Ankit Goyal

@UTDTrey Iriola is doing incredible with bournemouth. He has a better CV than carrick IMO. If villa win europa league, bournemouth could be in champions league

We have to start a serious conversation. Should we make Carrick our permanent manager or find someone else ?

@Investorike @AliceTalksFooty Lol i wish. Wouldnt have had feelings if i was one. the current state of man united really hurts. #Glineos out

@chamy5 @AliceTalksFooty Amorim bot spotted

Did it with a relegation side in a month but we had to wait 11 months for Amorim to do this😭

@thsottiaux Switch codex to use claude opus/sonnet and you’ll be there

What are we obviously not getting right with Codex?

@CharZiyanah @sampilger @Derminhio This Mic drop. Nothing else to add

Sam, why is it so hard to understand why fans have doubts? Surely you must know that there is always context to results? And that the job next season is a different beast? How about the fact that 3 of the 4 matches Carrick failed to win have been in weeks where we’ve played more than 1 match, much like what we’ll face next season? Or what about the low possession counter attacking football we’ve played for the majority of his time? And the memories of Ole as interim?

Manchester United were happy to keep a manager who finished 15th last season, but there are now doubts about keeping a manager whose form since taking over has him in 1st position?

@markgoldbridge @bethTmufc This is why you need to have proper performance reviews at united stand @markgoldbridge

@bethTmufc So one good game makes him class. He's factually been crap since Christmas

Can this silly narrative about Luke Shaw please end ? He’s absolutely class

@RhysSullivan And they don’t post all of their outages btw

we literally have no idea how to make software anymore

@trq212 Happy to partner and share my findings. I was forced to move to Codex and i am getting to used to it now. Much better than Claude Code in the last few weeks

we're doing a lot more of this, hunting down some of the most annoying bugs in Claude Code let me know if you have any white whales

$path the short interest here is getting to squeeze levels. Another $car in the making?

@GaryMarcus Precisely why you beed governance and control of what agents have access to

Total AI disaster, totally predictable

@mukund @AnthropicAI Isn’t this theoretically same as investors asking for royalty till they recoup 1.5x of their investment?

This is insanely circular. Why are the SEC and FTC not all over this? $AMZN invests $25B in @AnthropicAI Amazon is adding up to $25B to its previous $8B investment. This starts with $5 billion now, with $20 billion more unlocked by "commercial milestones." The deal values Anthropic at $380 billion. Anthropic has committed to spending $100 billion on AWS over the next 10 years. AWS GM - 40% - 45% AWS Net Margin - 25% - 30% At a ~35% operating margin, Amazon would see $35 billion in operating profit from Anthropic’s spending alone over the next decade. So, they get the investment return back in <4-5 years! Amazon isn't just an investor; they are the landlord, the power company, and the tool manufacturer for their own tenant.

@TENETTRADEGROUP @ripster47 what percentage of your overall calls do you get it right?

81% Green Today 📊 These aren't blind signal chasers - these are self-reliant traders executing setups with real confidence thanks to @ripster47 's teachings 🎯 This isn't luck and this isn't a bull market only story - we execute in bearish markets too 🔥

@codyhrountree Just vibe coding won’t be enough. They need more that, upgrade the talent at all levels in all departments. Most importantly get rid of people who treat that company as a retirement home Growth is abysmal, better opportunities elsewhere

@chamy5 how do you feel about UIPath now? I almost think this vibe code era will actually accelerate them getting through the backlog. Are you buying more at these levels or wait and see?

@saxena_puru Another big issue is why would someone want to marry a single inference provider when inference is a commodity

Claude Cowork and software vendors The latest agentic AI tools like Claude Cowork are powerful desktop assistants that can handle multi-step knowledge work i.e. researching, synthesising data, creating documents, managing files and running workflows autonomously on your machine. But here's the key reality most investors seem to be missing during the ongoing panic: these agents cannot operate effectively in isolation at enterprise scale. For real business use (especially in regulated environments with sensitive, large-scale data), they need deep integration with existing systems of record and governed platforms. Standalone agents on a laptop lack the secure data access, compliance controls, audit trails, permissions and team-scale reliability that enterprises demand. That's why the smart move from AI labs has been to build plugins, connectors and enterprise features that embed these agents into current software workflows rather than replacing them. The result? AI becomes a productivity layer that makes established data, observability, backup and integration platforms even more valuable and sticky. The initial stock sell-off in software names was classic fear-driven overreaction. In practice, widespread agent adoption is likely to increase demand for clean, governed enterprise infrastructure - not destroy it. Had an interesting discussion with a good friend who used to work alongside ex-CEO - Andy Grove at Intel and is currently responsible for building out AI systems within a rapidly growing global scale-up company. He is very knowledgeable and regularly invited as an expert speaker at Jefferies' conferences to talk about AI and software. Earlier today he told me that agents need the underlying enterprise-grade software platforms to truly shine. This isn't disruption through replacement; it's augmentation that strengthens the moat of serious enterprise software vendors which are integrating AI into their platforms. Long-term, the winners will be those building the intelligent layer on top of robust data foundations. alphatarget.com

@jakebrowatzke @UiPath @danieldines No problem, i do believe they have a very unique opportunity though. Hopefully they can capitalise. Time is running out

@chamy5 @UiPath Appreciate you sharing, even if it's not what I want to hear about UiPath. @danieldines

I'd genuinely like to hear from someone at @UiPath why the following strategic shifts shouldn't be considered. As a $PATH shareholder, I want this company to win and I think some honest conversation about the business model is overdue. 1. Rethink the licensing barrier, especially for small and mid-size customers. Someone close to me recently listened in on a UiPath sales call with a company of a few hundred employees. The prospect's head of AI automation was disappointed to learn they wouldn't work with UiPath directly and that the upfront licensing cost — comparable to a new employee's salary — was required before they could even begin building automations through a third party, with no guarantee of ROI. That friction may have been tolerable when the RPA market had few competitors. Today, open-source and AI-native tools let companies start automating for free and only pay when they see results. UiPath needs to meet that expectation. It was also concerning that the salesperson advised starting with small use cases, which seems to contradict the vision Daniel Dines has shared around using Maestro to tackle an organization's most significant and time consuming challenges first. If dropping the licensing fee requires changing the third party vendor model, do it! Perhaps UiPath should consider deeper investment in direct implementation capabilities — whether through selective acquisitions of top partners or expanded in-house professional services — to ensure the team building the product is closely connected to the people using it. 2. Move toward outcome-based or usage-based pricing. Yes, this could pressure revenue in the short term as new deal stop paying up front, but it's not like the market is rewarding UiPath right now anyway! A lower barrier to entry could dramatically expand the number of companies willing to try UiPath — potentially by an order of magnitude or more. If Maestro genuinely delivers the ROI that's been promised, customers will scale their usage and the revenue will follow. The products are improving, and engineering agents for faster time to value will be another leap forward. But product innovation alone isn't enough if the business model doesn't evolve alongside it. UiPath risks ceding the new market share created by AI to competitors who are simply easier to start with. I honesstly thought six months ago this was the direction $PATH was moving in, big automation outcomes with big outcome based pricing, but I have yet to see any evidence of any change in the pricing model. In fact in the Q4 call it sounded quite the opposite, like pricing hasn't and won't be changing. I'm reminded of something Alex Karp said early in Palantir's AIP rollout, when analysts pressed him on monetization. He said something to the effect of: "If we provide enough value, customers will pay us eventually, we're not worried about that right now." Crazy? Maybe. But that kind of conviction in one's products — prioritizing customer value over short-term revenue optimization — is what separates companies that capture defining market share in a new era from those that protect margins while the market moves past them. @danieldines

@elvissun Exactly, marrying to a single provider is a death sentence, you need a system that can help you change inference providers seamlessly

nice demo but i'm calling it now: this will end up dead like openai's agent builder x.com/claudeai/statu…

@jerryjliu0 The trap is locking down yourself to a single vendor. I would stay away from locking myself into a single inference provider, use open ai, anthropic and gemini as pluggable models and not use anything else from them

If you're an AI/agent builder, it's so important that you don't overbuild and overcommit on a specific toolset and infrastructure. Frontier labs are shipping not just the models, but the harnesses and surrounding tooling such that your existing stack might be obsolete next week. * e.g. if you had a super complex RAG stack, you may need to rip it out in favor of agents + sandboxes * e.g. if you spent a lot of time building the sandbox and serving layer, you may not need to anymore if you can just bootstrap the product with Claude Managed Agents The tradeoff is completely dependent on how good out-of-the-box these proprietary agent wrappers get. Back when the OpenAI Agent SDK came out, most people did not switch from frameworks because they were simply more powerful. Nowadays tools like the Claude Agent SDK + managed agent services are getting way better.

@DeryaTR_ My guess is they are burning through tokens and money at an incredible pace

So Anthropic now claims there was no token-eating bug after all, it was all our fault for using Opus 4.6! Such dishonesty! I cancelled my Max account and, for the first time, turned negative toward an AI company. They claim to have the highest ethics & we need to trust them!? 👎

@mukund @lydiahallie We’ve decided to roll back Claude code for our engineering team. Its clear the product, the team and the processes are immature, they have a long way to go

@lydiahallie Say a lot. Communicates nothing. Prime example.

Thank you to everyone who spent time sending us feedback and reports. We've investigated and we're sorry this has been a bad experience. Here's what we found:

@mukund @lydiahallie Exactly, i have no idea what this post means. Has nothing valuable