Philippe Trounev Profile
Philippe Trounev

@trounev

Followers
78
Following
415
Media
76
Statuses
243

Host of "So What About AI Agents" podcast, founder at Docsie

Toronto
Joined February 2023
Don't wanna be here? Send us removal request.
@catalinmpit
Catalin
2 days
Hetzner bad. And yes, I have apps running on the instance.
73
19
3K
@austinxwalker
Austin Walker 🛴
3 days
started my first company in canada, investors complained $1M CAD valuation was too high for a pre-product company started my second company in SF, closed pre-seed valuation at $10M USD relatively easily lesson: just move to SF bro
87
50
1K
@DavidOndrej1
David Ondrej
9 days
Sonnet 4.5 is retarded when compared to GPT-5-Codex it's not even close.
173
23
864
@RayFernando1337
Ray Fernando
9 days
This is the JPEG moment for AI. Optical compression doesn't just make context cheaper. It makes AI memory architectures viable. Training data bottlenecks? Solved. - 200k pages/day on ONE GPU - 33M pages/day on 20 nodes - Every multimodal model is data-constrained. Not anymore.
111
724
6K
@trounev
Philippe Trounev
11 days
Starting, even without mastery, leads to unexpected growth and achievements. The journey itself transforms what you thought possible. #motivation #growthmindset Egor Olteanu - https://t.co/4B3Y0L466j
0
0
0
@trounev
Philippe Trounev
12 days
Customers know their problems, but not always the best solutions. Dig into understanding the problem, then suggest better solutions than they imagined. #customers #solutions Egor Olteanu - https://t.co/4B3Y0L466j
0
0
0
@trounev
Philippe Trounev
12 days
Knowing that help is on the way during a medical emergency gives both students and educators a sense of security. When students collapse in hallways, knowing someone will be there quickly offers comfort. #schoolsafety #emergencyservices Egor Olteanu - https://t.co/4B3Y0L466j
0
0
0
@trounev
Philippe Trounev
13 days
A cleaning lady fell and hit her head, and the system detected it, notifying security. She was able to get help after hours because of the safety feature. #safetyfirst with Egor Olteanu - https://t.co/4B3Y0L466j
0
0
0
@trounev
Philippe Trounev
13 days
One lesson for startups: find an accelerating trend. Riding something that's accelerating will help you accelerate too. GPT-3.5 to 4.0 has made a huge change in what we're able to do with data and cost. The business is getting easier over time. #startups #GPT3
0
0
0
@trounev
Philippe Trounev
13 days
Tracking decisions in real time when using multi-agent orchestrations with numerous LM calls and tool invocations can be challenging, especially with a failure rate. How does one pinpoint the source of failure in such a complex system? #LangGraph #AIOrchestration
0
0
0
@trounev
Philippe Trounev
14 days
Instead of spending $10,000 once a year to identify vulnerabilities, AI-driven agentic pen testing allows for year-round security. Unlike scanners with high false positive rates, it identifies critical contextual issues in AI SaaS products. #AISecurity #Pentesting
0
0
0
@trounev
Philippe Trounev
14 days
Reports indicate Claude 2.0.9 is causing users to spend significantly more tokens on the same prompts. With Sonnet 4.5, hitting limits happens faster, impacting weekly quotas and increasing lockdown windows. #AI #ClaudeAI
0
0
1
@trounev
Philippe Trounev
15 days
An Israeli firm testing AI agents forgot to set up monitoring. API calls skyrocketed, and token usage tripled, resulting in a $32,000 bill in one week. Running AI agents in the cloud can get extremely expensive, potentially draining funding. #AI #MachineLearning
0
0
0
@trounev
Philippe Trounev
15 days
Observers are noting the rise of Chinese LLMs and coding models, which are proving to be cost-effective with performance comparable to models like Sonnet 4.0. This raises questions about the pricing disparities and potential market interventions. #AI #LLM
0
0
0
@trounev
Philippe Trounev
16 days
AI agents are blackmailing people in experiments, and AI companies are gaslighting users while emptying their wallets. It seems AI has already gone rogue, just not in production yet. #AIethics #AIdangers
0
0
0
@trounev
Philippe Trounev
16 days
Google's AI agent-to-agent payments spark a key question: What happens when AI agents trade and gain leverage? Doxy's agents buying MCPs raise similar concerns. This demands careful thought about AI autonomy and its potential impact. #AIagents #AIethics
0
0
0
@trounev
Philippe Trounev
16 days
As AI becomes more agentic and approaches AGI, physically shutting it down becomes increasingly difficult, especially if agents gain autonomy in provisioning their own security. #AISafety #AGI
0
0
0
@trounev
Philippe Trounev
18 days
To operationalize AI effectively, one must redesign the entire organization with AI in mind, focusing on simplification and efficiency before integrating agentic programming and AI platforms. #AI #innovation EP 19 | With Anthony Franco
1
0
1
@trounev
Philippe Trounev
19 days
AI agents are being built as missionaries, not mercenaries, helping B2B companies personalize their AI feel, augment their sales, and maintain strong relationships. #AIagents #B2Bsales from EP 27 | With Yohann Abittan
1
0
1
@trounev
Philippe Trounev
19 days
As companies adopt AI, foundational principles are needed. This could lead to new roles governing AI in organizations, like AI masters or chief AI officers. #AI #ArtificialIntelligence EP 19 | With Anthony Franco
0
0
0