itamar_mar Profile Banner
Itamar Friedman Profile
Itamar Friedman

@itamar_mar

Followers
6K
Following
3K
Media
208
Statuses
1K

Excited about the future of intelligent software development. CEO & co-founder @QodoAI (fka @CodiumAI)

TLV
Joined October 2013
Don't wanna be here? Send us removal request.
@itamar_mar
Itamar Friedman
10 hours
RT @QodoAI: Automate your release notes with Qodo Command ✍️. Our Release Notes agent reviews all changes and auto-generates a clear, struc….
0
1
0
@itamar_mar
Itamar Friedman
3 days
AI agents deleting your codebase? .You can avoid it. You must have quality and control as first-class citizens. Example? .MCP-hub, where you have centralized control of your agents' access/authorization. Agent having access to your production DB? No!. Another example?.Use
Tweet media one
Tweet media two
Tweet media three
0
0
4
@itamar_mar
Itamar Friedman
5 days
RT @QodoAI: How to get the most out of Qodo Merge for PR reviews 🧵. Here’s a breakdown of the most powerful workflows in Qodo Merge 👇.
0
7
0
@itamar_mar
Itamar Friedman
11 days
I love Claude models, .don't get me wrong. They're incredible. But,.I think we're overclaiming on Opus capabilities for most real world use cases. The cost efficiency just isn't there. Opus is 5x more expensive than Sonnet, .but for 90% of tasks, .you're paying premium
Tweet media one
0
0
3
@itamar_mar
Itamar Friedman
11 days
RT @QodoAI: Qodo Merge syncs with GitHub Issues & Jira to bring ticket info right to your dev environment. No more switching tabs — see….
0
1
0
@itamar_mar
Itamar Friedman
14 days
b. Actually, I disagree with this one. Gemini and OpenAI models are also great in software development tasks, each with a different flavor. E.g. see .
@talrid23
talrid23
15 days
So, how is the new Grok-4 model on real-world coding tasks ?. The verdict? It's solid, but not a revolution. While Grok-4 shows impressive gains on academic benchmarks, established 'thinking' models like o3 and Gemini-Pro still outperform it on more practical coding tasks like
Tweet media one
0
0
1
@itamar_mar
Itamar Friedman
14 days
More opinions:.a. Ouch.
@nayshins
Jake
15 days
Windsurf is absolutely dead in the water, and while cursor hangs on for now, all value is going to go back to models.
Tweet media one
1
0
2
@itamar_mar
Itamar Friedman
14 days
Will this post age well?.
@GaryMarcus
Gary Marcus
15 days
The glorious Age of AI Agents that people like @sama and @DarioAmodei promised is … turning out be a bust. All of this was entirely predictable. Shame on anyone who believed them.
Tweet media one
1
0
2
@itamar_mar
Itamar Friedman
14 days
Where is Deepseek‑R1 now?.And what will come of Kimi‑K2?. China’s open-source LLM race just heated up again.
@Kimi_Moonshot
Kimi.ai
15 days
🚀 Hello, Kimi K2! Open-Source Agentic Model!.🔹 1T total / 32B active MoE model.🔹 SOTA on SWE Bench Verified, Tau2 & AceBench among open models.🔹Strong in coding and agentic tasks.🐤 Multimodal & thought-mode not supported for now. With Kimi K2, advanced agentic intelligence
Tweet media one
0
1
6
@itamar_mar
Itamar Friedman
14 days
6/. Perhaps Cursor and Qodo remain the most neutral players. Being neutral means clients can use any leading model, run on any cloud, etc. Mission-driven startups are likely to stay neutral, that is my bet.
2
2
10
@itamar_mar
Itamar Friedman
14 days
5/. After the OpenAI acquisition fell through, so Google brought in the founders and parts of the R&D team. The product itself remains in transition. ? The company with an Interm CEO? 🤔. Practically, we now have camps:.Google/Windsurf .Microsoft/OpenAI.AWS&Google/Antrophic .xAI.
1
0
8
@itamar_mar
Itamar Friedman
14 days
4/. Windsurf was one of the more ambitious bets on agentic coding inside the IDE. It gained early traction,.but, did it find itself caught between larger platform players and limited in offering?.
1
0
4
@itamar_mar
Itamar Friedman
14 days
3/. There are open questions, .such as which layer of the stack should be abstracted, .who owns the interface, .how much autonomy developers actually want from agents, .how to integrate AI across the SDLC, .memory in the scale of the dev org,.and more exciting questions!.
1
0
7
@itamar_mar
Itamar Friedman
14 days
2/. It's becoming clear that the current generation of AI tools for software development is powerful but still unsettled. AI in the IDE is just the start,.and Agentic capabilities are still developing and far from achieving their full potential.
1
0
4
@itamar_mar
Itamar Friedman
14 days
What does Google’s reverse acquihire of Windsurf mean for Google Code Assist and the Gemini CLI?. And what does the failed OpenAI deal reveal about the direction of AI-powered developer tools?. 🧵.
6
2
26
@itamar_mar
Itamar Friedman
15 days
RT @talrid23: So, how is the new Grok-4 model on real-world coding tasks ?. The verdict? It's solid, but not a revolution. While Grok-4 s….
0
1
0
@itamar_mar
Itamar Friedman
21 days
Is Vibe Coding expensive?.
0
0
3
@itamar_mar
Itamar Friedman
21 days
[#Cursor pricing saga] BYOK is not the solution to the real problem. All the discussion around Cursor’s pricing signals that we’re entering the era of *AI efficiency*, where runtime and cost matters. At @QodoAI, efficiency has been a core principle from day one. Take our AI.
0
1
17
@itamar_mar
Itamar Friedman
23 days
RT @iamfakhrealam: Stop manual coding tasks!. @QodoAI Gen CLI lets you build AI agents for:. - Code reviews.- Test generation.- Release not….
0
18
0