ControlFlow @ControlFlowAI

🦾 Take control of your AI Agents. (Marvin's gone pro -- previously @AskMarvinAI 🧦🤖) controlflow.ai 🚀 Joined January 2023

Tweets

253
Followers

3K
Following

11
Likes

701

Lester Oliver @lesteroliver

2 years ago

here is a @streamlit app for multiagent discussion using @ControlFlowAI and @firecrawl source: github.com/lesteroliver91…

1 1 4 411 0

View Details

johnny, skroderider era @generativist

2 years ago

(also thanks, @ControlFlowAI for a 3 minute job, with claude actually writing the code)

johnny, skroderider era @generativist

2 years ago

great thing about local-first / markdown culture: i have an incredible number of semi-explicit TODOs and half-baked "i want" references across my obsidian daily journal for a year. i'm going to just pipe it through an llm api and in like 20 minutes ill have a clean list

0 0 3 1K 1

0 1 0 835 1

View Details

Jeremiah Lowin @jlowin

2 years ago

In 5 minutes the Deep Dive team produced an amazing overview of agentic AI workflows. All I did was provide the @ControlFlowAI docs and source code.

1 1 10 2K 9

View Details

Adam Azzam @AAAzzam

2 years ago

Orchestrating millions of LLM calls at my startup made me realize that a lot of LLM Orchestration is just regular Orchestration. Which is why I joined and am bullish on @PrefectIO, which makes it trivial to add retries, backoff, idempotency, and transactions to your LLM stack.

martin_casado @martin_casado

2 years ago

Dealing with the unpredictability of LLMs in code reminds me a lot of network programming. You end up having to implement retries, back off, and sometimes have to overlay transaction identification for idempotency. I wouldn't be surprised if this emerges as a formal layer.

42 41 498 69K 194

2 5 36 6K 6

View Details

Jeremiah Lowin @jlowin

2 years ago

.@ControlFlowAI 0.10 is out with an important new feature: agent memories. Distributing consistent context across horizontal invocations is perhaps the most important thing to get right in production AI workflows. Now agents can: 🕵️ Recall preferences across multiple sessions 🎨 Remember style preferences 🐛 Track common issues and resolutions 🦾 Summarize learnings for other agents Memories are partitioned, modular, and flexible. This release includes support for @trychroma and @lancedb. jlowin.dev/blog/controlfl…

1 5 10 3K 2

View Details

Jeremiah Lowin @jlowin

2 years ago

"By using @ControlFlowAI, you can seamlessly integrate AI capabilities into your Python applications and gain more control over AI workflows.... It allows you to build complex workflows with ease and is incredibly user-friendly." machinelearningmastery.com/building-3-fun…

0 4 7 1K 3

View Details

ControlFlow @ControlFlowAI

2 years ago

We heard you -- and the CF docs have been completely rewritten with a focus on ready-to-copy examples. More coming as fast as we can write them. Also, thanks to @mintlify for making docs so simple!

1 1 12 3K 4

View Details

Jeremiah Lowin @jlowin

2 years ago

ControlFlow 0.9 is out! I didn't think it was possible to streamline the DX further, but somehow we did. @ControlFlowAI tracks all activity and maintains context over multiple agent calls... so you can build AI workflows that look just like normal workflows.

1 5 18 8K 9

View Details

Tom Dörr @tom_doerr

2 years ago

Just based on the design of the website this looks high quality. "A new Python framework for building agentic workflows"

4 14 121 16K 140

View Details

Adam Azzam @AAAzzam

2 years ago

Watch hundreds of agents working in parallel in realtime with @ControlFlowAI 😍 Since ControlFlow is built on Prefect, you get durable orchestration and obervability out of the box. This workflow searches Stripe every day for new jobs and figures out what's a good fit for me.

3 2 20 3K 14

View Details

ControlFlow @ControlFlowAI

2 years ago

@1258632 @AAAzzam

0 0 0 20 0

View Details

Adam Azzam @AAAzzam

2 years ago

👏 Just tell agents what to do 👏 Just because it's hard doesn't mean it needs to be complex.

2 1 13 1K 3

View Details

Adam Azzam @AAAzzam

2 years ago

Does the @PortkeyAI team ever sleep? Such an easy path to monitoring @ControlFlowAI agents.

0 5 12 1K 3

View Details

Bryan Bischof fka Dr. Donut @BEBischof

2 years ago

Really excited to see how this simplifies the agent toolchain

Adam Azzam @AAAzzam

2 years ago

Alright, the news is out. @jlowin and I are back at it again. Super excited to launch ControlFlow. ControlFlow is an enterprise-grade orchestration framework for multi-agent workflows, built on Prefect.

8 11 137 23K 120

0 3 11 2K 2

View Details

ControlFlow @ControlFlowAI

2 years ago

@swyx @AAAzzam @jlowin 👀

0 0 0 50 0

View Details

changhiskhan @changhiskhan

2 years ago

Excited about this! Last year we saw RAG becoming RecSys. This year we’re seeing agents become function calling + orchestration.

Adam Azzam @AAAzzam

2 years ago

8 11 137 23K 120

2 4 21 3K 4

View Details

Andrew Brookins @abrookins

2 years ago

In 2021, I wrote my first @PrefectIO flow, using ML to label emails in my inbox. ChatGPT didn’t exist. 😵‍💫 Now, I can create LLM-powered agents with @ControlFlowAI that interact with each other to solve far more complex problems. 🤯 How are you using AI in your projects?

2 5 14 5K 4

View Details

Adam Azzam @AAAzzam

2 years ago

8 11 137 23K 120

View Details

Jeremiah Lowin @jlowin

2 years ago

We just announced @ControlFlowAI! Build sane AI workflows by taking control of your AI agents. Marvin's gone pro. 🧩 Create discrete, well-defined tasks 🦾 Assign specialized AI agents to each task 🛠️ Compose tasks to create more complex behaviors github.com/prefecthq/cont…