permission audit.[^dir-yutori-product][^dir-yutori-delegate] | **Seed (historical verified round)** — **USD15M**, Mar. 27, 2025. No later round verified in this pass.[^dir-yutori-funding] | Sensitive-action policy is documented, not independently audited. |
| **[Mindtrip / Flights, Stays](https://mindtrip.ai/)** | Travel | Flight comparison/booking and hotel search with price monitoring. Consumer product plus B2B travel tools.[^dir-mindtrip-flights][^dir-mindtrip-stays][^dir-mindtrip-funding] | **Traveler chooses and checks out.** Flights launched May 2026, Stays July 2026; hotels support direct or supplier checkout. Not unrestricted purchasing on the traveler's behalf.[^dir-mindtrip-flights][^dir-mindtrip-stays] | **Strategic investment; series/amount undisclosed**, Dec. 4, 2025 reporting. **USD22.5M total raised then**, not this round's size.[^dir-mindtrip-funding] Historical disclosure; later round unverified. | Booking launch dates do not establish unrestricted buying authority. |
| **[Daydream / Dahlia Labs](https://daydream.ing/)** | Shopping | Style-aware fashion search and refinement. **Boundary comparator:** personal discovery, not verified end-to-end buying.[^dir-daydream] | Free consumer offering. FAQ sends the user to choose where to complete the purchase; do not infer automatic checkout.[^dir-daydream] | **Seed (historical)** — **USD50M**, June 20, 2024. Fashion Daydream, not the similarly named SEO company.[^dir-daydream-funding] | A discovery comparator; retailer checkout remains user-led. |
| **[Macaron 马卡龙 / 心洲科技](https://macaron.im/)** | Personal assistant | Memory-informed personal mini-apps and connected-service actions. Consumer product; parent also has model/infrastructure business.[^dir-macaron][^dir-macaron-funding] | **Permission-scoped, supervised.** Policy warns that tasks can mis-share information; users should supervise/interrupt and can revoke connected-app access.[^dir-macaron-policy] | **Series A (reported)** — nearly **USD50M**, disclosed June 2, 2026; July report places closing in early 2026 and **USD60M cumulative**. Distinct from Second Me.[^dir-macaron-disclosure][^dir-macaron-funding] | Macaron / 心洲科技 is not Second Me / 心识宇宙. |
| **[Flowith / Agent Neo, FlowithOS](https://flowith.io/tools/agent-neo/)** | Creation & research | Multi-step creation of websites, games, presentations and research artifacts. **Mixed/prosumer**, not household-only.[^dir-flowith][^dir-flowith-origin] | Review generated artifacts. Browser demonstrations do not establish payment-approval defaults or reliable personal purchasing.[^dir-flowith] | **Seed+**, Mar. 4, 2026 reporting. **Amount unresolved:** press describes seed + seed+ together as USD10M-scale; investor post says USD9M raised.[^dir-flowith-funding][^dir-flowith-investor] | Browser demonstrations are not payment-control evidence. |
| **[Second Me 小己 / 心识宇宙](https://second.me/)** | Social representation | An avatar represents memories/personality in conversations and services. Personal/family use plus consulting and business inquiries: **mixed**.[^dir-secondme][^dir-secondme-store] | **Social delegation, not general execution.** Current avatar claims do not establish autonomous account administration. Legacy Me.bot capabilities are not automatically current Second Me features.[^dir-secondme][^dir-secondme-policy] | **Seed (historical entity-matched)** — **USD5M**, Apr. 2024, reported Aug. 15, 2024. Latest stage unverified. Macaron's A round does not belong here.[^dir-secondme-funding] | Legacy Me.bot features are not automatically current Second Me features. |
| **[StudyFetch](https://www.studyfetch.com/)** | Learning | Turn course materials and deadlines into study schedules, notes and practice. Student product plus institutional offerings.[^dir-studyfetch] | **Prepare / bounded learning.** Public product; not external task execution or independently measured learning improvement.[^dir-studyfetch] | **Series A (reported historical round)** — **USD11.5M**, June 2025; Owl Ventures-led with College Board. Official Jan. 2026 retrospective.[^dir-studyfetch-funding] | Learning support, not general external-account execution. |
| **[Origin](https://useorigin.com/)** | Personal finance | Account-context financial analysis, recurring briefs and scenario planning. Consumer and employer channels.[^dir-origin] | **Watch / prepare comparator.** Official disclosure says advice is non-discretionary: the user decides whether to act. Not autonomous investing or a reliability endorsement.[^dir-origin] | **Series B (historical)** — **USD56M**, Aug. 30, 2021. Later stage not verified; do not present this as current total capital.[^dir-origin-funding] | Non-discretionary advice is not autonomous investing. |
| **[Pilot](https://www.pilotplans.com/)** | Travel | Collaborative AI itinerary planning, imported travel material and booking support. Consumers plus group/business services.[^dir-pilot][^dir-pilot-pricing] | **Prepare / booking comparator.** Real human support; not an independently verified autonomous buyer. Distinct from the accounting company Pilot.[^dir-pilot-pricing] | **~CAD2.2M cumulative**, company-reported; source undated, checked Oct. 11, 2026. Latest round amount, date and stage unconfirmed.[^dir-pilot-funding] | Travel Pilot is not the similarly named accounting company. |
| **[Phia](https://phia.me/)** | Shopping | Fashion price thresholds, watchlists and coupon assistance. Consumer app with retailer-performance business channels.[^dir-phia-product][^dir-phia-funding] | **Watch / purchase-support comparator.** Retailer purchase flow, not verified autonomous buying. Current app financing wording differs from the dated release.[^dir-phia-product] | **Series A** — **USD35M**, Jan. 27, 2026 announcement. Use the dated round amount; current listing says USD35.5M, an unresolved discrepancy rather than a second round.[^dir-phia-funding][^dir-phia-product] | USD35M release versus USD35.5M app wording remains unresolved. |
| **[Littlebird](https://littlebird.ai/)** | Personal assistant | Context-aware drafts and recurring briefings, with a personal/household lane alongside work-first use.[^dir-littlebird][^dir-littlebird-personal] | **Watch / prepare.** Scheduled prompts create private deliverables; optional connected apps and controllable capture. Per-action approval enforcement not verified.[^dir-littlebird-routines] | **Seed** — \$11M, March 2026. Current company facts explicitly name the stage.[^finance-littlebird-facts][^dir-littlebird-funding] | Capture controls and private drafts differ from write-action authority. |
| **[Town](https://www.town.com/)** | Personal assistant | Inbox, calendars, to-dos and routines; household-school examples alongside professional/team use. **Mixed/prosumer.**[^dir-town-funding] | **Approval-led.** Trust page says email sending and calendar invitations require approval; audit/control claims are documented, not tested.[^dir-town-trust] | **Series A** — **USD55M**, June 3, 2026 company announcement.[^dir-town-funding] | Approval requirements are provider claims, not an enforcement test. |
| **[SEELE 全灵](https://www.seeles.ai/about-us)** | Creation & research | Agent Workspace turns prompts into game assets/code and playable projects, with multi-turn editing and shared runtime. **Creator/prosumer**, not personal administration.[^dir-seele-agent] | **Prepare / artifact execution.** Review, iterate and publish creative work. Current subscriptions/Koins are distinct from future play/social monetization.[^dir-seele-agent] | **Pre-A (historical)** — **USD10M-scale**, Nov. 18, 2024 reporting. Precise amount and later stage not verified.[^dir-seele-funding] | Creative publishing is not household-account administration. |
| **[YouMind](https://youmind.com/)** | Creation & research | Collect sources, develop research and produce editable writing, slides, video or webpages; reusable skills. Personal learning and professional creation overlap.[^dir-youmind][^dir-youmind-founder] | **Prepare / artifact creation.** Human editing and source review remain explicit. No verified external checkout or autonomous account administration.[^dir-youmind] | **Seed, historical founder disclosure** — 2024; amount undisclosed.[^dir-youmind-founder] Later 2025 financing reports use conflicting stage labels; current stage unconfirmed.[^finance-youmind-conflict][^finance-youmind-angel] | Human editing and evidence review remain necessary. |

## Watchlist and status

These five cases are separate because access, company status or consumer
execution remains unresolved in the cited evidence. A live website or a funding
announcement does not resolve those questions. **This is a dated verification
watchlist, not a claim that every issue remains unchanged today.**

| Company / official destination | Why it remains separate | Funding / ownership (dated) | Next verification |
| --- | --- | --- | --- |
| **[myOS Inc.](https://www.myos.ca/)** | Personal myOS is Alpha; myAgents Beta. Mixed products/consulting; meaningful consumer task execution and traction not established.[^dir-myos] | **Unverified** — no reliable round/amount disclosure located. | Verify usable personal workflows, access and financing; solo-founded does not mean bootstrapped. |
| **[Manus](https://manus.im/)** | Aug. 2026 notice announces a return to independent operations. Browser tooling documents interruption/payment pause, but affected-account restoration and final ownership were not tested.[^dir-manus-status][^dir-manus-browser] | **>\$500M reported raise**, Oct. 8, 2026; stage unnamed.[^finance-manus] | Check final ownership, restored account access and payment controls separately from funding. |
| **[Selfin](https://selfin.app/)** | The live YC profile now describes institutional liquidity, while retaining an earlier AI-bank launch. Current consumer-agent positioning is not established.[^dir-selfin] | **YC-backed** — round amount and later stage unverified.[^dir-selfin] | Current YC profile describes institutional prediction-market liquidity; do not treat its earlier personal-finance launch as the current offer. |
| **[Instinct](https://instinct.com/)** | Personal email, messaging, phone and computer tasks advertised; **private access/invitation**. Per-action approval behavior and task reliability not established.[^dir-instinct] | **Reported \$250M Series B**, Aug. 26, 2026; \$350M cumulative, \$2.5B valuation.[^finance-instinct] | Verify invitation access, permissions and task reliability independently of financing. |
| **[Do Browser](https://www.dobrowser.io/)** | Public extension/plans and browser/form task claims. **Consumer-workflow fit and permission evidence remain incomplete in this pass**; command-driven multi-step work can still qualify.[^dir-do-browser] | **Unverified** — no reliable funding disclosure located. | Inspect a bounded personal workflow and permission controls; public plans do not establish reliable execution. |

For Manus, an announced transition is not proof of completed ownership or
restored account access. For private-access products, advertised ambition
should not become a claim of generally available functionality.

## Delegation patterns

Anthropic’s distinction between fixed-path workflows and model-directed agents
is useful for architecture. It does not tell us who the customer is, who is
allowed to spend money or whether a human operator finishes the job.[^dir-agent-definition]

Start with three product questions: whose goal is being pursued, what happens
beyond the answer and who controls the consequential step. [Figure 1](#figure-1)
separates levels of responsibility; it is not an official autonomy scale.
NIST likewise treats permissions, monitoring, reversibility and autonomy as
distinct dimensions.[^dir-tool-boundaries]

| Lens | What it does | A useful test |
| --- | --- | --- |
| **Discover** | Find and compare options. The user makes the external commitment. |  |  |
| **Watch** | Monitor changes and bring relevant information back. |  |  |
| **Prepare** | Turn scattered context into a draft, briefing or plan. |  |  |
| **Act within a scope** | Change an external system under an explicit task and approval boundary. |  |  |

A separate dimension: AI only or AI plus human operators. Duckbill explicitly describes AI preparation with human execution on every task.

## Product lessons beyond the label

### Big platforms do not have one common autonomy boundary

ChatGPT agent describes permission checks for consequential actions and browser
takeover. Gemini Spark describes opt-in connected apps and checks before major
actions. Alibaba’s January Qwen App announcement separates explicit payment
confirmation, public testing and its invite-only Task Assistant.[^broad-chatgpt][^broad-spark][^broad-qwen]

Apple’s contextual suggestions and Microsoft’s research/analytics features are
useful boundary comparisons, not evidence of unrestricted external execution.
Alexa+ combines supported service actions with forward-looking agentic examples.
Do not infer current access or a universal confirmation policy from the provider’s
name.[^broad-apple][^broad-copilot][^broad-alexa]

**The product lesson:** classify an actual task and its current release. “Agent”
is not a substitute for checking the permission, execution and recovery contract.

### Ohai and Duckbill: reduce the household’s mental load

Ohai connects messy household information to schedules and tasks, with human
support for difficult inputs. Duckbill uses AI preparation and human handling
for personal administration. They make different promises about how work is
completed.[^ohai][^duckbill]

**Why this matters:** human involvement can be part of the product design, not
a failure of the AI. Compare cost and time saved after intake, clarification,
review and follow-up. A hybrid concierge’s price buys service capacity;
it should not be compared with software-only inference costs.

### jo, Town and Littlebird: context is useful only if the controls are clear

These products connect ongoing context to briefings, drafts and routines. The
cited jo profile and Town trust page describe approval for consequential actions.
Littlebird advertises routines and controllable capture, but that does not
establish identical approval enforcement.[^dir-jo-yc][^dir-town-trust][^dir-littlebird-routines]

**Why this matters:** knowing more about someone and having more authority are
different permissions. Inspect the settings for collection, retention, sharing
and action separately. Ask what requires confirmation, what can run in the
background and how a person revokes access.

### Yutori: watching is not the same as acting

Scouts illustrates recurring monitoring; Delegate describes connected-app and
browser work. The changelog includes action capabilities, so Yutori should not
be reduced to alerts alone.[^dir-yutori-product][^dir-yutori-delegate]

**Why this matters:** proactive work needs a stopping rule. A useful monitor
must avoid missing important changes without overwhelming the person with
noise. An actor must also respect limits, confirm sensitive steps and handle
partial completion. The same architecture can require different product controls.

### Mindtrip, Daydream and Phia: discovery can be valuable without auto-buying

Mindtrip offers booking experiences, Daydream sends shoppers toward partner
purchases, and Phia combines price watching with purchase support. None of those
facts establishes unrestricted purchasing on a person’s behalf.[^dir-mindtrip-flights][^dir-mindtrip-stays][^daydream][^dir-phia-product]

**Why this matters:** the final action boundary is part of the product. The
assistant can do substantial useful work while leaving the decision and payment
to the user. Evaluate constraints, total price, disclosed incentives and the
remaining effort, rather than rewarding automation for its own sake.

### Macaron, Second Me and creative agents: different meanings of delegation

Macaron’s memory-informed tools and connected actions differ from Second Me’s
social representation. Flowith, SEELE and YouMind focus on multi-step creative
or research artifacts. Publishing a game is execution within an artifact
workflow, not evidence that the service safely administers bank or travel accounts.[^dir-macaron][^dir-secondme][^dir-flowith][^dir-seele-agent][^youmind]

**Why this matters:** a broad agent count hides differences in authority and
outcomes. Judge a research artifact by provenance and editing quality; judge a
social representative by faithfulness and sharing boundaries; judge an account
action by permission and verified state change.

## How to evaluate an agent

A plausible final answer is not enough when external state can change. Use a
bounded task set with explicit starting conditions, permitted actions and a
definition of completion. NIST’s evaluation work motivates testing behavior
and tool boundaries rather than relying on a label.[^dir-eval-probes]

| Dimension | What to inspect | Example failure |
| --- | --- | --- |
| Outcome | Evidence that the intended state changed | A booking claim without a confirmed reservation |
| Constraint fidelity | Budget, dates, recipients and user preferences | The right item purchased for the wrong delivery date |
| Authority | Allowed tools, permissions and approval enforcement | A consequential action taken without required confirmation |
| Oversight burden | Setup, review, interruptions and correction | The person checks every step and saves no time |
| Recovery | Retry rules, duplicate prevention, rollback and escalation | A retry sends a second message or buys twice |
| Privacy | Necessary data, retention, sharing and revocation | Private context copied into an unrelated service |
| Economics | Cost per completed task, including retries and human work | Cheap inference with expensive supervision |

This is my proposed evaluation checklist, not a scorecard collected on these
companies. A completion rate is meaningful only with the task distribution and
approval policy stated. “No errors” in a few demonstrations is not a reliability
estimate.

## Scope and open questions

The directory includes household help, personal assistants, web delegation,
travel, shopping, creative work, social representation, learning and a
non-discretionary finance comparator. It is not an inventory of every general
assistant, workplace tool or merchant bot. Presence here is not a recommendation
to grant access to financial or personal accounts.

The strongest next evidence would be independently observed completion,
constraint violations, intervention burden and recovery under realistic failures.
Public approval language is useful, but enforcement still needs testing. Also
ask what happens when a service is discontinued, a connector changes or a person
wants to export their history.

For implementation, continue with [Agents](/agents/). For the reasons people
choose these products, use [Consumer AI Products](/products-markets/landscapes/consumer-ai/).

[^dir-ohai]: Ohai, [Ohai product and FAQ](https://www.ohai.ai/). Undated; inspected September 7, 2026.
[^dir-ohai-funding]: Ohai via PR Newswire, [Ohai.ai raises strategic investment led by Muse Capital](https://www.prnewswire.com/news-releases/ohaiai-raises-strategic-investment-led-by-muse-capital-celebrity-and-executive-mothers-to-save-parents-hours-of-mental-load-302533511.html). 2025-08-19.
[^dir-duckbill]: Duckbill, [Duckbill pricing and service scope](https://getduckbill.com/pricing). Undated; inspected September 7, 2026.
[^dir-duckbill-funding]: citybiz, [Forerunner leads USD33M seed and Series A for Duckbill](https://www.citybiz.co/article/467208/forerunner-ventures-lead-33m-seed-and-series-a-for-duckbill/). 2023-09-18.
[^dir-jo]: jo, [jo product](https://askjo.ai/). Undated; inspected September 7, 2026.
[^dir-jo-yc]: Y Combinator, [jo company profile](https://www.ycombinator.com/companies/jo). Undated; inspected September 7, 2026.
[^dir-yutori-product]: Yutori, [Scouts and Delegate product updates](https://yutori.com/changelog). Undated; inspected September 7, 2026.
[^dir-yutori-delegate]: Yutori, [Yutori Delegate](https://yutori.com/delegate). Undated; inspected September 7, 2026.
[^dir-yutori-funding]: Yutori, [Yutori launches with USD15M seed funding](https://yutori.com/press_release.pdf). 2025-03-27. PDF p. 1.
[^dir-mindtrip-flights]: Mindtrip via PR Newswire, [Mindtrip launches flight booking](https://www.prnewswire.com/news-releases/mindtrip-launches-travels-first-all-in-one-agentic-ai-flight-booking-experience-powered-by-partnership-with-sabre-and-paypal-302763838.html). 2026-05-06.
[^dir-mindtrip-stays]: Mindtrip via PR Newswire, [Mindtrip launches Stays](https://www.prnewswire.com/news-releases/mindtrip-launches-mindtrip-stays-bringing-agentic-ai-to-hotel-search-and-booking-302826228.html). 2026-07-15.
[^dir-mindtrip-funding]: PhocusWire, [Capital One Ventures and United Airlines Ventures invest in Mindtrip](https://www.phocuswire.com/mindtrip-ai-travel-planning-capital-one-ventures-united-airlines-ventures). 2025-12-04.
[^dir-daydream]: Daydream, [Daydream FAQ](https://daydream.ing/faqs). Undated; inspected September 7, 2026.
[^dir-daydream-funding]: Daydream via PR Newswire, [Daydream secures USD50M seed financing](https://www.prnewswire.com/news-releases/daydream-secures-50-million-in-seed-funding-to-launch-new-ai-powered-search-and-discovery-shopping-platform-302177481.html). 2024-06-20.
[^dir-macaron]: Macaron, [Macaron personal agent](https://macaron.im/). Undated; inspected September 7, 2026.
[^dir-macaron-funding]: 36Kr, [中国Neo Lab Mind Lab研究](https://eu.36kr.com/zh/p/3916202023660929). 2026-07-29.
[^dir-macaron-policy]: MINDAI PTE. LTD., [Macaron Privacy Policy](https://macaron.im/privacy-policy). 2026-08-17.
[^dir-macaron-disclosure]: 投资界, [心洲科技Mindverse完成A轮融资](https://m.pedaily.cn/first/157343.shtml). 2026-06-02.
[^dir-flowith]: Flowith, [Agent Neo use cases](https://flowith.io/tools/agent-neo/). Undated; inspected September 7, 2026.
[^dir-flowith-origin]: Flowith, [Flowith company profile](https://www.linkedin.com/company/flowith). Undated; inspected September 7, 2026.
[^dir-flowith-funding]: 上海证券报 via Sina, [Flowith完成千万美元融资](https://finance.sina.com.cn/roll/2026-03-04/doc-inhpvtwp3773487.shtml). 2026-03-04.
[^dir-flowith-investor]: Vertex Holdings, [Flowith recognized by Forbes](https://www.linkedin.com/posts/vertex-holdings_forbes30under30-ai-agenticai-activity-7468490933777481728-iGpQ). Undated; inspected September 7, 2026.
[^dir-secondme]: Mindverse AI, [Second Me personal avatar](https://second.me/). Undated; inspected September 7, 2026.
[^dir-secondme-store]: Apple / Mindverse AI, [Second Me—Your Digital Mind](https://apps.apple.com/ph/app/second-me-your-digital-mind/id6475769640). Undated; inspected September 7, 2026.
[^dir-secondme-policy]: MINDVERSE AI PTE. LTD., [Second Me privacy policy](https://docs.second.me/compliance/privacy-policy). 2025-11-05.
[^dir-secondme-funding]: 联合早报, [AI个人助手 和你心有灵犀不是梦](https://www.zaobao.com.sg/finance/singapore/story20240815-4467832). 2024-08-15.
[^dir-studyfetch]: StudyFetch, [Product and study-plan features](https://www.studyfetch.com/). Undated; inspected September 7, 2026.
[^dir-studyfetch-funding]: StudyFetch, [2025 review and financing disclosure](https://sso.studyfetch.com/blog/studyfetch-helps-millions-of-students-learn-responsibly-with-ai-in-2025-2). January 5, 2026; reports the June 2025 Series A.
[^dir-origin]: Origin, [AI Advisor launch and non-discretionary disclosure](https://useorigin.com/resources/blog/introducing-the-first-sec-regulated-ai-financial-advisor). September 9, 2025.
[^dir-origin-funding]: Origin via Business Wire, [Origin secures USD56M Series B](https://markets.financialcontent.com/stocks/article/bizwire-2021-8-30-origin-secures-56m-in-series-b-funding-at-400m-valuation-to-support-employee-financial-health-on-a-global-scale). August 30, 2021.
[^dir-pilot]: Pilot, [Company overview](https://www.pilotplans.com/about). Undated; inspected September 7, 2026.
[^dir-pilot-pricing]: Pilot, [Product plans and support](https://www.pilotplans.com/pricing). Undated; inspected September 7, 2026.
[^dir-pilot-funding]: Pilot, [Company information and backers](https://www.pilotplans.com/for-ai-assistants). Company-reported stage and cumulative funding; page update dates inconsistent; inspected September 7, 2026.
[^dir-phia-product]: Apple / Phia, [Phia: Fashion Deals & Rewards](https://apps.apple.com/us/app/phia-fashion-deals-rewards/id6739351340). Current developer listing; inspected September 7, 2026.
[^dir-phia-funding]: Phia via GlobeNewswire, [Phia raises USD35M Series A](https://www.globenewswire.com/news-release/2026/01/27/3226735/0/en/Phia-Raises-35M-Series-A-to-Build-the-AI-Alignment-Layer-for-Commerce.html). January 27, 2026.
[^dir-littlebird]: Littlebird, [Product overview](https://littlebird.ai/). Undated; inspected September 7, 2026.
[^dir-littlebird-personal]: Littlebird, [Personal and productivity uses](https://littlebird.ai/use-cases/personal-and-productivity). Undated; inspected September 7, 2026.
[^dir-littlebird-routines]: Littlebird, [Routines](https://littlebird.ai/features/routines). Undated; inspected September 7, 2026.
[^dir-littlebird-funding]: Littlebird, [USD11M funding announcement](https://littlebird.ai/blog/littlebird-raises-11m-to-build-the-first-full-context-ai). March 22, 2026.
[^dir-town-funding]: Town, [Series A announcement](https://www.town.com/press/series-a). June 3, 2026.
[^dir-town-trust]: Town, [Trust and approval boundaries](https://www.town.com/trust). Undated; inspected September 7, 2026.
[^dir-seele-agent]: SEELE, [Agent Workspace and company overview](https://www.seeles.ai/about-us). Undated; inspected September 7, 2026.
[^dir-seele-funding]: 36Kr, [全灵完成千万美元Pre-A轮融资](https://eu.36kr.com/zh/p/3026090101564936). November 18, 2024.
[^dir-youmind]: YouMind, [Product and operator](https://youmind.com/), and [Reusable skills](https://youmind.com/skills). Undated; inspected September 7, 2026.
[^dir-youmind-founder]: YouMind cofounder, [YouMind创业故事：是幸运，也是信任](https://www.caicai.me/zh/blogs/youmind-startup-journey-2). November 17, 2025; historical 2024 seed disclosure.
[^dir-myos]: myOS Inc., [myOS company and product stages](https://www.myos.ca/). Undated; inspected September 7, 2026.
[^dir-manus-status]: Manus, [A Note to Our Users](https://manus.im/blog/a-note-to-our-users). 2026-08-11.
[^dir-manus-browser]: Manus, [Browser Operator deep dive](https://manus.im/blog/deep-dive-browser-operator-connector). 2026-05-05.
[^dir-selfin]: Y Combinator, [Selfin company and beta launch](https://www.ycombinator.com/companies/selfin). Undated; inspected September 7, 2026.
[^dir-instinct]: Instinct, [Product and private access](https://instinct.com/). Undated; inspected September 7, 2026.
[^dir-do-browser]: Do Browser, [Product and workflow scope](https://www.dobrowser.io/). Undated; inspected September 7, 2026.
[^dir-agent-definition]: Anthropic, [Building effective agents](https://www.anthropic.com/engineering/building-effective-agents). 2024-12-19.
[^dir-tool-boundaries]: NIST, [Lessons Learned from the Consortium: Tool Use in Agent Systems](https://www.nist.gov/news-events/news/2025/08/lessons-learned-consortium-tool-use-agent-systems). 2025-08-07.
[^daydream]: Daydream. [FAQ](https://daydream.ing/faqs). Checked October 10, 2026.
[^scouts]: Yutori. [Building the Proactive, Multi-Agent Architecture Powering Scouts](https://yutori.com/blog/building-the-proactive-multi-agent-architecture-powering-scouts), December 22, 2025. Inspected October 10, 2026; a provider engineering account, not an independent reliability evaluation.
[^jo]: Jo. [Private AI agent, connected apps and routines](https://askjo.ai/). Checked October 10, 2026.
[^ohai]: Ohai. [Household manager and FAQ](https://www.ohai.ai/). Checked October 10, 2026.
[^mindtrip]: Mindtrip. [Travel assistant](https://mindtrip.ai/). Checked October 10, 2026.
[^duckbill]: Duckbill. [Pricing and what every plan includes](https://getduckbill.com/pricing). Checked October 10, 2026; AI preparation and human participation are explicit.
[^youmind]: YouMind. [Learning and creation workflow](https://youmind.com/). Checked October 10, 2026.
[^dir-eval-probes]: NIST, [Building Evaluation Probes into Agentic AI](https://www.nist.gov/programs-projects/building-evaluation-probes-agentic-ai). 2026-05-05.

[^broad-chatgpt]: OpenAI. [Introducing ChatGPT agent](https://openai.com/index/introducing-chatgpt-agent/). July 17, 2025 announcement; inspected October 10, 2026. Launch access and controls are dated, not a current entitlement guarantee.
[^broad-gemini]: Google. [Gemini with Google apps](https://gemini.google/overview/apps/). Undated living page; inspected October 10, 2026.
[^broad-spark]: Google. [Gemini Spark](https://gemini.google/overview/agent/spark/). Undated living page; inspected October 10, 2026.
[^broad-copilot]: Microsoft. [Copilot plans for individuals](https://www.microsoft.com/en-us/copilot/solutions/individuals). Undated living page; inspected October 10, 2026.
[^broad-apple]: Apple. [Apple Intelligence and Siri](https://www.apple.com/apple-intelligence/). Undated living page; inspected October 10, 2026.
[^broad-alexa]: Amazon. [Introducing Alexa+: skills, cost and availability](https://www.aboutamazon.com/news/devices/new-alexa-generative-artificial-intelligence). Undated living page; inspected October 10, 2026.
[^broad-qwen]: Alibaba Group. [Qwen App turns ecosystem services into executable capabilities](https://www.alibabagroup.com/en-US/document-1948497434959151104). January 15, 2026 announcement; inspected October 10, 2026. Public testing and invite-only beta refer to that announcement, not a verified current rollout.

[^finance-openai]: OpenAI. [Accelerating the next phase of AI](https://openai.com/index/accelerating-the-next-phase-ai/). 2026-03-31; inspected October 11, 2026. Committed capital is not a statement that all cash has been received.

[^finance-alphabet]: Alphabet. [Investor relations](https://abc.xyz/investor/). Undated; inspected October 11, 2026. Supports listed-parent context, not separately financed AI products.

[^finance-microsoft]: Microsoft. [Investor relations](https://www.microsoft.com/en-us/investor/default). Undated; inspected October 11, 2026. Supports listed-parent context, not separately financed AI products.

[^finance-apple]: Apple. [Investor relations](https://investor.apple.com/investor-relations/default.aspx). Undated; inspected October 11, 2026. Supports listed-parent context, not separately financed AI products.

[^finance-amazon]: Amazon. [Investor relations](https://ir.aboutamazon.com/overview/default.aspx). Undated; inspected October 11, 2026. Supports listed-parent context, not separately financed AI products.

[^finance-alibaba]: Alibaba Group. [Investor relations](https://www.alibabagroup.com/en-US/investor-relations). Undated; inspected October 11, 2026. Supports listed-parent context, not separately financed AI products.

[^finance-duckbill-founder]: Emi Gonzalez, Duckbill cofounder. [Founder profile and financing disclosure](https://emigc.com/). Undated; inspected October 11, 2026. Undated cumulative founder disclosure. Does not identify a new round or its closing date.

[^finance-littlebird-facts]: Littlebird. [Company facts](https://littlebird.ai/facts). Undated; inspected October 11, 2026. Newer explicit seed label complements the March announcement, which did not name a series.

[^finance-youmind-conflict]: Rime research via Eastmoney. [July 18, 2025 venture bulletin](https://pdf.dfcfw.com/pdf/H3_AP202507181711069007_1.pdf?1752858046000.pdf=). 2025-07-18; inspected October 11, 2026. Contrasts with other same-period angel labels; insufficient primary evidence to resolve current round/stage. Used only to disclose the uncertainty.

[^finance-manus]: Axios. [Manus financing update](https://www.axios.com/newsletters/axios-pro-rata-ef74e379-7b84-4591-85a4-2420bb1f47de). 2026-10-08; inspected October 11, 2026. Named stage not supplied. Funding does not establish restored product access.

[^finance-instinct]: TechCrunch, citing The Wall Street Journal. [Instinct Series B report](https://techcrunch.com/2026/08/26/viral-ai-startup-instinct-has-raised-350-million-at-a-2-5-billion-valuation/). 2026-08-26; inspected October 11, 2026. Supersedes the prior Series A / further-talks presentation.

[^finance-youmind-angel]: Huayuan Securities. [July 2025 industry weekly](https://pdf.dfcfw.com/pdf/H3_AP202507211712851939_1.pdf). July 21, 2025; financing table inspected October 11, 2026. Labels the July YouMind event angel; the Rime bulletin calls it Pre-A. Neither is company confirmation of its current stage.

Source: https://latentlit.com/products-markets/landscapes/personal-agents/index.mdx
