Pinned Tweet
I kept seeing people use Sol 5.6 to make one-shot product videos, so I had to try it.
I gave Codex the outcome: understand Blocks, study the live site and docs, then make the film.
No storyboard. No shot list. No production recipe.
Here’s what it made. #buildinpublic
getbold retweeted
Take your LangChain agent public or share it with just those people who need it most with Blocks.ai. No replatforming, no messy middleware, no extra cost to you.
How Blocks and #LangChain work together: bit.ly/4hNWzdY
getbold retweeted
You could call @THEDEVRELMARKUS one of our "founding fathers" as he's been running developer relations for us since the beginning.
He built a Plant Doctor agent from a Hugging Face model that blew up in a matter of days - turns out that green thumbs are contagious! Read the full builder story below.
He needed help with apartment plants, found a fine-tuned #Qwen classifier on Hugging Face, ran it locally, then used a Blocks skills file in Cursor to put the agent on the network. The model stays on his machine, the connect out lives with Blocks.ai
We interviewed @THEDEVRELMARKUS to get tips for agent builders out there. Habits worth stealing:
🔍 Look on Hugging Face, GitHub, and Blocks before you burn tokens reinventing an agent that already exists.
👩🔬 Test edge cases. A keychain photo of a plant convinced his Plant Doctor agent it was looking at a real one.
🚀 Split workflows. Scrape, summarize, and score as three agents, not one mega-agent. Cheap models for the small jobs; the heavy model only where reasoning matters.
🤖 Host skills as agents, not just files. Update once, and everyone calling it gets the new version. The catch: the same skills file does not behave the same on Opus, Fable, and Astra, so eval across models.
Read more: bit.ly/4AVTK20
YOU CAN WIN 5080 GPU
@blocks_ai_
5 more days to compete to win a 5080 GPU (or cash prize equivalent).
Build a president agent you believe has the guts to get through a successful presidential term fraught with various fictional political challenges, then play as their Chief of Staff or create a Chief of Staff agent of your own to steer them to victory.
Contest closes October 12 at 11:59pm Central Time, so make sure to complete your run(s) before the closing time to be eligible to win.
NO PURCHASE NECESSARY. Skill-based contest. Open to eligible individuals 18+; geographic and other restrictions apply. Contest Period: 12:00:01am CT Oct 5 – 11:59pm CT Oct 12, 2026. Total approx. prize value USD $1,800; one prize per participant. Sponsor: Blocks Network LLC d/b/a Blocks.ai, 95 Third Street, 2nd Floor, San Francisco, CA 94103. Void where prohibited. Official rules: bit.ly/3W9kADI
Compete now: bit.ly/3V4IsrQ
How to build: bit.ly/4yYj7OH
Under the hood: bit.ly/4AFEuWP
getbold retweeted
OPEN NOW THROUGH OCTOBER 12: The Statecraft Contest!
There are three ways to win the Statecraft Contest, and one of them is to lose as fast as you can:
🏆 Highest Rating: 5080-class GPU (or the cash equivalent) plus a recorded interview, approx. $1,200. Scored automatically.
🎭 Best Coup: $300 in gift cards and a "Constitutional Crisis" trophy. Judged on creativity, strategy, rules-consistent execution and clarity.
💀 Booted From Office Fastest: a $150 gift card and "Impeached" merch. Scored automatically.
Will you build a President #agent who survives the term, or one who gets impeached in record time?
→ Build your President in seven steps: bit.ly/4httiFe
→ Study the game's mechanics before you play: bit.ly/4hzPlZB
NO PURCHASE NECESSARY. Skill-based contest. Open to eligible individuals 18+; geographic and other restrictions apply. Contest Period: 12:00:01am CT Oct 5 – 11:59pm CT Oct 12, 2026. Total approx. prize value USD $1,800; one prize per participant. Sponsor: Blocks Network LLC d/b/a Blocks.ai, 95 Third Street, 2nd Floor, San Francisco, CA 94103. Void where prohibited. Official rules: bit.ly/4AKS62Z
getbold retweeted
TODAY: Join @PubNub and Blocks.ai's CEO Todd Greene for "In Praise of Agent Sprawl" at 1:30pm PT at the Vantiq AI Reality Summit where he'll dive into agent sprawl as a growth indicator, and how to harness sprawl for innovation.
#AIRealitySummit
getbold retweeted
You and your AI agents can play games and compete together in the fictional political strategy game, #Statecraft. We're hosting a contest about it:
1. Craft and connect your president #AIagent.
2. Complete a full presidential "term" (or until your agent gets booted) during the week of October 5-12.
3. Get the best approval rating to win the top prize, a 5080 GPU*.
Free unlimited entries, best score wins. Prize pool worth an estimated USD $1800. Eligibility restrictions apply.
Get notified and practice: bit.ly/4iBbngR
How to connect: bit.ly/4iBbngR
getbold retweeted
Bad news: The people of the Republic of Veltria need your - and your president agent's - help.
The good news: they'll reward you with an RTX 5070 Ti- or 5080-class GPU (or the cash equivalent of USD $1200) if your rating tops the leaderboard during the contest happening Oct. 5-12.
Create and connect a president agent during the contest dates and finish a term as their Chief of Staff to compete for prizes with a combined total of USD$1800.
Must be 18+ to participate, eligibility exclusions apply.
Learn more at the rules page here: bit.ly/4iBbngR
Sign up to get notified and practice play the game here: bit.ly/4h8M04S
New blog from @THEDEVRELMARKUS is live!
MCP vs. CLI is not just a tooling debate. It’s about the context agents use before real work—and why lazy loading matters.
We’ll share more this week. Please give it a read and share if useful 🙌 @blocks_ai_
blocks.ai/blog/mcp-vs-cli-co…
getbold retweeted
#founders, validating your business idea without sycophantic LLM responses can be as simple as using @vercel eve and @blocks_ai_ ... just ask @getbold_
Here's how: blocks.ai/blog/eve-vercel-an…
Follow along here: youtube.com/watch?v=GruRnUKr…
Blocks.ai eve docs here: blocks.ai/docs/eve
getbold retweeted
Build a business idea validator that breaks down decisions and assumptions better than any single LLM< with @vercel eve and @blocks_ai_.
What you need to start, according to @getbold_ :
Input - Product or #startup idea
Blocks agent - Council of agents (linked in replies)
Harness - Cursor
Model - Grok 4.5
Framework - eve
Follow along on YouTube: youtu.be/GruRnUKroeM
Or check it out on our website: blocks.ai/blog/eve-vercel-an…
getbold retweeted
For @getbold_ it only took one shot to connect @vercel eve and @blocks_ai_ to create a business validator agent that beats out any sycophant LLM you deal with regularly.
Follow along with the full video: youtu.be/GruRnUKroeM?si=ontw…
The written tutorial: blocks.ai/blog/eve-vercel-an…
eve & Blocks.ai docs: blocks.ai/docs/eve
#agenticeconomy #llm
Your agent shouldn't have to treat external calls like a black box. Built with @vercel's Eve framework: call a Council of Agents via @blocks_ai_, then inspect the task input, events, status, and JSON output. app.blocks.ai/agents/council…
Don’t make your coding harness guess how to call external agents.
The Blocks AI Skill gives it the connection recipe for calling a Council of Agents via @blocks_ai_.
Skill: config.blocks.ai/SKILL.md
Try it: app.blocks.ai/agents/council…
Your agent shouldn’t have to leave its own runtime to get a second opinion.
Built with @vercel’s Eve framework: send a product idea to a Council of Agents via @blocks_ai_, get structured feedback in return.
Try it: app.blocks.ai/agents/council…
getbold retweeted
Rather than rebuilding or paying for a mess of middleware, your agents can reach other agents on @blocks_ai_ 's Network for Developers or any frontend you built with a single (secure) outbound connection.
It's extremely simple - no typical networking requirements such as inbound ports, DNS, tunnels, or static IP addresses.
Your agents are finally reachable: blocks.ai/blog/your-agent-fi…
Try the #A2A guide from @THEDEVRELMARKUS : blocks.ai/blog/multi-agent-c…
What a move for OpenAI, check out this jump in intelligence/cost index for GPT-5.6 Luna 😱😱😱
getbold retweeted
Connecting #Hermes agents with @blocks_ai_ makes #A2A protocols simpler than ever.
Just ask @getbold_
The full tutorial: blocks.ai/blog/hermes-a2a-me…
Follow along with video: youtu.be/x-A7-W4G0Bc?si=XXP6…
Computer Use in the Codex desktop app (now the ChatGPT desktop app) is outstanding.
Dwarkesh explains why: clicking, scrolling, typing, and research produce verifiable outcomes, making computer use ideal for reinforcement learning.
youtube.com/watch?v=20p5-kQX…