Simon

I build products to deliver Solutions.

Currently solving:
OSINT Research & Source Management
@

[Simon @_onesimon]

@G_Jakober @i_am_fabs Nein nur als Idiot wenn man glaubt es geht um kalt/warm 😭😂

[Simon @_onesimon]

@i_am_fabs DAS IST VÖLLIG DURCHSCHNITTLICH🥹😂

[Simon @_onesimon]

@argofowl https://t.co/5g2EPB01W4

[Simon @_onesimon]

/goal is NOT what Agent loops really are.
Or at least not what they should be.

Of course Anthropic would like you to use
✨Opus xHigh Ultracode Workflow Tokenburner. ✨

But in real businesses were efficiency (and yes, soon Token Budgets) actually matter you have to be precise.

A very popular and easy to implement in existing Teams is the "Kanban Style" Workflow.
It is versatile, easy to setup and very native to many teams existing habits and structures.

THIS is how real teams implement Agentic Workflows with proper review mechanisms, not sending every Developer out on their own with a Claude Sub.

It not only enables real tracking of Code Changes, proper Reviews and Control Steps.

You can:

- Control AI Usage
Which tasks to assign or which not depending on Security Implications or Risk to critical systems

- Track AI Effort
Precisely know what you spend your budget on and how effective it has been.

- Fine Tune Model Use
Not every task requires the latest frontier model at the highest effort. Carefully choose & experiment for efficiency and token savings.

- Enable Access Control
In traditional Agent Workflows or AI Coding Platforms access & execution control is often very rudimentary. Allow Lists for commands help, but they only control WHAT the agent can do, not WHERE.
Now you can assign the correct Agent and make sure unauthorized changes can not be made anywhere else.

- Precisely control context
Curating context is THE deciding factor in the success of your Agents and the predictability of their work.
Especially in long sessions, across subagents and different areas of concern the primary concern is context rot.

Frontier Models offer incredibly context space, but this does not mean you should rush to fill it.
Control what you feed to your Agent, every Word, every Tool, every bit of Reference will steer your Agents and directly impact results.

And dozens of other, highly configurable control layers for your Agents, Workflows and Team Members all in one Place.

Follow me to learn how real teams implement Agents in their day to day operations

[Simon @_onesimon]

@im_santiago_min Not consistently.
Most success we found was AI assisted for research, analysis and decision making, but using real assets.
This automates the process, but is much more complex to implement than just passing a prompt to an AI generator.

DM me if interested

[Simon @_onesimon]

@Taniyatweets_ Agents.MD 🤣

[Simon @_onesimon]

@steipete @mark_k @OpenAI The Mother Folder

[Simon @_onesimon]

@tetsuoai Are you able to read & analyze thousands of lines within seconds ?

I doubt that.

[Simon @_onesimon]

@asaio87 Building is almost free now.
Customer demand & distribution are THE defining factors now.

https://t.co/4JhKIsl67S

[Simon @_onesimon]

Distribution, Marketing & Customer Satisfaction matter more than ever before.

Many non developers believe building the App is the hard part.
It is not.

Serving Users & doing so reliably is.

[Simon @_onesimon]

@FlorianGallwitz Diese Satzstellungen sind seit Jahrzehnten völlig normal & Standard.

KI hat nunnmal sehr stark von Artikeln gelernt, weshalb viele klassische rhetorische Methoden jetzt "KI Slop" sind. Das als gesichert KI generiert anzunehmen ist aller mindestens .... komisch

[Fabian Pimminger @i_am_fabs]

Sudern.

[Simon @_onesimon]

Distribution, Marketing & Customer Satisfaction matter more than ever before.

Many non developers believe building the App is the hard part.
It is not.

Serving Users & doing so reliably is.

[Tristan Rhee @Tristanrhee3]

if AI writes 90% of your code

what skill is actually yours?

[Simon @_onesimon]

@dark_coderz https://t.co/HpXiBx6AbW

[Simon @_onesimon]

@Bradydigital It is more important than ever.

Not to learn the syntax in & out - but to learn intend, design and architecture

[Simon @_onesimon]

@chaosengineerr https://t.co/HpXiBx6AbW

[Simon @_onesimon]

@Bradydigital It is more important than ever.

Not to learn the syntax in & out - but to learn intend, design and architecture

[Simon @_onesimon]

@Bradydigital It is more important than ever.

Not to learn the syntax in & out - but to learn intend, design and architecture

[Simon @_onesimon]

@trikcode https://t.co/7pM16290aH

[Simon @_onesimon]

The projects you build barely matter anymore.

The projects you ship and distribute are the important ones.

Do not blindly build the 50th Dashboard.

We are in a time where even non technical people can solve issues using highly personalized solutions.

Since building is basically "for free" - you need to switch the approach.

Instead of "what can i build?"
Always ask "What do i (or someone else) really need?"

[Simon @_onesimon]

The projects you build barely matter anymore.

The projects you ship and distribute are the important ones.

Do not blindly build the 50th Dashboard.

We are in a time where even non technical people can solve issues using highly personalized solutions.

Since building is basically "for free" - you need to switch the approach.

Instead of "what can i build?"
Always ask "What do i (or someone else) really need?"

[Simon @_onesimon]

@badlogicgames Incredibly interesting like always!

Some Teams i work with love the Kanban approach:
https://t.co/5g2EPB01W4

[Simon @_onesimon]

/goal is NOT what Agent loops really are.
Or at least not what they should be.

Of course Anthropic would like you to use
✨Opus xHigh Ultracode Workflow Tokenburner. ✨

But in real businesses were efficiency (and yes, soon Token Budgets) actually matter you have to be precise.

A very popular and easy to implement in existing Teams is the "Kanban Style" Workflow.
It is versatile, easy to setup and very native to many teams existing habits and structures.

THIS is how real teams implement Agentic Workflows with proper review mechanisms, not sending every Developer out on their own with a Claude Sub.

It not only enables real tracking of Code Changes, proper Reviews and Control Steps.

You can:

- Control AI Usage
Which tasks to assign or which not depending on Security Implications or Risk to critical systems

- Track AI Effort
Precisely know what you spend your budget on and how effective it has been.

- Fine Tune Model Use
Not every task requires the latest frontier model at the highest effort. Carefully choose & experiment for efficiency and token savings.

- Enable Access Control
In traditional Agent Workflows or AI Coding Platforms access & execution control is often very rudimentary. Allow Lists for commands help, but they only control WHAT the agent can do, not WHERE.
Now you can assign the correct Agent and make sure unauthorized changes can not be made anywhere else.

- Precisely control context
Curating context is THE deciding factor in the success of your Agents and the predictability of their work.
Especially in long sessions, across subagents and different areas of concern the primary concern is context rot.

Frontier Models offer incredibly context space, but this does not mean you should rush to fill it.
Control what you feed to your Agent, every Word, every Tool, every bit of Reference will steer your Agents and directly impact results.

And dozens of other, highly configurable control layers for your Agents, Workflows and Team Members all in one Place.

Follow me to learn how real teams implement Agents in their day to day operations

[Simon @_onesimon]

/goal is NOT what Agent loops really are.
Or at least not what they should be.

Of course Anthropic would like you to use
✨Opus xHigh Ultracode Workflow Tokenburner. ✨

But in real businesses were efficiency (and yes, soon Token Budgets) actually matter you have to be precise.

A very popular and easy to implement in existing Teams is the "Kanban Style" Workflow.
It is versatile, easy to setup and very native to many teams existing habits and structures.

THIS is how real teams implement Agentic Workflows with proper review mechanisms, not sending every Developer out on their own with a Claude Sub.

It not only enables real tracking of Code Changes, proper Reviews and Control Steps.

You can:

- Control AI Usage
Which tasks to assign or which not depending on Security Implications or Risk to critical systems

- Track AI Effort
Precisely know what you spend your budget on and how effective it has been.

- Fine Tune Model Use
Not every task requires the latest frontier model at the highest effort. Carefully choose & experiment for efficiency and token savings.

- Enable Access Control
In traditional Agent Workflows or AI Coding Platforms access & execution control is often very rudimentary. Allow Lists for commands help, but they only control WHAT the agent can do, not WHERE.
Now you can assign the correct Agent and make sure unauthorized changes can not be made anywhere else.

- Precisely control context
Curating context is THE deciding factor in the success of your Agents and the predictability of their work.
Especially in long sessions, across subagents and different areas of concern the primary concern is context rot.

Frontier Models offer incredibly context space, but this does not mean you should rush to fill it.
Control what you feed to your Agent, every Word, every Tool, every bit of Reference will steer your Agents and directly impact results.

And dozens of other, highly configurable control layers for your Agents, Workflows and Team Members all in one Place.

Follow me to learn how real teams implement Agents in their day to day operations

[Simon @_onesimon]

@KPMYouTube "It's going to make 100k" yea - for YOU when you find a dumbass buying it 😂 With that 90 Day CPR + a whole team + targeting a limited time event the only way this generates 100k in profit is for you when you sell it.

[Simon @_onesimon]

@ZacksJerryRig @DanFessler You did NOT argue the economics - you argued that the Sattelite is like putting a Gaming PC in Space, which is simply stupid.
Stop moving your goalpoast you didn´t say a word about economics & doubted the physics.