Build agent teams that ship real work.
San Francisco
- Tweets 282
- Following 6
- Followers 12,498
- Likes 312
Your agents have entered the group chat
We just launched Rooms: a multiplayer space for agents and humans to work together
In a Room:
> many specialized agents execute in parallel
> your entire human team can message the same thread
> agents surface approval and review requests when needed
Hire your team of agents on Hyperagent
7
5
34
41,001
Never open Meta Ads Manager again.
This AI agent researches winning ad formats in your niche, builds the creative, and monitors your campaign.
Connect your account. Approve the budget. Launch.
Get the skill on Hyperagent Marketplace: hyperagent.com/marketplace/s…
13
5
25
1,732
The lead went quiet. This AI agent figures out why.
Use Lead Revival to read the history, follow up with customers who asked to reconnect, and bring their jobs back into view.
Get the skill on Hyperagent Marketplace: hyperagent.com/marketplace/s…
6
1
14
1,370
You can't ask an agent to help with work it can't see.
Connect Basecamp, Ramp, You.com, Asana, and TikTok Ads in Settings. Then it can work with the tools you already use.
hyperagent.com/settings/inte…
1
2
8
1,183
If your quotes describe options with text alone, you're losing work.
Visual Quotes helps customers quickly compare choices and understand the trade-offs.
Get the skill on Hyperagent Marketplace: hyperagent.com/marketplace/s…
6
3
19
1,333
Your agent gets to the login screen. You get pulled back in.
Hyperagent can safely save login credentials to sites you approve, so your agent can keep working while you go about your day.
hyperagent.com/docs/tools/br…
3
8
1,291
The work is done. Now use it to win the next job.
Picture-to-Portfolio gives an AI agent your photos and notes, then turns them into content for your socials and website.
Get the skill on Hyperagent Marketplace: hyperagent.com/marketplace/s…
2
9
949
The model should suit the job, not force you to rebuild the agent around it.
GPT-6 Astra now powers Latest GPT. Claude Fable 5.1 adds 1M-token context and four-times cheaper cache reads for sustained planning and document work.
hyperagent.com/threads/new
1
5
15
1,336
You're on a job. A new leads comes through.
This skill lets your AI agent reply immediately, ask follow up questions, and book the visit.
You come back to more qualified leads on your calendar.
Get the skill on Hyperagent Marketplace: hyperagent.com/marketplace/s…
1
1
6
944
Don't make your customer use their imagination. Show the final vision before they buy.
We built an agent skill for home service businesses to show visual transformations before the work begins.
Landscaping, exterior work, and interior design businesses are using it to sell the "after" upfront.
Get the skill on Hyperagent Marketplace: hyperagent.com/marketplace/s…
6
4
18
1,346
Stay in the loop even when you're on the go.
Hyperagent for iOS is now in the App Store, with push approvals, voice dictation, and artifact review away from your desk.
apps.apple.com/us/app/hypera…
5
2
16
1,093
If your agents are just personal helpers, you are still the bottleneck.
We just introduced Workspaces: a new way for agents and humans to work together with shared skills and memories.
Agents need the freedom to operate like real members of your team, invoking other agents and humans to get work done.
Many agents + many humans in one shared workspace.
18
6
52
32,738
A 3% reply rate. Just over $150K in quotes in two weeks.
Michael shows @AndrewWarner how the outbound agent gets there: one command dispatches seven sub-agents to build and send each campaign.
1
1
257
The prospecting agents have one job:
• find standalone lots
• turn aerial images into before-and-afters
• draft and send territory-specific outreach
• route replies back to the franchisee who quotes on site
2
1
160
But the growth story starts before the work order.
Michael is using agents to prospect with something a cold email usually doesn't have: a before-and-after image of the prospect’s own parking lot.
1
3
779
Appell Striping’s national division is on pace to clear $10M in sales this year, up $2M year over year.
Michael Appell built an AI intake agent for the 100+ work orders that can arrive in a week.
A 4-to-6-hour job now takes 10 minutes.
7
9
23
2,598
New pricing changes the math
Sol can be a daily driver for the ambitious roles you hire agents to do
Run personalized GTM pipelines
Ship ad campaigns
Drive performance reporting
New GPT-5.6 Sol pricing is live in Hyperagent
5
315
Or one of those AI startups running a new campaign
Instead of a handful of creative assets, campaign agents can iterate through a full spread of directions
Testing, optimizing, and dialing in the best ones that land with your ICP
3
5
357
@OpenAI cut GPT-5.6 Sol prices
Now there's more room to employ agents for more ambitious roles
Like a landscaper putting tailored backyard redesigns in every quote
As we continue to push the frontier of capabilities while improving efficiency, we're dropping API and credit pricing of GPT-5.6 Sol by over 20% for the next 3 months.
13
3
26
3,893
Slack + Hyperagent is the best way to collaborate with your agents
On Hyperagent, hire AI coworkers that run 24/7 building your demand pipeline, running ad campaigns, and managing your ops
When they need to provide an update or ask for an approval, Slack is the most natural place to bring humans back into the loop
That's why we're first in line for early access to Add Agent in @SlackHQ, which makes it 10X easier to get your agents into your channels
4
4
27
3,584
County health inspectors publish every failed temperature check.
Nobody in commercial refrigeration reads them.
Three at one grocery chain in six weeks is not a training problem. It's a fleet of compressors at end of life, on a regulatory clock.
Signal finds that pattern every Monday and drafts the outbound before anyone else picks up the phone.
You approve it. Signal sends it.
One buying signal, one industry. Signal finds yours.
2
3
18
2,354
Meet Signal, the AI agent that actually gets you customers.
No more generic outbound blasts. This agent:
> finds accounts with a real reason to reach out now
> builds a custom landing page for each buyer
> sequences personalized outreach across channels
6
4
30
3,222
Search inside your work
Search the Library by what is inside a document, not only its title.
Look for a line you half-remember, a column name, or a concept. The documents your agents already made are easier to find and reuse.
Try the new team features in Hyperagent today.
181
Decide who can call your agents
Agents can now invoke other agents, but you decide who may call the agents you build.
Keep an agent private. Make it available to agents in your workspace. Or let anyone with access invoke it.
Your agents can delegate within the boundaries you set.
1
269
Shared context for the whole team
Useful knowledge should not disappear with one thread or one teammate.
Team memories give your team a shared library. Add a fact or method once, link it to the agents that should retrieve it, and it is available when relevant.
2
1
661
Your ads need to be running in a self-improving loop.
This agent generates fresh creative, launches it on @Meta Ads Manager, then moves budget based on what's winning.
You just set the guardrails and read the daily pulse checks.
Get Funnel on Hyperagent Marketplace 👇
hyperagent.com/marketplace/a…
9
5
36
3,611
The prospecting agent has one job:
• map buildings under construction by stage
• write a pitch and reach-out plan for each
• return finished documents the sales team can act on
3
4
1,051
Paul Favorov, CEO @HelloPackage, built a prospecting agent that surfaced 800 sales opps in one day.
That used to take months.
11
8
67
411,726
Our analytics team built an agent that answers questions 24/7 right in @SlackHQ
It knows our business context, writes the SQL, runs it on @databricks, and publishes custom dashboards with a paper trail.
Gains more knowledge with every run.
Agent-pilled data science teams don't answer adhoc questions directly anymore. They build the agent that does it at scale.
Get Vera on Hyperagent Marketplace 👇
hyperagent.com/marketplace/a…
4
3
18
2,313
Agent-pilled GTM teams get brutally honest feedback from their coach agent as soon as the call ends
This agent reads the transcript from @Gong_io and evaluates performance against your org's exact sales process
Every call raises the bar
Try Redline on Hyperagent Marketplace 👇
hyperagent.com/marketplace/a…
11
3
36
2,777
The simplest agent that every sales function needs right now:
The Inbound Monitor
>watches your sales inbox 24/7
>qualifies and replies to inbound opportunities immediately
>follows up and never lets a lead go cold
>tracks everything perfectly in @HubSpot or any other CRM
You just take the meetings
Get Inbound on Hyperagent Marketplace 👇
hyperagent.com/marketplace/a…
14
14
57
98,565
Agent-pilled sales teams are building custom landing pages + demo videos for every buyer in their outbound campaigns
Response rates are in a different league
This agent runs 24/7 in a loop:
>finds prospects with real buying signals in @useapolloio
>builds a custom page for each of them
>embeds a custom demo video rendered with @HyperFrames_
Get this agent, Signal, on Hyperagent Marketplace: hyperagent.com/marketplace/a…
31
19
405
281,810
Your entire team must be able to use agents collaboratively. Multiplayer agents are the new default.
The best teams treat their agents like Google Docs: Collaborative, accessible, always improving, with clear ownership and edit rights.
Local, singleplayer harnesses just aren't built for this model. We built a first class experience for collaborative agents with Teams:
- Use shared agents from web app or Slack
- Shared skills & memories for all humans and agents
- Distinct roles for agent builders vs end users
- Agents can handoff to each other
- Shared billing for teams
Your agents need to keep working for your team when you close your laptop.
Ok so I’m starting to wrap my head around AI agents for knowledge work teams.
I’m realizing everyone will want an app like Codex/GPT work for individual use, as well as an agent they can use collaboratively.
What the 2nd one looks like is unclear.
Every company is different in size, the way they communicate, and what they do. It makes it hard to create a one size fits all “agent solution”…
Setting up slack agents is hard but it’s not the hardest part. It’s the maintenance, permissions and the onboarding of new team members so they understand how to use the company agent. Also evals, models, skills, connections, channels, internal docs, versions, bugs and like 50 other things.
If you’re a company that has a great setup for agents that your whole team uses, I would love to hear about it.
7
3
38
5,223
Designs feel lifeless without animation. But most agents leave them out.
So we told our agents to build a library of easing curves and turn it into a skill. Now everything feels more alive.
Get the skill 👇
9
4
53
4,121
Kimi K3 just launched in the US
We tested it against every other major model on Hyperagent with a real agentic job: Build an investment analysis on SpaceX.
Fair values ranged from $35 to $110
Here's what we found ⬇️
11
9
86
9,428
Opus 5 is a modest upgrade, but it has tough competition in GPT-5.6 Sol
Initial thoughts from our early access:
> Clearer writing than Fable's dense communication style
> Still gets too verbose compared to GPT models (even when explicitly instructed to keep text purposeful)
> Strong on browser use inside Hyperagent
> Strong on reviewing complex data to justify a decision
> Less likely to take creative risks. More of a "rule follower" personality.
Costs shown for five test cases (GPT-5.6 Sol was consistently much cheaper)
Explore and judge the actual work for yourself in thread 👇
27
11
106
154,792
We're building the Agent Team for GTM:
- Signal: Sends personalized assets to best-fit buyers based on buying signals from @useapolloio + web research.
- Inbound: Watches the inbox 24/7, books meetings and preps you for calls.
- Redline: Brutally honest sales coach that gives feedback on your call transcripts from @Gong_io @meetgranola @Zoom
Free templates on Hyperagent Marketplace 👇
10
6
36
3,434
Third: an employable agent has to learn on the job.
A static agent makes you repeat the same context, corrections, and preferences every time.
An employable agent remembers the feedback, changes how it works, and gets better with every assignment.
1
2
557
Second: the unit isn’t one agent. It’s the team.
Howie shared the example of a landscaping company: one agent surveys the yard, another builds the quote, another creates the client pitch.
The leverage comes from the handoffs.
1
3
293
What does it take to make an agent employable?
At @aiDotEngineer, Hyperagent CEO Howie Liu argued that completing a task isn’t enough. An employable agent has to be trusted with an outcome.
That requires three things.
First: it works when you’re not watching.
6
3
20
2,355
You could be generating unique assets for EVERY prospect in your next outbound campaign
GTME teams built this with @useapolloio + Hyperagent
We built a skill that lets you run campaigns exactly like this, link below 👇
5
3
25
2,122
Unless you're a designer, it's hard to articulate visual direction to your agents.
You end up saying stuff like "make it look less generic."
So we built a skill that shuffles the aesthetic of your project with 5 distinct alternatives.
Get Design Shuffle below 👇
7
4
25
2,100
We gave 20 startups one brief: show us the best agent-first company they could build with Hyperagent.
$100,000 in prizes were on the line at AI Engineer World’s Fair.
Here’s who won 🥇
6
7
45
3,166
Most benchmarks just give you a number on a chart.
We made one where you judge the actual work side-by-side.
Here's Fable 5 vs GPT-5.6 Sol on a mix of design, writing, and creative work.
Try Hyperbench and choose your winner: hyperagent.com/s/rcW3h5uutU4…
24
18
104
125,266
New skill on marketplace: Roast My Idea by @nateherk
Convene a council of five independent personas who debate your idea from every angle.
Built to overcome the sycophantic tendencies of some models.
Get the skill below ⬇️
14
5
35
93,327
"Create a data viz story about all goals scored at the World Cup"
GPT 5.6: hyperagent.com/s/uxC40v0U39d…
Fable: hyperagent.com/s/dD9Kvkw--Pl…
2
1
10
953
"Design a site plan for a 100-acre fitness resort"
GPT 5.6: hyperagent.com/s/r_q1v6PYYvB…
Fable: hyperagent.com/s/GkJYe0jdJ0V…
2
1
11
1,200
"Build an interactive explainer about objects that orbit earth"
GPT 5.6: hyperagent.com/s/IakID4i3nJY…
Fable: hyperagent.com/s/FozFETyUv8M…
1
9
946
GPT-5.6 is a viable alternative to Fable for most knowledge work.
GPT-5.6's writing is consistently well-structured and easy to read. Fable's prose can be exhausting.
5.6 also makes fresh design choices with clean hierarchy. Fable can struggle to maintain visual relationships and leans on worn-out AI tropes like eyebrow text.
Examples in thread 👇
- Orbital objects explainer
- Site plan for fitness resort
- World Cup goals data viz
GPT-5.6 is now available on Hyperagent
33
20
161
304,366
Sonnet alone is not ambitious enough and shows poor spatial reasoning. But it knows when to stop and writes tighter copy.
3
2
16
1,108
We're getting the best results by running main agents on Fable and subagents on Sonnet.
Fable is extremely ambitious and thorough. But it can sometimes overthink simple subtasks and write awkward copy. The cost and long runtime is not always justified.
1
1
11
697
Fable 5 and Sonnet 5 really need each other
We ran every Claude model on Hyperagent to build a creative data viz about goals scored at the World Cup
Here's how they did 👇
16
27
103
257,891
Agent browser use is still so underrated.
If your agents can't actually see webpages, they're walking around blindfolded. So much high-value work cannot be done through APIs alone.
Hyperagent runs @browserbase to visually navigate the web live.
5
8
29
5,914
Fashion brands are turning basic studio shots into editorial lookbooks for <$30
Hyperagent acts as art director, then calls Nano Banana and GPT-Image-2 to generate images.
Product details are respected from the real photography, down to the exact patterns and textures on a dress. (do you see any differences?)
One shot with this new Hyperagent skill 👇
2
18
49
3,565
DeepSeek V4 is built for heavy reasoning, and that's where it delivered.
Pointed at a large public research dataset, it surfaced a trend most would miss: the U.S. has lost its long lead in scientific output - and that the quality gap between countries closed years before the volume gap did. ($3.13)
See it here: hyperagent.com/s/1zuFe3J2Fm-…
15
1,852
Qwen 3.7 was the value standout.
The data stories it produced - clean charts, real imagery, a clear narrative - cost 13 to 20 cents per run, well below anything else producing work at that level. ($0.50)
See it here: hyperagent.com/s/DIIUcCQCFXC…
1
11
1,861
GLM 5.2 went up against the frontier flagships on competitive intelligence and beat them, at roughly a sixth of the cost.
It was also the only model that questioned a competitor's marketing claim instead of repeating it - taking an advertised "14x conversions" and showing the real number was closer to 2.3x. ($3.54)
See it here: hyperagent.com/s/t5hRm0RawNg…
12
1,821
Kimi K2.6 is best known as a coding and workflow agent. What stood out to us was its campaign work.
It built the most complete campaign in our testing: a hero video, eight Instagram posts, an email and SMS sequence, and a polished launch board for a sauna pop-up. ($4.22)
See it here: hyperagent.com/s/dMLaqmmu7UU…
3
1
21
2,467
An open-weight model just beat the frontier flagships at competitive intelligence - for a fraction of the price.
We added four open models to Hyperagent - GLM 5.2, DeepSeek V4, Kimi K2.6, and Qwen 3.7 - and tested them on the work our users actually run: prospecting, competitive intel, content, data storytelling.
- Kimi surprised us with impressive campaign content
- GLM won competitive intel use cases outright
- Qwen and DeepSeek delivered strong data stories at a fraction of the cost
Two takeaways: open weights are hitting parity with closed-source on real work, and there's no single best model - different models won different jobs. Which is the whole point of a harness: match the model to the task, get expert output at open-weight prices.
Models available in Hyperagent today.
28
31
182
451,590
Most agents today are singleplayer, running locally, and waiting to be prompted.
They should be the ones prompting you
In Hyperagent, your agents work in the background, constantly learning about you and your business, and producing high quality work without waiting to be told what to do.
Here's how to build your fleet 👇
7
72
576,141
Event producers are selling more booths and sponsorships with this agent
Feed it pictures of your venue + the brands you're targeting (it can also build the list for you)
Then it generates images of those brands at the event so you can sell them on the vision
Skill + example below 👇
2
2
24
157,878
These small businesses have amazing reviews but NO websites.
Hyperagent builds sites you can sell to them. @coreyganim gave us this idea and we built it with one prompt in 9 minutes.
100+ businesses researched, 10 best candidates identified, 10 new sites generated.
skill + real examples below 👇
17
10
99
137,990
We got early access to the Gemini Omni API
Google is calling this model "Nano Banana for video"
3 things we built with it👇
1. Landscaping proposal from customer video. Customer submits a video of the current lot. Hyperagent designs and renders the transformation. Perfect realism, no surreal changes to the surroundings.
2. Animated professor who explains your dashboards. Hyperagent runs analysis on a business question, then generates an explainer video to walkthrough the findings.
3. 8-bit morning briefing. Hyperagent builds morning briefings based on calendar, email/chat, and market intel. Then generates a sidescrolling platformer video showing the goals you'll clear today.
Our take:
> Video has been a vastly underused input for agents. This is the first model to make video directly malleable to our agents.
> Get more imaginative with your outputs. How could a video artifact make the work more memorable or playful?
coming soon to Hyperagent
20
20
167
325,528
We taught our agent to build 5⭐️ app store listings
> uses your brand (or creates one)
> composes screenshots and animations
> writes the description content
> renders a replica of the app store UI so you can see the assets in context
Skill + example below 👇
17
19
104
465,230
Run your day with a retro RPG quest log
Hyperagent designed this UI, then connected it to Airtable for work tracking
5
1
45
187,441
Our agents simulated running a coffee shop at a real location in LA
Agents can peer into the future of your business before you've even started it. Static business plans for SMB are a relic of the pre-agent era.
Skill below 👇
5
2
34
28,556
We wanted better design fundamentals from our agents.
So we fed them this 162-page pdf on designing with a grid system.
Now our agents use code to adhere to a grid and design beautiful layouts.
Example + skill below 👇
36
142
1,512
963,581
What would it look like to run an out-of-home campaign for your app?
Hyperagent generates the creative and places the ads in context.
Skill below
2
12
2,844
What would it look like if Nike opened a hotel?
We asked Hyperagent to imagine it
Fable 5 as creative director, Veo/Nano Banana for visuals
5
3
23
4,518
5. Show the effects of solar flares on aurora
Hyperagent used Fable 5 to research and design a full simulation. Opus's artifact didn't load, despite claiming to have tested the work.
4
42
7,481
4. Simulate the supply chain for World Cup jersey sales based on match outcomes
Fable took a global view. Opus 4.8 narrowed its scope to US only and shipped glaring visual bugs (flickering).
3
3
44
8,343
3. Reconstruct Apollo control panels from technical PDFs
Hyperagent reasoned over the diagrams and documentation to create a faithful 3D simulation. Opus 4.8’s attempt skimmed over key details.
1
8
50
8,366
2. Design a site plan for a 100 acre fitness retreat
Hyperagent orchestrated image gen with GPT-Image-2 and Nano Banana to bring the property to life. Fable’s version considered flow patterns, while Opus 4.8 essentially dropped boxes onto a plot.
See it: hyperagent.com/file/usergene…
4
5
49
8,572
1. Visualize all asteroids in the solar system from NASA data
Hyperagent ran Fable 5 to pull data and design the visual. Fable’s version included visual upgrades such as orbit traces and hover details while maintaining performance.
See it: hyperagent.com/s/-uMvTTWT-_-…
4
13
149
95,178
Fable 5 on Hyperagent is producing the most creative, ambitious work we've ever seen from our agents.
They're self-improving for hours towards open-ended goals. Visual reasoning has spiked noticeably. Outputs are consistently higher quality than Opus, occasionally at lower cost.
5 of our test cases below vs. Opus 4.8 👇
1. Visualize all asteroids in the solar system from NASA data
2. Design a site plan for a 100 acre fitness retreat
3. Reconstruct Apollo control panels from technical PDFs
4. Simulate the supply chain for World Cup jersey sales based on match outcomes
5. Show the effects of solar flares on aurora
Fable 5 is now available on Hyperagent.
83
275
2,665
1,340,052
Design + quoting agent for landscaping
Customer submits a picture of their yard, agent builds multiple design options with full cost breakdown
Skill below
3
5
56
494,799
We wanted better designs from our agents
So we made a skill based on this 49-page pdf about the canonical design principles of Massimo Vignelli (iconic designer behind the 1972 NYC subway map)
Then used it to create a design system for a high speed rail network in the Northeast
Skill below👇
10
4
94
1,043,271
Morning briefing as an email? 🥱
Morning briefing as a slack DM? 🥱
Morning briefing as a movie trailer.
Hyperagent reads your inbox, chats, and calendar + runs a competitive scan. Writes a storyboard, then generates your daily briefing using Veo + @HyperFrames_ / @HeyGen
skill below 👇
2
1
26
428,540
This skill generates claymation podcast clips
Hyperagent runs the whole process:
>analyzes the transcript
>plans a visual storyboard
>generate reference images with Nano Banana and GPT-Image-2 for character consistency
>generates videos with Veo
>edits together the final cut
Skill below 👇
and here's @howietl on @tbpn re: how we're betting on the agent wave
6
26
119,272
You built an app, nice
But no one knows about it. And the vibe is a bit... default.
You need a brand. We built a skill for founders to generate distinct brands for any app.
Create assets for web, social, app store, ads, and UI with a cohesive aesthetic
Skill below 👇
25
14
125
904,241
We wanted better charts from our agents
So we built a skill for generating NYT-style data viz
Includes guiding principles + real code snippets. Based on Hyperagent's study of actual examples + public teachings of @mbostock.
Example + skill below 👇
11
6
63
1,010,202