Jet

Level up with AI Automation.
AI Agents obsessed, solving problems with evolving AI systems, scaling business with 100% AI.

Founder of Ex-Bytedance.

[Addy Osmani @addyosmani]

https://t.co/hIe0UX7z6T

[Jet @jettonext]

8/ Reply with your sales stack:

CRM:
Call recorder:
Slack/Teams:
Lead source:

I’ll tell you the first AI teammate I’d build.

[Jet @jettonext]

Your sales team does not need another AI SDR.

It needs a CRM janitor.

Most B2B teams are trying to automate prospecting before they fix the mess already inside HubSpot.

Bad idea.

Dirty CRM data kills every “AI agent” you build on top of it.

[Jet @jettonext]

7/ If you run a B2B sales team, audit this today:

How many minutes does one rep spend after each sales call updating CRM?

Now multiply by:
calls per week
number of reps
52 weeks

That number is your admin tax.

[Jet @jettonext]

Excalidraw now supports MCP.

This saves a ridiculous amount of time.

No more manually dragging icons, drawing arrows, or cleaning up messy layouts.

Just connect the MCP server, type what you want:
“Draw me an outbound system like xxx.”

And AI can generate a high-quality, fully editable Excalidraw diagram in seconds.

The setup is surprisingly easy too.

Excalidraw is becoming one of the best visual workspaces for AI.

[Jet @jettonext]

@mikefutia COWORK

[Jet @jettonext]

Your payroll is leaking through boring work.

Not bad strategy.
Not lazy employees.
Not “lack of AI adoption.”

Boring work.

Managers chasing updates.
Sales reps cleaning CRM fields.
Ops teams moving data between tabs.
Engineers writing reports no one reads.

That is not productivity.

That is expensive human middleware.

The highest-paid people on your team should not be doing the lowest-leverage work in your company.

Pay them for judgment.
Pay them for taste.
Pay them for closing, building, deciding, and moving the business forward.

Then automate the rest.

Most teams do this backwards.

They hire another operator before killing the work.
They buy another dashboard before fixing the workflow.
They ask for “AI strategy” before naming the 5 repetitive tasks bleeding 40 hours a week.

Start here:

Pick one boring workflow.
Put a dollar value on it.
Build the n8n spine.
Use AI only where judgment is needed.
Add human tripwires.
Ship it in 2 weeks.

That is how lean teams get dangerous.

Not by replacing people.

By stopping $120K employees from doing $12/hour work.

[Jet @jettonext]

I never opened openclaw chat UI again!
After throwing 10 openclaw agents into a chat group, and work with them for a week!

I've found what open claw need today:
1. A better long-term memory system and prompt cache.(it's kind of expensive)
2. A communication framework that can let agents really communicate with other agents, share chat histories and memories.
3. A better way to manage the api keys.

[Jet @jettonext]

@mohbii No offense, but I paid $2k bills for Cursor EACH MONTH. But I think this will change soon.

[Jet @jettonext]

Some people say “Cursor is DEAD. It’s OUTDATED.”
Because we’re entering the era of Agentic AI.

But the real question isn’t Cursor.

The real question is:
When will there be a true AI-native tech stack for SaaS startups?

Agents are evolving extremely fast.

But today’s tools — even things like OpenClaw — are not infrastructure yet.
They are mostly agent systems for individuals or small teams.

Useful, yes.
But not foundational.

What we are waiting for is something bigger:

An agent infrastructure layer.

Imagine OpenClaw evolving into something like this:

• A foundation service for SaaS companies
• Agents interacting with APIs autonomously
• Entire workflows executed by AI
• Software designed primarily for agents, not humans

Because the reality is:

The future users of SaaS products will not be humans.
They will be agents.

Agents will:

• call APIs thousands of times per hour
• trigger workflows automatically
• generate massive traffic to SaaS systems

Even if 90% of the actions are imperfect,
agents will still operate orders of magnitude faster than humans.

This will fundamentally change SaaS economics.

Traditional SaaS was built for:

human login → human action → human workflow

The next generation SaaS will look like:

agent request → agent orchestration → agent execution

This means:

• pricing models will change
• API design will change
• infrastructure will change
• product UX will change(less UI for humans)

The companies that survive will not be the ones with the best dashboards.

They will be the ones that design for agents first.

Startups need to move faster.

Because AI is already operating at a speed that startups themselves struggle to match.

[Jet @jettonext]

@mohbii I'm not talking about models, I'm talking about maybe for most people, they probably won't need an IDE at all, except for a small group of experts.

We let agentic AI to do this, just imagine what will happen if openclaw gets 10x better.

[Jet @jettonext]

https://t.co/COUonXVI4t

[Jet @jettonext]

@rovvmut_ Couldn't agree more, in the end, seedance 2.0 might become a privilege for Hollywood.

[Jet @jettonext]

Creating a video was just as complicated as coding.
Coding are largely solved by AI today.
Creating videos are still not, but with the new model like seedance 2.0 and LLM like sonnet or opus, it starts feeling like this will be solved in the next coming year.