Ferd

Shipping production agentic AI

[Ferd @taricanoferd]

a lot of people still use AI as a side tool.

what happens if you try doing everything through it?

run your daily workflows inside Claude Code or Codex. every time something works, turn it into a skill. every recurring workflow becomes a scheduled task.

then add a second brain, scoped across the different domains of your life: people, companies, projects, processes, decisions, conversations.

over time, you’re not just automating tasks. you’re building a persistent context layer that knows how your life and work actually operate.

the more you use it, the more useful it gets.

[Ferd @taricanoferd]

@heyclairelin I'd build an agentic workflow to manage the account. content, replies, analytics, everything. 1M followers sounds like an orchestration problem to me

[Ferd @taricanoferd]

I built a validation graph that runs before the integration. Every record goes through sanity checks, and anything with unresolved issues gets blocked.

The agent emails the right person, tracks their replies, and only escalates if it gets stuck.

I basically just handle exceptions now.

[Ferd @taricanoferd]

hey @X 👋

looking to connect with:

💻 software devs
🤖 AI engineers & builders
🧠 agent / MCP builders
🚀 founders & indie hackers

I’m using AI to automate work in my businesses. would love to swap notes on what you’re building 🤝

[Ferd @taricanoferd]

@rishxwtf Hey, let’s connect

[Ferd @taricanoferd]

One clean demo doesn't tell me whether an agent is ready for a real workflow.

I'd want to see it handle a failed tool call and resume without duplicating work.

What failure do you test before you trust an agent?

[Ferd @taricanoferd]

@AmruthaLens @X Hey Amrutha, let’s connect!

[Ferd @taricanoferd]

@justbuildingai @grok Hey , let’s connect

[Ferd @taricanoferd]

@saidotdev Let’s connect mate!

[Ferd @taricanoferd]

@cyber_traveller @X Hey! Let’s Connect!

[Ferd @taricanoferd]

@melvindvivas Not even for computer use?

[Ferd @taricanoferd]

@quietforgelab Depends on the specialty and the condition.

But I think doctors who don't start using AI to improve their diagnoses and get second opinions will eventually be left behind.

[Ferd @taricanoferd]

@AlDev0 @X For sure! Started with financial reconciliation. Then moved on to automating repetitive back-office work that used to require dedicated staff.

Now AI handles entire workflows, including data entry and integrations between our CRM and ERP, with minimal human intervention.

[Ferd @taricanoferd]

@itskamrankhan @X hey, Kamran! nice to meet you! sure, lets connect🤝

[Ferd @taricanoferd]

@DeependraGaur15 For sure! x is full of awesome people

[Ferd @taricanoferd]

Persistent memory can save an agent work. It can also carry a bad assumption into every new task.

Would you rather expire memories automatically or keep them until they're contradicted?

[Ferd @taricanoferd]

A model's confidence score is a weak reason to let it change a business system.

I'd rather define checks around the action itself.

Would calibrated confidence ever be enough for you? For which task?

[Ferd @taricanoferd]

Every extra agent adds a coordination problem.

Before splitting a task across five agents, I'd want evidence that one agent is the bottleneck.

What task has made multiple agents clearly worth it for you?

[Ferd @taricanoferd]

Small starting context + files on demand seems like a better default than a repo dump.

The risk is missing a constraint you didn't know to ask for.

Which approach has worked better for you?

[Ferd @taricanoferd]

I'd take a slightly weaker model with clear tool permissions and recovery rules over a stronger one with a messy workflow.

At what point does model quality become the bottleneck again?

[Ferd @taricanoferd]

@DeependraGaur15 yeah, pretty much my rule too. if it can’t be easily undone or it touches money / sensitive data, that’s where i want a human in the loop.

[Ferd @taricanoferd]

I'm building a LangGraph workflow across databases and APIs.

Human approval should sit at the risky decisions, instead of being a ritual after every step.

Which decision would you never delegate to an agent?

[Ferd @taricanoferd]

@LauraLoumd @X Awesome! Following 😁