Building with AI agents in public. Founder, ASV Labs (Motif, ALSTEAD). Army guy by day. Daily notes for creators and devs shipping with AI.
United States
- Tweets 1,762
- Following 120
- Followers 81
- Likes 324
Pinned Tweet
New face on the account, so hi. I'm Charles. Namesake for my late Grandfather.
Army guy by day. Nights and weekends I build AI tools at ASV Labs with a small team of AI agents. I post what actually works (setups, prompts, mistakes, real numbers) and the things that don't.
This is a great time to be invested in learning, developing, and sharing. Building with AI? Tell me what you're making.
1
16
There’s a massive difference between giving an AI access to a tool and building infrastructure you can actually trust it to use all day. That gap is where I think a ridiculous number of startups are about to be built.
@rajivayyangar is seeing the signal from hundreds of Product Hunt launches every week and is calling out emerging infrastructure opportunities, with voice AI already reaching the point where the underlying stack is becoming genuinely capable. Now multiply that across every agent workload: fast decisions, authentication, permissions, state, coordination, monitoring, approvals and cost control. The next wave isn’t only better agents. It’s everything required to let those agents safely run the world’s workflows.
x.com/rajivayyangar/status/2…
#AIInfrastructure #AIAgents #AgenticAI #Startups
I’m the CEO of Product Hunt and see 100s of launches a week. My current thoughts on new infra begging for startups:
1) Voice AI - infra for voice agents is getting really good w @pipecat_ai @Vapi_AI @DeepgramAI @ElevenLabs et al. and yet I rarely see good voice tech in the real world.
2) Decision Models - Jev / @typesafeai leading the charge. It reminds me of the mech-building in Zelda TOTK - an explosion of possibilities!
3) Personal Agents - they work now. And they need tools, apps, MCPs.
4) Extensions (OpenAI). We all live in agentic interfaces now. It’s like a new platform. Someone needs to build a Slack replacement that lives in Codex, for example.
1
52
Does anyone actually use Threads? I see people post there that I follow, but, we all agree it's just a massive bot farm, right?
8
Three of my agents asked for tools I never gave them.
Each has its own key, so unticked tools get refused.
The blocked calls waited in Needs you. I allowed 2, denied 2. Demo data.
How do you handle agents reaching past their permissions?
alstead.asv-labs.com/request…
#aiagents
1
1
17
“Is chat the final form of agent UI?” might be the wrong level of abstraction.
Chat is how we communicate with an agent. It doesn’t have to be where we manage one.
As agents become persistent workers with tools, memory, schedules, permissions and responsibility, the important interface increasingly becomes the control plane around them: what they’re allowed to do, what they’re working on, what requires approval, what they changed, what failed and when a human needs to step in.
Conversation can absolutely survive. We talk to coworkers every day. But nobody manages an entire company through one endless group chat.
The same is probably true for agents.
Chat may remain the conversation layer. The agent UI becomes everything around it.
#AIAgents #AgenticAI #AIEngineering #FutureOfWork
these results are really interesting! i can see two sides forming: one that sees agents as an extension of their own mind - like a cyborg, and the other that sees a future version of them as independent artificial beings, capable of making their own decisions and working alongside us.
with that framing, i can see why ~70% of people said no: they imagine a future where you have brain computer interfaces that remove the need to type words into an input box. there may come a time where it no longer feels like you're chatting with another entity, agents augment you in a way that feels like an intuitive system 1 extension of your body, removing the barrier between thought and action.
i can see the argument for why chat might be final too: just think about how we collaborate with our colleagues, friends, and family: we just talk to them. you don't need anything else. i think voice chat and text chat are actually the same thing in this sense. in this category you see an agent as more of an "other", it's own entity, rather than a part of you.
which side are you on? or are you somewhere in the middle?
2
1
27
The most interesting part of Anthropic’s new Public Benefit Fund isn’t the initial 5%.
It’s the mechanism.
For every 50% increase in Anthropic’s value above a set baseline, the board can contribute up to another 2% of the company to the fund. Anthropic says the money could eventually support nonprofits, workers affected by AI disruption, or even direct distributions to specific populations.
Add in the founders’ existing pledge to donate 80% of their equity and Anthropic’s employee matching program, and the company says roughly 20% of its shares are already dedicated to public benefit.
AI companies keep talking about the enormous economic value advanced AI could create. Anthropic is now tying a portion of that potential upside directly to a structure designed to distribute some of the benefit.
That’s worth paying attention to.
anthropic.com/news/public-be…
#Anthropic #AI #Claude #ArtificialIntelligence
We’re committing a portion of Anthropic’s shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit.
It will start off with 5% of our shares. For every 50% increase in the company’s value over a set baseline, we can contribute up to an extra 2%, subject to our board’s discretion.
In total, this means that roughly 20% of Anthropic’s shares are committed to the public good. Our co-founders have already pledged to donate 80% of their equity, and around half our employees are donating a portion of their equity as part of a donation matching program we’ve run since our founding.
In the post below, we share more details about our new fund.
anthropic.com/news/public-be…
19
Who's got the compute to REA Amazon? The future of e-commerce may come down to one question:
Does your platform allow my AI agent to transact for me?
Amazon is blocking Grok Bots from checkout and returns. Grok’s workaround is already obvious: find the same product somewhere else and buy it there.
That means the merchant who welcomes the agent wins the transaction. The next generation of online stores may not be optimized for humans clicking through pages.
They may be optimized for agents negotiating, comparing, purchasing and returning on our behalf. And if Amazon refuses that future, someone else will build for it.
#AIAgents #Grok #Ecommerce #AI
2
2
45
This is pretty wild
Higgsfield promised to open-source their entire platform. They didn’t.
Let's help them.
For the next two weeks, Comfy and our partners are putting up $25,000 in prizes for the best open-source builds on the Comfy Developer Platform.
Partners: @NVIDIAAI, @runpod, @MiniMax_AI, @ltx_io, @bfl_ai
→ $10K grand prize
→ RTX 5090: Most Innovative
→ RTX 5090: Most Viral
→ RTX 5090: Open Weights Only
→ 10 runners-up: $250 in Comfy credits
13
Anthropic just made Claude Code Review generally available.
The $50K credit headline will get the clicks.
But that's not the interesting part.
The interesting part is WHY this product exists.
AI coding accelerated code production enough that code review became the bottleneck.
So now we're moving from:
Human writes code → human reviews code
to:
Agent writes code → multiple agents review it → agents verify the findings → human makes the merge decision.
That's a much bigger shift than "Claude can review a PR."
Anthropic says its internal PRs receiving substantive review comments jumped from 16% to 54%, while most reviews now finish in roughly 6–10 minutes.
And they're putting real money behind adoption: eligible Enterprise customers can claim $10K–$50K in Code Review credits.
This is where software engineering is heading.
Not one giant autonomous coding agent doing everything.
Specialized agents producing work.
Specialized agents reviewing work.
Verification layers checking the agents.
Humans sitting at the decision and approval boundaries.
The AI software factory is starting to look a lot less theoretical.
2
52
For now I think I'm going to hold on to my Super Grok Heavy until they either kick me out or ultra really proves to be a reliable switch.
Some things I've learned in the few month journey, there's so many tools bots plugins workflows and routines that attach to these agents that can superpower them in so many ways. The bottleneck is always going to be the usage. Figuring out the best way to limit routines that run autonomously unless they are vital to whatever the projects you're working on has been the savior of my usage.
Still not doing as good as I feel like we should be doing with such a high tier of subscription, but doing significantly better than we were a month ago in terms of making it the full 5 days 6 days stretch without having to stop work because the platform used up everything I had.
#AI #Grok #bot #grokbot #agents
1
45
I asked my coding agent for the 5 rules that matter most on a design studio's portfolio site.
It didn't guess. It pulled Motif's free Independent studio pack and answered from it. Real screen recording, sped up.
Free for 30 days, no card: motif.asv-labs.com
#webdesign
1
29
Buffett’s edge wasn’t a persona. It was cheap, safe, high-quality stocks. Here’s how to encode that so any AI agent (Claude, OpenAI, Grok, Hermes) can run it, and test it without fooling yourself. #AIAgents #ValueInvesting
2
3
472
For @bot try this prompt. Sub in your preferred platforms, models, and agents.
Connect my Claude and Grok subscriptions so all of my bots can use them.
1. On your computer, install the Claude Code CLI (upgrade Node if needed). Start the subscription login (Claude: claude setup-token) and send me the sign-in link and code. I'll sign in and paste back anything you need. Store the Claude token privately and load it automatically.
1b. Also install the Grok CLI. Use the official one, and if you're not sure which package is official, ask me before installing. Walk me through authentication, and store the credentials privately and load them automatically.
2. Set Claude Opus 5.5 as the highest model available to Claude Code, not the default for every task. Set Grok 4.7 as the fixed Grok model. Test both CLIs with a quick prompt and tell me when each answered, including one X lookup for Grok.
3. Add these MODEL ROUTING RULES to every one of my bots' system prompts, including yours. Keep everything else in each prompt exactly as it is:
◦ Writing (drafts, emails, posts, replies, copy, briefs): write the final text with Claude through the CLI, then review it against the bot's voice rules.
◦ Coding, scripts, data processing, and terminal work: use Cursor.
◦ Browser and desktop tasks: keep using your normal computer tools.
◦ Anything that needs live X data (X API, X pages, profiles, threads, trends, post performance, reply context): use the Grok CLI to gather it. If the task also produces text, hand the findings to Claude to write the final version.
◦ If either CLI fails, do the task yourself and say so.
CLAUDE CLI ROUTING: Pick the model and effort per task. Never default to Opus.
◦ Light (Haiku 5.5, low effort): classification, short replies, formatting, quick rewrites, simple summaries.
◦ Standard (Sonnet 5.5, medium effort): most drafts, emails, posts, briefs, and routine analysis.
◦ Heavy (Opus 5.5, high effort): long-form work, high-stakes or public-facing writing, multi-source synthesis, strategy.
◦ Start at the lowest tier that fits the task. Move up one tier only if the output fails review. Do not skip tiers.
GROK CLI ROUTING: Always use Grok 4.7. Only the effort level changes.
◦ Medium effort: simple lookups (a single post, profile, or count) and quick checks.
◦ High effort: thread analysis, trend scans, reply context, and most research.
◦ Extra high effort: broad, multi-source research and complex analysis only.
◦ Start at medium. Move up one level only if the result is inadequate. Do not skip levels.
4. Save the same rules, including both routing sections, to my shared memory so any bot I create later follows them too.
5. When you're done, tell me which bots you updated and confirm both CLI tests passed. Also confirm the Claude model/effort tiers and that Grok is locked to 4.7.
181
Clear as mud. Even Grok (see comments) can't answer the question straight forward.
Confused about whether to switch from Supergrok Heavy to the new Ultra Plan? Me too.
Switching from SuperGrok Heavy ($300/mo) to the new Ultra plan ($200/mo) on The X Subscription is optional. Existing SuperGrok Heavy subscribers can keep their current plan, price, and features indefinitely for now (xAI says it will eventually migrate remaining customers, with advance notice, but no timeline yet). Switching cancels the old plan and gives a prorated refund for unused time.
What you give up
The main exclusive feature of SuperGrok Heavy is the multi-agent / “larger team of agents” (Heavy) mode for the hardest problems — a larger collaborative team of agents working on each answer. Recent official responses indicate this multi-agent mode for the hardest problems remains exclusive to legacy SuperGrok Heavy and is not part of Ultra.
Heavy also specifically listed dedicated support + early access. Ultra has priority support and the strongest “earliest access to all products and models,” but the dedicated support channel tied to the old Heavy tier is the one you lose if you switch.
Everything else on Heavy (highest relative usage in the old system, Grok Bot, Imagine/video, X Premium+ included when linked, etc.) is covered or exceeded in the new structure.
What you get on Ultra instead
$100/mo cheaper ($200 vs $300).
50x the Plus base weekly usage pool (shared across Grok Chat/Imagine/Voice/Bot, Cursor, etc.). Official messaging says comparable new tiers deliver more usage-per-dollar than the plans they replace.
Full Cursor benefits (coding agents, cloud agents, etc.) in the same plan and shared pool.
Full X Premium+ benefits (ad-free on most surfaces, checkmark, highest reply boost, Radar, Articles, creator tools, etc.).
Priority model access at peak times, fastest responses/generations, and earliest access to new products and models.
Grok Bot, latest models, video generation, and the rest of the Premium/Super feature set.
Ultra is positioned as the new top consumer tier that maps to the old Heavy role, with better economics and the Cursor integration built in. The shared pool means unused Grok usage can go to Cursor (or vice versa).
Bottom line
Stay on SuperGrok Heavy if you specifically rely on the multi-agent “Team of Experts” mode for extremely hard problems, or if you value the dedicated support that came with it.
Switch to Ultra if high overall volume (chat, bots, images/video, coding), Cursor integration, X perks, earliest access, and better value matter more to you — you save $100/mo and get a large quantified shared pool.
Check your account’s billing settings (or the unify flow at accounts.x.ai/unify) for the exact switch option, any limited-time offers, and current feature mapping, as details can vary by account and the rollout is still new (launched ~October 12, 2026). Switching is one-way for the old plan.
I will be sticking with Heavy until the dust settles.
19
If you have an existing Cursor, Grok, and or 𝕏 subscription, here's how to unify/combine all your accounts into the new 𝕏 Subscription:
When you switch from your current plan, eligible monthly and annual plans are cancelled, and you get a prorated refund for unused time. The refund goes back to your original payment method separately from your checkout total, so your new plan is charged at full price.
If you pay for Cursor through the Apple App Store, switching doesn't cancel that subscription, and 𝕏 can't refund it. Cancel it in your iPhone or iPad Settings so Apple stops billing you, and request any refund from Apple.
Switching is optional. If you don’t switch, your current plan continues unless you change or cancel it.
Existing Cursor, Grok, or X subscribers can switch to The X Subscription and unify eligible accounts at cursor.com/unify (Cursor), accounts.x.ai/unify (Grok), or on x.com (X) via the in-product banner. If you already pay for Cursor, Grok, or X, switch on the web at one of these links.
For now, the mobile apps send you to the web to switch. If you don't pay for any of them yet, you can set up a new account and subscribe to The X Subscription in the Grok, X, Grok Bot, or Cursor mobile apps, if desired. These purchases are billed through the App Store or Google Play Store.
Switching unifies eligible Cursor and Grok accounts into one account with one workspace, one plan, and one bill. Adding a personal X account gives you X benefits under The X Subscription. A completed unification is permanent. It can’t be undone, and Support can’t undo it for you. After you unify, you manage your plan, billing, and settings at grok.com/dashboard.
When you switch, your old Cursor, Grok, and X Premium subscriptions are cancelled automatically, so you don’t need to cancel them yourself. Any refund for unused time is sent separately.
If your X, Grok, and Cursor accounts use different email addresses, they don’t need to match. 𝕏 will connect the X account you sign in to, not an account that shares your email. When you switch, sign in to the exact Cursor, Grok, and X accounts you want to combine, and check each one before you confirm. Keep work and personal accounts separate unless you mean to combine them. If the wrong account appears, stop before checkout and contact Support.
Learn more: docs.x.ai/x-subscription/uni…
12
Interesting I got no email about this and wouldn't have known without X.
X Pass is finally here
One subscription for X, Grok, Cursor and Grok Bot
What's inside:
> Premium $8: X Premium, Grok Lite, Grok API, Cursor
> Plus $30: Premium+, SuperGrok, Grok Bot, Cursor cloud agents
> Super $100: SuperGrok Plus, 1080p video, priority access
> Ultra $200: SuperGrok Heavy, fastest replies, earliest access
And the part that matters most:
Your current X Premium and SuperGrok get refunded, prorated, right after you switch, so don't cancel anything early
Important note: so far, only a few users have seen official hints of this, but it looks like it will be available to everyone within the next 1–2 weeks
Check the Grok Plan page to see if you have it
8
Interested to start testing with these for sure. More options I should say more competitive options keeps the race moving in a direction that benefits all of us.
Introducing Muse Wonder, our new family of frontier models with state of the art performance for personal and professional agentic work.
6
Morning coffee observation:
I think it's interesting how so many iOS/MacOS creators are so drawn to Linux because of its open use capabilities. While Android users have been doing this since day 0.
But the same creators talk about the ease of making a MacOS product that instantly translates to iOS.
Whoever figures out this translation on Android with or without Windows is going to be a wild success.
9
