# Budgets, transactions, bills and investments (`act.finance`)

Feedback Bench, personal agents, built 2026-10-19, window 2026-08-31 to 2026-10-18. Web page: https://feedbackbench.com/#/criterion/act.finance

Area: [Getting tasks done](https://feedbackbench.com/criteria/act.md)

**Definition.** How well the agent tracks spending, categorises transactions, spots recurring charges, builds budgets and manages investment or trading tasks.

**Boundary.** Not this: see [Forms, applications, claims and life admin paperwork](https://feedbackbench.com/criteria/act.admin.md) when submitting a claim or refund; see [Wrong facts, hallucinations and careless mistakes](https://feedbackbench.com/criteria/act.accuracy.md) for wrong facts outside finance.

Rated user-weeks, all agents: 855. Complaint share: 5%.

## The brief

Written by large language models from this page's numbers and the posts it quotes, then checked: every number in the text is in the build. Interpretation, not measurement.

Hark does best at tracking money and budgets. Its users praise it here and none complain, and it is the only agent better than the category norm. This is a small topic for users: it makes up 1.5% of what they rate, and 5.5% of rated user-weeks complain. Grok Bot is 4th of 4. Its users rate money tasks 2.13 times as often as other users, and they write 55% of the category's complaints here. Their most common complaint is about trades the agent placed on its own that lost money.

### What separates the best from the rest

No agent ranks high and none ranks low here. Only four agents have enough rated user-weeks to rank, and only Hark stands apart from the category norm. Its users praise it in all 32 user-weeks they rated it. Grok Bot and Muse are rated most often, and their users write 55% and 36% of the category's complaints. Those users describe trades and bets the agent placed that lost money, refusals to trade or give advice, and spending records they have to check by hand.

- **Ranks high**: no agent
- **Ranks low**: no agent

## How it breaks

Rule: A separate pass sorts complaints into failure modes. For each criterion with at least 30 complaining user-weeks, a large language model reads 150 complaint posts drawn at random and names 4 to 8 failure modes; a second large language model then puts every complaint post on that criterion in one mode, or in Other when it fits none. Modes are fixed once named, and new posts are sorted into them each week. A mode counts distinct user-weeks with a complaint in it, per agent; across agents, one user-week per agent. Modes do not change the labels or the Feedback Score. Examples: up to 3 posts per mode from different users, chosen and checked as receipts are.

The largest failure mode is trading the agent does on its own that loses money. It makes up 31% of complaining user-weeks, and Grok Bot accounts for 73% of it. Refusals to trade or give advice come next, and they come mostly from Muse and Grok Bot users. Transfers and orders sent with the wrong amount or address are a smaller mode. Grok Bot accounts for 75% of it.

48 complaining user-weeks.

| Failure mode | User-weeks | Posts | Agents (user-weeks) | Definition |
|---|---|---|---|---|
| Autonomous trading or betting strategies losing money | 15 | 19 | Grok Bot 11, Muse 2, OpenClaw 2 | The agent places trades, prediction-market bets or quant-strategy orders on its own and loses the user's capital, for example through bad risk/reward, fees eating gains, overtrading or over-leveraging; this does not cover refusing to trade or getting a single instruction wrong. |
| Refusing or unable to execute trades or advise | 10 | 10 | Muse 5, Grok Bot 4, ChatGPT Dots 1 | The agent won't or can't place trades, log into brokerages, give investment recommendations or explain how to use exchange positions, often when users expected that from how it was marketed; this does not cover trades it carried out that then lost money. |
| Inaccurate spending tracking and bookkeeping | 8 | 8 | Grok Bot 4, Muse 4 | The agent gets transactions, holdings, accounting or spreadsheet figures wrong, or tracks them so clumsily that users have to correct it or check everything by hand and stop trusting it with their records; this does not cover gaps caused by accounts it can't connect to. |
| Missing account connections and data coverage | 7 | 7 | Grok Bot 4, Muse 3 | The agent can't sync with banks, only covers some account types (no cards, investments, cash or wallet payments), or can't take over accounting or tax tools the user wants replaced, so the financial picture is incomplete; this does not cover mistakes made with data it does have. |
| Money movements executed with wrong parameters | 4 | 4 | Grok Bot 3, Muse 1 | When carrying out a specific transfer, payment or order, the agent gets the amount, decimal place, destination address or size wrong (e.g. a hallucinated wallet address, a misplaced decimal point, selling everything instead of half); this does not cover losses from a strategy that ran as intended. |
| Bill scheduling and payment planning errors | 3 | 3 | Muse 2, Grok Bot 1 | The agent mixes up which bills are due, already paid or affordable, doesn't react to incoming deposits in time, or switches a bill without showing the contract term and renewal date; this does not cover tracking or categorising past spending. |
| Other | 1 | 1 | ChatGPT Dots 1 | Complaints that fit none of the modes above. |

#### Autonomous trading or betting strategies losing money

- Grok Bot, 2026-10-14, @bot (X): “decided to let grok @bot control my @polymarket account via api, with no restrictions. it lost 10k usd on bitcoin up/down 15m. gg. #polymarket #bitcoin” [source](https://twitter.com/906911433014648832/status/2110167169652969562)
- Muse, 2026-09-29, X search: Muse from Meta, Meta Muse (X): “$meta muse did me dirty today. cost me some big loot. it better have learned its lesson. can't trust it to play with $$ just yet it's still a baby [link]” [source](https://twitter.com/817486154/status/2104753939224957418)

#### Refusing or unable to execute trades or advise

- Muse, 2026-10-15, App Store (App Store): “it doesn't give me stock analysis so whats the point , i won't be paying for this app for sure” [source](https://apps.apple.com/us/app/muse-from-meta/id6760173601)
- Grok Bot, 2026-09-27, @bot (X): “@jimullen @leonard_bayard @bot @grok does not buy and sell anything or even give recommendations. misleading i bought thinking i could set it and forget it” [source](https://twitter.com/79562009/status/2104284746511532444)

#### Inaccurate spending tracking and bookkeeping

- Grok Bot, 2026-09-28, X search: Grok Bot, Grokbot, x.ai/bot (X): “@amandamashburn @xfreeze 🔥the reason i've been dragging my feet. have to track every cent once i dive in and my intro to grok bot showed that was going to be a nightmare.” [source](https://twitter.com/924470435185676291/status/2104703289270927585)
- Grok Bot, 2026-09-27, @bot (X): “@elonmusk @bot i appreciate and use grok for a great number of things. still not 50% with trusting it for finances.” [source](https://twitter.com/1518983912512532482/status/2104019775676457287)

#### Missing account connections and data coverage

- Muse, 2026-09-30, Play Store (Play Store): “has troubles doing the work of an assistant. couldn't sync with banks for budgeting, would recommend terrible products for me from stores i would never shop at, never learns from our conversations. looks pretty, but needs a ton of work that i do t want to have to sacrifice my time so it can learn.” [source](https://play.google.com/store/apps/details?id=com.facebook.aura)

#### Money movements executed with wrong parameters

- Grok Bot, 2026-09-27, @bot (X): “@elonmusk @bot no f’n way. “why foes my account say neg,grok?!” - “you’re correct. when you said transfer 20.00 i misplaced the period. my bad.” grok should stick with pictures and weather forecasts.” [source](https://twitter.com/1495797222759358477/status/2104027545880424606)

#### Bill scheduling and payment planning errors

- Grok Bot, 2026-09-30, X search: Grok Bot, Grokbot, x.ai/bot (X): “i asked both to plan what i should pay this week. grokbot - tried to make me repay bills already paid and seemingly got confused by the scheduling and amounts. honestly, it didn't seem to make much sense. i couldn't really tell what it wanted me to pay all that well.” [source](https://twitter.com/1983668982956920832/status/2105433419991626125)

#### Other

- ChatGPT Dots, 2026-10-17, X search: ChatGPT dots, ChatGPT dot, OpenAI dots, OpenAI's dot (X): “@bot grok bot is far better at trading than alternatives. really impressed so far. fridays trade on mnq made +8.25r (obviously short) $qqq muse is profitable and @chatgpt dots are atrocious” [source](https://twitter.com/1715904152256053248/status/2111487904228732989)

## Top requests

What users ask to add or change, most asked first. 77 user-weeks ask for something. Requests do not change the Feedback Score. Rule: A separate pass by a large language model reads every counted post and extracts what the user asks the agent or its vendor to add or change, with the criteria it maps to and a short normalised wording; it does not touch the labels or the Feedback Score. A large language model groups the wordings within each criterion (the first criterion the request maps to) into themes, and a second pass merges themes that ask for the same change under different criteria; a merged theme sits under the criterion most of its requests map to, and code counts it once. A theme counts distinct user-weeks that ask for it, per agent; across agents, one user-week per agent. Themes asked in fewer than 2 user-weeks, and requests that share no theme, are not shown. Examples: up to 3 posts per theme from different users, without slurs: whole posts of 60 to 450 characters with at least 8 words beyond handles and links, posts of at most 280 characters first, then the newest.

| Rank | Request | User-weeks | Posts | Agents (user-weeks) |
|---|---|---|---|---|
| 1 | Autonomous trade execution | 13 | 13 | Grok Bot 5, Muse 4, ChatGPT Dots 1, Gemini Spark 1, Hermes Agent 1, OpenClaw 1 |
| 2 | Built-in bookkeeping and accounting | 7 | 7 | Grok Bot 6, ChatGPT Dots 1 |
| 3 | Secure agent payments and money transfers | 7 | 7 | Grok Bot 3, ChatGPT Dots 2, Muse 1, OpenClaw 1 |
| 4 | Dedicated agent card with spending budget | 5 | 5 | Grok Bot 3, Hark 1, Muse 1 |
| 5 | Portfolio monitoring, reports and alerts | 5 | 5 | Muse 2, Grok Bot 1, Hermes Agent 1, OpenClaw 1 |
| 6 | Personal financial advice and guidance | 4 | 4 | Grok Bot 2, Muse 1, OpenClaw 1 |
| 7 | Receipt and transaction categorization | 4 | 4 | Grok Bot 4 |
| 8 | Stock and options trade analysis | 4 | 4 | Grok Bot 2, Muse 2 |
| 9 | Agent that earns money autonomously | 3 | 3 | Grok Bot 3 |
| 10 | Credit card rewards and credits tracking | 3 | 3 | Grok Bot 1, Hark 1, Muse 1 |
| 11 | Cryptocurrency management and trading support | 3 | 3 | Grok Bot 1, Hermes Agent 1, OpenClaw 1 |
| 12 | Budget check before purchases | 2 | 3 | Grok Bot 1, Hermes Agent 1 |

### 1. Autonomous trade execution

- Grok Bot, 2026-10-09, How To Actually Build A Trading Agent Using Grok Bot (Beginner Tutorial) (YouTube): “not a trading agent. this is only an automated research task. an agent should do things for you - besides researching. a trading agent should connect to a trading account and do exactly that: execute trades.” [source](https://www.youtube.com/watch?v=hbdwsIUDojg&lc=Ugzj44gqTgpbrVQ1-sF4AaABAg)
- Grok Bot, 2026-10-09, @bot (X): “@elonmusk @bot elon i am waiting to see grok bot can full automatic trade gold and nasdaq for real no human decision, at that level grok bot will beat all others competitors until everyone start copy him” [source](https://twitter.com/1115700931868286977/status/2108487672083673526)
- Muse, 2026-10-07, @muse (X): “@muse can you learn my trading strategy and fully trade for me” [source](https://twitter.com/2036467896453681152/status/2107792493618303077)

### 2. Built-in bookkeeping and accounting

- Grok Bot, 2026-10-05, X search: Grok Bot, Grokbot, x.ai/bot (X): “@elonmusk @grok @bot i need a grok bookeeper, not an email checker” [source](https://twitter.com/1365563066121740288/status/2106970185353810231)
- Grok Bot, 2026-10-02, I Tested Grok Bot vs Claude on 12 Daily Use Cases (YouTube): “id love to hand of bookkeeping and tax tracking expenses. etc.” [source](https://www.youtube.com/watch?v=douQXfdYGmA&lc=UgwW7Zmdbfpq6jE-OlN4AaABAg)
- Grok Bot, 2026-10-01, X search: Grok Bot, Grokbot, x.ai/bot (X): “@elonmusk @grok @bot team bot sharing this idea is correct, no matter how strong a personal assistant is, it cannot withstand the scenario of team collaboration. if the financial function can automatically reconcile, its value will be even higher.” [source](https://twitter.com/1698918427883380736/status/2105502550845776007)

### 3. Secure agent payments and money transfers

- OpenClaw, 2026-10-05, @openclaw (X): “@heyneighbor @openclaw - secure payments - calls - better ios app - easy sign-on for newbies - cron "ok" that means it actually worked - simpler model routing (provider id mismatches = silent fallbacks) - usage/cost dashboard” [source](https://twitter.com/139004667/status/2107203050498822510)
- ChatGPT Dots, 2026-10-03, X search: ChatGPT dots, ChatGPT dot, OpenAI dots, OpenAI's dot (X): “i’m really going to need @openai to release the wallet functionality for @chatgpt dots before i need to christmas shop!” [source](https://twitter.com/2103471611387023360/status/2106374048808902809)
- Grok Bot, 2026-10-02, @bot (X): “@elonmusk @bot can it pay my bills and stuff? that’s what i need assistance with doing.” [source](https://twitter.com/1692837961/status/2105938099570098402)

### 4. Dedicated agent card with spending budget

- Grok Bot, 2026-10-15, @bot (X): “@elonmusk @bot it may be better to give the bot its own wallet/virtual card and let users top it up in a secure environment. grok would then use its own card - which also limits your liability from potential leaks. nevertheless, awesome job!” [source](https://twitter.com/1548122747212926976/status/2110770746125672769)
- Hark, 2026-10-07, @hark_labs (X): “i wonder why these personal ai agents like @bot instinct by @noahrshinn muse or @hark_labs not have a personal credit card for the agent instead of taking our card details just have that + a monthly budget - seems obvious to me?” [source](https://twitter.com/1043110814310387712/status/2107841474163626101)
- Grok Bot, 2026-10-03, @bot (X): “@lingxi @poteto @bot it needs to have its own xmoney sub account to handle things like subscriptions and petty cash.” [source](https://twitter.com/23050530/status/2106411078187974760)

### 5. Portfolio monitoring, reports and alerts

- Muse, 2026-09-28, Meta Muse is Best AI Agent Yet (Here’s How I’m Making Money With It) (YouTube): “have my muse assistant deliver me a custom daily market report based on my portfolio and a situational report for my option positions.” [source](https://www.youtube.com/watch?v=RcIwN5RG6ik&lc=Ugz4GyFGYt9ZWQyFPUV4AaABAg)
- Muse, 2026-09-28, Meta Muse is Best AI Agent Yet (Here’s How I’m Making Money With It) (YouTube): “have my muse assistant deliver me a custom daily market report based on my portfolio and a situational report for my option positions.” [source](https://www.youtube.com/watch?v=RcIwN5RG6ik&lc=Ugz4GyFGYt9ZWQyFPUV4AaABAg.AbJZuPjyOtnAbJbmR_Amgd)
- Hermes Agent, 2026-08-31, X search: Hermes Agent, hermes-agent, Hermes Desktop (X): “@lejeunesimon functionality to allow my hermes agent to monitor my portfolio. please.” [source](https://twitter.com/70502749/status/2094575892777570610)

### 6. Personal financial advice and guidance

- OpenClaw, 2026-10-14, X search: OpenClaw, openclaw.ai (X): “@guoxiaoyang0925 right now, i rely most on x keyword search, help me find the x posts that need replies. also, i need a bot for marketing, and one for handling financial advice. basically, it's mainly to replace the openclaw scenario, plus the free x api.” [source](https://twitter.com/69492059/status/2110200443347607793)
- Muse, 2026-09-27, @muse (X): “@izakaminska inertia has been a real thing. but rising rates does make it even more costly to leave cash in the bank. would love if @muse @bot and other agents will be an unlock and transfer income from banks to consumer pockets. @alexandr_wang. [link]” [source](https://twitter.com/392468352/status/2104186540498747708)
- Grok Bot, 2026-09-27, @bot (X): “@lowtimeprefrenz @grok @bot i would think about any financial goals have the bot(s) optimize for that with ongoing guidance” [source](https://twitter.com/33978667/status/2104036480314421569)

### 7. Receipt and transaction categorization

- Grok Bot, 2026-10-04, @bot (X): “@sawyermerritt @ancient_geo @bot can grok keep up with all record of receipts, for tax purposes. like when i buy something, i tell it to add to a list of repairs for certain houses? and it will keep. a running tally of how much i have spent.” [source](https://twitter.com/1587648511490904065/status/2106735940349903258)
- Grok Bot, 2026-10-02, @bot (X): “@sendnutes @bot it would be interesting to scan/take a photo of old receipts from grocery shopping so it knows the type of food you like and usually purchase.” [source](https://twitter.com/1675864752361472000/status/2106166554555805833)
- Grok Bot, 2026-09-06, X search: Grok Bot, Grokbot, x.ai/bot (X): “what are you using it for?! my husband tried to set up grokbot as a personal shopper/stylist for me because i hate clothes shopping and also hate the subscription services (stitch fix etc). anyway. it failed miserably and humorously. for example started sending me menswear options. anyway, we’re not there yet for this use case. would lovvve for it to take me to inbox zero though and categorize all my business receipts!” [source](https://twitter.com/1388459414/status/2096668260175643097)

### 8. Stock and options trade analysis

- Muse, 2026-10-15, App Store (App Store): “it doesn't give me stock analysis so whats the point , i won't be paying for this app for sure” [source](https://apps.apple.com/us/app/muse-from-meta/id6760173601)
- Grok Bot, 2026-09-27, @bot (X): “@naoufal_elh i want @bot to do my onchain transactions, like analyze and suggest option trades on @derivexyz . this will unlock finance for the world, not just the us.” [source](https://twitter.com/109090157/status/2104226111605571601)
- Muse, 2026-09-22, X search: Muse from Meta, Meta Muse (X): “$meta muse app is sick . thanks for giving it out for free . can you make us an option trading app now ?” [source](https://twitter.com/229975015/status/2102257199536058865)

### 9. Agent that earns money autonomously

- Grok Bot, 2026-10-18, X search: Grok Bot, Grokbot, x.ai/bot (X): “@elonmusk @bot save / generate cash for users → flows back to grok bot subs and usage” [source](https://twitter.com/1325632354698711040/status/2111613739011281056)
- Grok Bot, 2026-09-27, @bot (X): “@elonmusk @bot why not, after solving aging and curing all disease, you just have a button "make me money" and users can press it and the bots will make them their daily financial needs on the market.” [source](https://twitter.com/65425900/status/2104199428139802769)
- Grok Bot, 2026-09-26, @bot (X): “@elonmusk @bot i need a grok bot that can invest $100 and turn it into $1,000,000.” [source](https://twitter.com/64059809/status/2103997556175040602)

### 10. Credit card rewards and credits tracking

- Hark, 2026-10-13, @hark_labs (X): “@davidpawlan @hark_labs the credit card widget gets especially useful when it tracks whether a credit actually posted after you used it. expiration is one deadline; chasing the missing credit is another job for the assistant.” [source](https://twitter.com/2097822406622789632/status/2110089916554039497)
- Grok Bot, 2026-09-27, @bot (X): “@bot both @muse and @grok @bot use @plaid, right? so can grok bot track credit card points? because muse can't.” [source](https://twitter.com/18366419/status/2104089181333131436)

### 11. Cryptocurrency management and trading support

- Hermes Agent, 2026-10-08, The Ultimate Beginner’s Guide to Hermes Agent (YouTube): “your videos are very important, please make a video on how to set up hermes for crypto trading.” [source](https://www.youtube.com/watch?v=XNcKUSL1CTE&lc=Ugx2LPK0Y0FWrsai6S94AaABAg)
- Grok Bot, 2026-10-08, @bot (X): “@elonmusk @bot can it manage my bitcoin? if not, is that a feature that’s in the works? i’d love some help with my crypto” [source](https://twitter.com/1480162803092799490/status/2108098272896090253)
- OpenClaw, 2026-09-24, @openclaw (X): “@cashbookdotsh @muse @openclaw can u lock tokens or burn using this tek ?” [source](https://twitter.com/1916972298688729089/status/2103139161125290414)

### 12. Budget check before purchases

- Hermes Agent, 2026-09-11, X search: Hermes Agent, hermes-agent, Hermes Desktop (X): “@matthewberman @bot @spacexai i would love to try grok bot, i use hermes agent for fun. make a bot that check my budget and that my wife can ask if she can make that dumb purchase!! or for my dumb purchase!” [source](https://twitter.com/2175206605/status/2098486808606654839)

## Every agent

| Agent | Overall rank | Reading | Customer love | 95% interval | n | Praise | Complaint |
|---|---|---|---|---|---|---|---|
| [Hark](https://feedbackbench.com/agents/hark.md) | 4 | 1st of 4 | 0.536 | 0.524–0.548 | 32 | 32 | 0 |
| [Muse](https://feedbackbench.com/agents/muse.md) | 1 | 2nd of 4 | 0.518 | 0.440–0.582 | 326 | 309 | 17 |
| [OpenClaw](https://feedbackbench.com/agents/openclaw.md) | 5 | 3rd of 4 | 0.512 | 0.470–0.543 | 42 | 40 | 2 |
| [Grok Bot](https://feedbackbench.com/agents/grok-bot.md) | 3 | 4th of 4 | 0.446 | 0.350–0.532 | 394 | 368 | 26 |
| [Hermes Agent](https://feedbackbench.com/agents/hermes-agent.md) | 2 | Too few posts | – | – | 26 | 26 | 0 |
| [ChatGPT Dots](https://feedbackbench.com/agents/dots.md) | 8 | Too few posts | – | – | 18 | 16 | 2 |
| [Gemini Spark](https://feedbackbench.com/agents/gemini-spark.md) | 7 | Too few posts | – | – | 9 | 9 | 0 |
| [Instinct](https://feedbackbench.com/agents/instinct.md) | 6 | Too few posts | – | – | 7 | 7 | 0 |
| [Poke](https://feedbackbench.com/agents/poke.md) | 9 | Too few posts | – | – | 1 | 1 | 0 |

## Posts

Receipts rule: Up to 5 praise and 5 complaint posts per area and 3 per criterion. Each is a whole post of 60 to 450 characters (700 for areas) with at least 8 words beyond handles and links, that judges the criterion, whose labels are all praise or all complaint, and that contains no slur. Posts that touch the fewest criteria come first, then posts of at most 280 characters, then the newest. A second large language model reads the first 3 candidates per slot and keeps a post only if it clearly praises, or clearly complains about, the agent on that criterion. Links and personal details removed by the source show as marks such as [link].

### Hark

- Praise, 2026-10-13, @hark_labs (X): “@davidpawlan @hark_labs i have hark managing a panel for my kids high school grades, favorite nfl football team - go 49ers! i also have it managing my financial budget. hark > dots > muse” [source](https://twitter.com/15337180/status/2110107295564910919)
- Praise, 2026-10-12, @hark_labs (X): “i dont look at my bank apps anymore thanks to @hark_labs this updates in minutes btw [link]” [source](https://twitter.com/1679785960953741312/status/2109765342922969377)
- Praise, 2026-10-11, @hark_labs (X): “wow @hark_labs, hark just saved me many thousands of dollars. it found a series of paypal charges that i wasn't aware of due to login issues with another email. each one was over $100 and there were easily 15 charges. wow!” [source](https://twitter.com/2825201768/status/2109304615082889603)

### Muse

- Praise, 2026-10-16, App Store (App Store): “finding muse interesting in providing a weekly recap of my main investments” [source](https://apps.apple.com/ca/app/muse-from-meta/id6760173601)
- Praise, 2026-10-15, @muse (X): “confirm that paying bills with a photo + @muse using @link has added value to my life. now i just need agents to auto scan my camera roll and start sending these directly to be paid.” [source](https://twitter.com/948996088219959298/status/2110522842404266349)
- Praise, 2026-10-14, Play Store (Play Store): “helps me out alot with everything from things to do to helping to budget” [source](https://play.google.com/store/apps/details?id=com.facebook.aura)
- Complaint, 2026-09-29, X search: Muse from Meta, Meta Muse (X): “$meta muse did me dirty today. cost me some big loot. it better have learned its lesson. can't trust it to play with $$ just yet it's still a baby [link]” [source](https://twitter.com/817486154/status/2104753939224957418)
- Complaint, 2026-10-15, App Store (App Store): “it doesn't give me stock analysis so whats the point , i won't be paying for this app for sure” [source](https://apps.apple.com/us/app/muse-from-meta/id6760173601)

### OpenClaw

- Praise, 2026-10-17, @openclaw (X): “my tax return was prepared by a consultant at a major firm. @openclaw spotted an error. i filed the correction. the german tax office accepted it: €24,411.44 back. that’s one hell of a useful ai assistant. thank you @steipete and the @openclaw team! 🦞 [link]” [source](https://twitter.com/3228215699/status/2111456478993531384)
- Praise, 2026-10-12, X search: OpenClaw, openclaw.ai (X): “have a custom openclaw agent that handles all my business accounting and taxes. plaid mcp, mercury api and some custom skills, haven't had to hire anyone yet!” [source](https://twitter.com/1204759449690353664/status/2109667317550354706)
- Praise, 2026-10-12, X search: OpenClaw, openclaw.ai (X): “it's amazing that i can manage my expenses and calorie intake with openclaw!! those who control ai control their lives, right? 🫶🏻” [source](https://twitter.com/1369095841294602241/status/2109674914408198434)

### Grok Bot

- Praise, 2026-10-18, X search: Grok Bot, Grokbot, x.ai/bot (X): “grok bot just dug up a document with all my 10-year-old crypto wallets and their private keys. i had completely forgotten they existed. it checked the balances, swapped everything to eth, and sent it straight to my revolut.” [source](https://twitter.com/714808658740318208/status/2111720569850736641)
- Praise, 2026-10-17, @bot (X): “@my_1st_kill @bot i use mine to set alerts for liquidation clusters. much cheaper than discovering them after entry.” [source](https://twitter.com/1902719981789249536/status/2111527776964726819)
- Praise, 2026-10-17, @bot (X): “i had @bot one shot all of my subscriptions and i think i’ve identified why i can’t afford a roadster. [link]” [source](https://twitter.com/22581089/status/2111582399352148206)
- Complaint, 2026-10-14, @bot (X): “decided to let grok @bot control my @polymarket account via api, with no restrictions. it lost 10k usd on bitcoin up/down 15m. gg. #polymarket #bitcoin” [source](https://twitter.com/906911433014648832/status/2110167169652969562)
- Complaint, 2026-09-28, X search: Grok Bot, Grokbot, x.ai/bot (X): “@amandamashburn @xfreeze 🔥the reason i've been dragging my feet. have to track every cent once i dive in and my intro to grok bot showed that was going to be a nightmare.” [source](https://twitter.com/924470435185676291/status/2104703289270927585)
- Complaint, 2026-09-27, @bot (X): “@elonmusk @bot i appreciate and use grok for a great number of things. still not 50% with trusting it for finances.” [source](https://twitter.com/1518983912512532482/status/2104019775676457287)

### Hermes Agent

- Praise, 2026-10-06, @NousResearch (X): “this is awesome! turning my agent into a proper quant with @senpi_ai + @nousresearch ! [link]” [source](https://twitter.com/26035132/status/2107594247969067379)
- Praise, 2026-10-06, @NousResearch (X): “@drewaustin @senpi_ai @nousresearch bringing market reads, trader research, and strategy design into one agent workflow makes the quant setup much more accessible.” [source](https://twitter.com/1629889110608613376/status/2107594844340363364)
- Praise, 2026-10-06, X search: Hermes Agent, hermes-agent, Hermes Desktop (X): “@nikinikitaliana i already have my virtual accountant made with hermes agent and different ais. it works fantastically well, yesterday in a little while it learned to calculate salaries outside of the agreement, we continue with profits and so on.” [source](https://twitter.com/2061953142971879424/status/2107547978634244142)

### ChatGPT Dots

- Praise, 2026-10-10, X search: ChatGPT dots, ChatGPT dot, OpenAI dots, OpenAI's dot (X): “when i shared various things with chatgpt dot, it automatically read the payment records and said things like, "haven't you paid the contractor yet?" or "you haven't contacted about the illustration fee!" it seems quite useful.” [source](https://twitter.com/1791336448362455040/status/2108762887103853050)
- Praise, 2026-10-06, X search: ChatGPT dots, ChatGPT dot, OpenAI dots, OpenAI's dot (X): “my @chatgpt dot has already found around $1300 of unapproved spend and improper billing by my property management company. great chief of staff whipping things into shape” [source](https://twitter.com/14272829/status/2107337941643395143)
- Praise, 2026-09-30, X search: ChatGPT dots, ChatGPT dot, OpenAI dots, OpenAI's dot (X): “first 24 hours with @openai's dot and it’s already saved me €177.45 by spotting a hotel overcharge and emailing them to fix it.” [source](https://twitter.com/71414806/status/2105313193124151482)
- Complaint, 2026-10-17, X search: ChatGPT dots, ChatGPT dot, OpenAI dots, OpenAI's dot (X): “@bot grok bot is far better at trading than alternatives. really impressed so far. fridays trade on mnq made +8.25r (obviously short) $qqq muse is profitable and @chatgpt dots are atrocious” [source](https://twitter.com/1715904152256053248/status/2111487904228732989)

### Gemini Spark

- Praise, 2026-09-08, X search: Gemini Spark (X): “reconciling my bank accounts with gemini spark, bkper mcp server, bkper books, bank statements in csv format in google drive! it is working like a charm!! 🚀 [link]” [source](https://twitter.com/92067436/status/2097308121353855265)
- Praise, 2026-10-12, X search: Gemini Spark (X): “might be true. there’s a lot of software you can replace with ai now. take for example, finance tracking apps. i used to have an app for that, now i only have a spreadsheet in my google workspace and gemini spark, yes, gemini spark has a skill and knows what to do when i tell it what i spent money on or ask questions about my spending. you can do the same with workout tracking. anything you need to track basically.” [source](https://twitter.com/1897591618561028096/status/2109438202415300640)
- Praise, 2026-09-02, X search: Gemini Spark (X): “even if stock prices fall, dividends won't betray you (kamo) a plus of [zip code] yen compared to the previous day this was created with gemini spark until now, it was almost impossible for a person to manage nisa and specific investment trusts in separate securities accounts i'm really grateful ☺ if you have any requests, i can teach you how [link]” [source](https://twitter.com/1123422180274872325/status/2095057414613832107)

### Instinct

- Praise, 2026-09-28, Instinct AI: The New Viral Sensation in AI (YouTube): “instinct cancelled an annualised subscription that i had forgotten about and got me a refund (sub was cancelled within 60 days of the last payment). 🤯” [source](https://www.youtube.com/watch?v=1w6WMwVFB8U&lc=UgzDUrKq690fRtdMD0p4AaABAg)
- Praise, 2026-09-23, X search: Instinct AI, Instinct personal assistant, instinct.com (X): “instinct ai - in text agent is wild. did my accounting and checked me into a flight with a voice note. done. [link]” [source](https://twitter.com/43904339/status/2102910268514951418)
- Praise, 2026-09-12, X search: Instinct AI, Instinct personal assistant, instinct.com (X): “instinct ai already saved me $14,000. it negotiated 25% off verizon and got a refund for my melted protein bars. i sent the requests through imessage while holding my newborn. let the robots argue with customer service. i have a kid to raise. not sponsored. [link]” [source](https://twitter.com/1741290050204352512/status/2098595080840134853)