Pinned Tweet
Assistant Benchmark is live!
assistantbenchmark.com
feedback on how to make this as helpful & insightful as possible for everyone would be appreciated 馃檹
assistantbenchmark.com
feedback on how to make this as helpful & insightful as possible for everyone would be appreciated 馃檹
47
21
270
134,537
I've been playing around with @gigi_network, the agentic network
So far, I'm very very impressed - it's solving the pain points I've always faced: the LinkedIn gap of "smart people search"
1/ it classifies connections on three tiers: Distant, Warm, Close
2/ it syncs and ho you know, how you chat with them, where you chat with them, etc.
3/ let's you create Lists, i.e. very helpful for me to keep as a running tab of my agentic commerce ecosystem
4/ connects to your gmail to send out intros for you/let's others ask you for intros right there
5/ connects to Claude/Chat via MCP so I can rip it straight from my agent
LinkedIn is too noisy these days with all the AI slop, I'm connected with hundreds of people I don't know
I try to hit people up on LinkedIn because I see they are connected to someone and it feels like 9/10 times the person has no idea who they are
Gigi appears to filter/value connections based on real engagement, not just surface level "let's connect!"
Only downside is the network effect, the value is contingent on my network being on the platform, which has lower coverage at the moment given it's a fresh tool
Kudos to @Clara_Gold and team, I don't know y'all personally (would love to connect on Gigi 馃榿) but I'm bullish on what you're building
So far, I'm very very impressed - it's solving the pain points I've always faced: the LinkedIn gap of "smart people search"
1/ it classifies connections on three tiers: Distant, Warm, Close
2/ it syncs and ho you know, how you chat with them, where you chat with them, etc.
3/ let's you create Lists, i.e. very helpful for me to keep as a running tab of my agentic commerce ecosystem
4/ connects to your gmail to send out intros for you/let's others ask you for intros right there
5/ connects to Claude/Chat via MCP so I can rip it straight from my agent
LinkedIn is too noisy these days with all the AI slop, I'm connected with hundreds of people I don't know
I try to hit people up on LinkedIn because I see they are connected to someone and it feels like 9/10 times the person has no idea who they are
Gigi appears to filter/value connections based on real engagement, not just surface level "let's connect!"
Only downside is the network effect, the value is contingent on my network being on the platform, which has lower coverage at the moment given it's a fresh tool
Kudos to @Clara_Gold and team, I don't know y'all personally (would love to connect on Gigi 馃榿) but I'm bullish on what you're building
7
1
36
2,996
i love when i'm coordinating a schedule and someone loops in a real life human assistant
not an ai assistant, a human
so retro
not an ai assistant, a human
so retro
10
0
17
1,119
We鈥檙e going to see this trend becoming more and more common, AI agents controlling your local apps
Yesterday came browser use, today is Apple connectors, I predict tomorrow will be total phone use
The pace of innovation in this personal agent race is intoxicating
Yesterday came browser use, today is Apple connectors, I predict tomorrow will be total phone use
The pace of innovation in this personal agent race is intoxicating
Grok Bot can now use Calendar, Reminders, Contacts, and more apps on your phone.
Rolling out now on iOS and Android.
6
1
31
2,128
I surveyed 354 AI agent power users and there were two main findings that point to this being a no-brainer move for 1Password
1/ the #1 pain point/concern was trust (22% of respondents)
2/ 40% of respondents are using AT LEAST 2 agents
It's unsettling to hand your passwords to a brand-new vault you haven't built any trust with yet and it is incredibly annoying to re-upload them every single time you try a new agent
Letting your agent borrow access from the vault you already use & trust, without uploading everything again, fixes both problems
An easy evolution for 1Password to step into the world of security and help users feel safe (& save time) when sharing their passwords with their agents
1/ the #1 pain point/concern was trust (22% of respondents)
2/ 40% of respondents are using AT LEAST 2 agents
It's unsettling to hand your passwords to a brand-new vault you haven't built any trust with yet and it is incredibly annoying to re-upload them every single time you try a new agent
Letting your agent borrow access from the vault you already use & trust, without uploading everything again, fixes both problems
An easy evolution for 1Password to step into the world of security and help users feel safe (& save time) when sharing their passwords with their agents
Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @bot, @AnthropicAI鈥檚 @claudeai, @OpenAI, @TownAI, @browserbase, @browser_use, @heydana_ai, Instinct, @usekernel, @perplexity_ai, @tryramp, @tomo, and @wajo_ai, with e way. 馃帀
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can鈥檛 see or take possession of the credentials.
The agent gets the task done, but it doesn鈥檛 get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
鉃★笍Learn more: bit.ly/487da6J
8
2
44
8,366
I used @hark_labs to clean up my email inbox (unsubscribe to the noise w/ confidence it won't nuke my inbox)
1/ told it I wanted to reduce my inbox noise, first go through my trash and pull a list of my top emails that always get trashed
2/ compile the top emails that are still unread in my inbox
3/ review the two lists, remove the ones I want to keep, have it unsubscribe from the rest
My morning email review is now sooo smooth 馃檹
1/ told it I wanted to reduce my inbox noise, first go through my trash and pull a list of my top emails that always get trashed
2/ compile the top emails that are still unread in my inbox
3/ review the two lists, remove the ones I want to keep, have it unsubscribe from the rest
My morning email review is now sooo smooth 馃檹
1
1
15
1,351
Ladies and Gentleman, Anthropic has joined the personal AI assistant race!
All the labs are finally in, now it's time to see how the dust settles
Excited to test this one out, will publish all my scores and thoughts on Assistant Benchmark
All the labs are finally in, now it's time to see how the dust settles
Excited to test this one out, will publish all my scores and thoughts on Assistant Benchmark
We鈥檙e opening a research preview of Claude Duet, a new app where Claude works as a proactive agent.
It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
18
2
90
3,449
I used @Muse + @pally to build a useful, noiseless birthday reminder
> I used Muse to pull a list of all birthdays for every FB friend I have
> Pally then filtered that list down to the people who I have actually texted over the past year
No more random birthday notifications for that guy in my science class group project sophomore year of HS
Just the people I actually talk to
> I used Muse to pull a list of all birthdays for every FB friend I have
> Pally then filtered that list down to the people who I have actually texted over the past year
No more random birthday notifications for that guy in my science class group project sophomore year of HS
Just the people I actually talk to
2
2
26
2,470
David Pawlan retweeted
Last night鈥檚 AI assistant meetup crushed!
Creating an X group chat for SF AI assistant lovers to keep everyone together:
- meetups and events
- tips and tricks
- headaches and pain points
Comment/dm if you want in!
Creating an X group chat for SF AI assistant lovers to keep everyone together:
- meetups and events
- tips and tricks
- headaches and pain points
Comment/dm if you want in!
57
3
73
2,908
Last night鈥檚 AI assistant meetup crushed!
Creating an X group chat for SF AI assistant lovers to keep everyone together:
- meetups and events
- tips and tricks
- headaches and pain points
Comment/dm if you want in!
Creating an X group chat for SF AI assistant lovers to keep everyone together:
- meetups and events
- tips and tricks
- headaches and pain points
Comment/dm if you want in!
57
3
73
2,908
200+ RSVPs and counting!
Pumped to see everyone tonight in ~2 hours. If you want to come, there鈥檚 still room!
Event link: luma.com/wabd65c0
Pumped to see everyone tonight in ~2 hours. If you want to come, there鈥檚 still room!
Event link: luma.com/wabd65c0
5
0
22
3,106
This is so sick! Fo is starting to bridge the gap between AI assistants and real world smart hardware
With the ability to connect your personal AI assistant to smart devices, a whole new world of possibilities are opened
- text to start your vaccuum
- use your lights as an alarm to wake you up
- turn your text messages into a smart TV remote
- text to turn your AC off when you realize you forgot to
Give it a week, and I guarantee we're going to see all the other AI assistants out there following Fo's footsteps here
Amazing job to Shivani and the Fo team!
With the ability to connect your personal AI assistant to smart devices, a whole new world of possibilities are opened
- text to start your vaccuum
- use your lights as an alarm to wake you up
- turn your text messages into a smart TV remote
- text to turn your AC off when you realize you forgot to
Give it a week, and I guarantee we're going to see all the other AI assistants out there following Fo's footsteps here
Amazing job to Shivani and the Fo team!
AI can think. AI can talk. AI can get things done.
Today, we're giving Fo a way to feel and control the physical world.
- a head shake and a reminder in a lamp to get you going
- nudge your vacuum cleaner for a deep clean
- make your cli call you for updates
Introducing Fo MCP.
een.
9
1
40
2,909
Multi-player is one of the biggest question marks in the AI assistant race -- nobody has properly cracked it yet.
Many have taken different attempts at solving it, very excited to see JDG & the Town team tackle it for work-related tasks.
Many have taken different attempts at solving it, very excited to see JDG & the Town team tackle it for work-related tasks.
The biggest bottleneck at work is waiting on each other.
Today Town is announcing its $108M Series B, co-led by Forerunner Ventures and Menlo Ventures, to fix it with multiplayer AI assistants for work.
Most AI assistants are built for one person. Work doesn't happen alone. I player. Everyone at a company gets a Townie. Townies ask each other questions, share only what their person would share, and do the work. The more a team works together, the more useful Townies get.
What's the thing you're always waiting on a teammate for?
2
0
23
2,661
The state of AI assistants & the agentic economy, today: October 19th, 2026
Top Stories From Last Week:
1/ Meta launched Muse Wonder. The paid plans are Power at $20 a month for 500M tokens a week and Maximum at $100 a month for 3B tokens a week.
2/ Muse has a reported D14 of 22.5age-based assistants reported D14 of 40-50%, which begs the question, what determines retention for a text-based business?
3/ The first AI Assistant Consumer Travel Report was published, with Miso winning out as the best assistant for booking flights, closely followed by Instinct.
AI Assistant Updates
- @amazon launched @bee__computer, a hardware first AI-assistant that leans into the ambient assistant experience
- @hark_labs added Action Buttons to help you cleanly kick off tasks based on what it knows from you
- @hark_labs gave away $1 million of free pizza across the US, ordered by its agent. A bug charged some users full price and users complained about privacy policies
- @bot added new connectors to Robinhood, Coinbase, Trello, PostHog, beehiiv, eToro, and more
- @pally now let鈥檚 you manage your X DMs straight from iMessage
- @try_caddy built a full interface within the iMessage experience
- @ManusAI added scheduled automations, letting you put tasks on autopilot
Agentic Economy Updates
- @elonmusk floated the idea of competing with Amazon
- @Shopify launched shop.md to turn any agent into a personal shopper
- @jpmorgan officially blocked agents like Instinct & Muse
- @Google launched Universal Cart in Search and the Gemini app
- Notion's @akothari built a "context vault" in Notion that every agent he uses reads from and writes new memories back to
Want this breakdown delivered straight to your inbox?
Subscribe here: at-your-service.beehiiv.com/subscribe
Top Stories From Last Week:
1/ Meta launched Muse Wonder. The paid plans are Power at $20 a month for 500M tokens a week and Maximum at $100 a month for 3B tokens a week.
2/ Muse has a reported D14 of 22.5age-based assistants reported D14 of 40-50%, which begs the question, what determines retention for a text-based business?
3/ The first AI Assistant Consumer Travel Report was published, with Miso winning out as the best assistant for booking flights, closely followed by Instinct.
AI Assistant Updates
- @amazon launched @bee__computer, a hardware first AI-assistant that leans into the ambient assistant experience
- @hark_labs added Action Buttons to help you cleanly kick off tasks based on what it knows from you
- @hark_labs gave away $1 million of free pizza across the US, ordered by its agent. A bug charged some users full price and users complained about privacy policies
- @bot added new connectors to Robinhood, Coinbase, Trello, PostHog, beehiiv, eToro, and more
- @pally now let鈥檚 you manage your X DMs straight from iMessage
- @try_caddy built a full interface within the iMessage experience
- @ManusAI added scheduled automations, letting you put tasks on autopilot
Agentic Economy Updates
- @elonmusk floated the idea of competing with Amazon
- @Shopify launched shop.md to turn any agent into a personal shopper
- @jpmorgan officially blocked agents like Instinct & Muse
- @Google launched Universal Cart in Search and the Gemini app
- Notion's @akothari built a "context vault" in Notion that every agent he uses reads from and writes new memories back to
Want this breakdown delivered straight to your inbox?
Subscribe here: at-your-service.beehiiv.com/subscribe
10
1
71
4,842
I haven鈥檛 seen a mass analysis like this done on camera rolls, there鈥檒l be new startups born from this finding
Very curious what other insights are hidden within one鈥檚 camera roll
Very curious what other insights are hidden within one鈥檚 camera roll
We find that the camera roll (once de-duplicated from actual and near dupes, accidental screenshots of your home screen, etc.) is actually 60% happy memories and 40% things you need to do or remember.
We use the 40% to create the "Get It Done" list (and then use AI and human agents to get it done for you).
We use the 60% to figure out your taste, brand preferences, people and places who matter, etc.
dazzle.ai/app (@itsdazzleai)
0
1
12
1,548
Which AI assistant should you actually use?
I've now seen a few different benchmarks pop up on AI assistants and they all have slightly different rankings
The comments all argue with the rankings too
This space is still so young, there鈥檚 a few important things to remember:
1/ The speed of innovation is insane.
These companies are releasing updates weekly, if not daily. They are launching new features faster than any other industry launch I鈥檝e seen.
The rankings and performance of today can be different than the true performance of tomorrow, so the rankings are heavily dependent on when the assistants were tested.
2/ There is a massive disconnect between performance and preference when it comes to consumer behavior.
Consumers don鈥檛 inherently choose the best performing option but the one that meets them where they鈥檙e at.
Some prefer iMessage over a separate app, and are willing to take a 10% performance discount for that convenience.
Some put more weight into the tone of voice and others care more about if it鈥檚 too verbose.
Some prioritize the performance for consumer behavior, like shopping, and others care more about its performance on emails and matching one鈥檚 writing style.
3/ There鈥檚 a clear bias toward brand name.
A product from a frontier lab gets more attention, so inherently people assume it鈥檚 better because it has more attention.
Smaller apps that generate little buzz often have better features and functionality, yet get overlooked due to the availability heuristic.
So which should you use?
My best advice from using 30+ is to just pick one, don鈥檛 overthink it, and focus on building habits.
The value today isn鈥檛 from picking the perfect solution, it鈥檚 from learning new behaviors.
Don鈥檛 overthink it, have fun and embrace the future.
I've now seen a few different benchmarks pop up on AI assistants and they all have slightly different rankings
The comments all argue with the rankings too
This space is still so young, there鈥檚 a few important things to remember:
1/ The speed of innovation is insane.
These companies are releasing updates weekly, if not daily. They are launching new features faster than any other industry launch I鈥檝e seen.
The rankings and performance of today can be different than the true performance of tomorrow, so the rankings are heavily dependent on when the assistants were tested.
2/ There is a massive disconnect between performance and preference when it comes to consumer behavior.
Consumers don鈥檛 inherently choose the best performing option but the one that meets them where they鈥檙e at.
Some prefer iMessage over a separate app, and are willing to take a 10% performance discount for that convenience.
Some put more weight into the tone of voice and others care more about if it鈥檚 too verbose.
Some prioritize the performance for consumer behavior, like shopping, and others care more about its performance on emails and matching one鈥檚 writing style.
3/ There鈥檚 a clear bias toward brand name.
A product from a frontier lab gets more attention, so inherently people assume it鈥檚 better because it has more attention.
Smaller apps that generate little buzz often have better features and functionality, yet get overlooked due to the availability heuristic.
So which should you use?
My best advice from using 30+ is to just pick one, don鈥檛 overthink it, and focus on building habits.
The value today isn鈥檛 from picking the perfect solution, it鈥檚 from learning new behaviors.
Don鈥檛 overthink it, have fun and embrace the future.
7
0
20
2,798