AI/job update
Thought I’d bump this topic. At a large MM and the scope of tools that have been released in the last 2 months have been game changers. I work at a mega pod, v tenured PM, good relationship w management, ability to take some macro tilts and looser vol leash w a good level of expense budget so we’ve tried a range of the tools and just generally have a process which is more top down and cross-asset. Some thoughts:
1) Shortcut/Rogo/Claudecode: I haven’t built a model or done any excel work since I got a series of skills set up. Model buildouts, sensitivities and data work is basically all done through my Claude now and it automatically makes the model changes I need. Insane, work I was doing on backlog triangulation would take the entire day and now takes maybe 30min of active time (+ the time it takes the agent to build out)
Claude code also MCPs directly into some of our alt data pipelines and ofc broker notes through Alphasense, very good at reading the data vs qualitative comments but can’t quite connect it to an actual read yet.
Has been connected to PnL + positioning data for a while but doesn’t really know how to utilise it
2) idea gen/predictive work: firm has rolled out an AI analyst which I’ve played around with. It can forecast earnings better than most analysts (apparently), I’ve played around with it but not hit an earnings season with it yet so can’t confirm. But directionally its forecasts align with what I’d have put in.
We have played around with an ecosystem of agents who model + put ideas through our process. Not great but velocity of ideas seems promising.
Excited to see where we are in 6m. I think we’re a long way off from analysts with actual sector expertise being replaced, but at least for us we probably would’ve hired 2 juniors/interns by now if we didn’t have these tools.
All in all I guess junior/entry level hiring gets hit. In 18m we can see where things stand but difficult to see senior analyst demand waning any time soon.
AI analyst thing might dox you btw. Feel like I've seen an article on this or at least we've been told internally about one fund doing this (I actually can't remember which fund).
Within the banks/sell-side (and I'm probably in the most AI-focused broker here), I don't think agentic stuff is quite yet where it could be and we don't yet have Claudecode etc. But even so I still feel like I'm not fully taking advantage of this. I really need to sit down and spend like a couple days doing a deep dive into everything I have available. Any advice into how to go from "I have all these tools" to "this is what I can automate and I have these embedded processes"? It's very overwhelming even if everything's in front of you, trying to figure out exactly what to do with it, where limits are etc.
The documented fund is not the only fund doing this. Range of shops with a prototype out just different ranges of transparency.
Without terminal access to Claude code it’s difficult as nothing is truly embedded into your system and it’s tricky to integrate MCPs (basically APIs for AI) from stuff like alphasense or positioning etc into it. Some of the better tools are vendor stuff (rogo/shortcut basically mean I don’t touch excel anymore, even for ad hoc small modelling touch ups to see how my numbers are impacted by whatever news update).
Sounds dumb but I’d just watch YouTube videos on setting up agents, MCPs etc to get a feel for how to build skills and the such.
Comment removed
Tldr:
Many shops have an AI analyst prototype
Watch videos on skills/agents to understand how they work and how to build them. Also make sure you’re comfortable with MCPs and integrating data pipelines into your AI tool (Claude code)
Thanks,
But yeah have very recently been told about a particular large MM rolling this out. Using the same kind of terminology you used.
Very heplful. Can you talk about the skills you have set up to build the models? (at a high level)
What’s the most direct knock on effect of all these productivity gains?
PMs running light and fast teams with less analysts?
Analysts cranking through more ideas faster? Is that result in getting more ideas in the book as an analyst because you’re spending more time on idea gen/risk-return eval vs. monkey work?
Analysts working less? (doubt)
None of the above because still too early?
Last question, with all this extra time have you finally gotten laid or still an aspey HF virgin?
lol but they're arguing in the other thread that it is useless. you should look in to local open source models. are you not worried about claude getting all of your training data? feed it more context, it gets better.
At SM. Claude Code changed the game and agreed on mcp integrations w/ data vendors. my MD / sub-PM did not hire another analyst when he was fired and instead is covering the incremental 40 names himself now.
Heard of a pm who has a $750m and just one analyst. 2 ppl pod. think brett is largely right about jr hf roles.
This stuff is helpful, but still can’t get to a decision or action that you can trust.
And you still need people watching the screens and absorbing and distilling all the extra information.
It certainly helps with the basics. And yeah it’s better than most junior analysts. But not senior analysts. It can’t figure out what’s important or the key KPIs or synthesize it all cohesively when everything is changing all the time.
And if you don’t hire juniors, you won’t have seniors in the future.
BUMP
Id molestias vero sunt dignissimos ut et. Qui rem ab odit nihil. Saepe error assumenda veniam cum. Aut magnam aut molestias hic qui suscipit. Ex maiores voluptatem veniam perspiciatis.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...