Robert Wayne

automating myself, one agent at a time

[Robert Wayne @robertjwayne]

@zeroxtlt @samnsci Also super helpful for just keeping track of all of your sessions and which items needed your attention. That way you’re just talking to one thread and not clicking around everywhere

[Robert Wayne @robertjwayne]

For example, I’ve found orchestrators are super helpful for unblocking my agents on redundant questions. Many times my agents want me to sign-off on really insignificant things to get unblocked and orchestrators can easily manage these types of blockers without interfering with the rest of your process (i.e. you can still be in the loop for the rest)

[Robert Wayne @robertjwayne]

@zeroxtlt @samnsci I’d recommend trying firstmate or setting up your own orchestrator. Either way you can just explain to your orchestrator which parts of the process you want to be involved in vs. which decisions are okay for it to make for you

[Robert Wayne @robertjwayne]

@orca_build @noahlisk Thanks @orca_build - The issues that I ran into were with the Orca orchestration skills / commands running through the native Claude / Codex harnesses from within Orca though, not the chat UI.

[Robert Wayne @robertjwayne]

@noahlisk @orca_build (i was using orca for months and just recently switched to T3 for the orchestration reliability improvements)

[Robert Wayne @robertjwayne]

@noahlisk @orca_build once you start doing a bunch of orchestration in orca you will likely start hitting bugginess like failed messages between agents. if you hit those issues, i highly recommend trying T3 code. the orchestration functionality is much more reliable.

[Robert Wayne @robertjwayne]

@arthurcd @theo @kunchenguid okay nice, i'll work off of your fork for now then

[Robert Wayne @robertjwayne]

@arthurcd @theo @kunchenguid @arthurcd are you still using T3 + firstmate? if so, do you still like it more than herdr / anything I can help with to get this merged into firstmate?

[Robert Wayne @robertjwayne]

@ssbrouhard @poteto @kunchenguid @ssbrouhard are you still using this setup?

[Robert Wayne @robertjwayne]

to move away from human review, do you have an opinion on the “best” setup for automated code review? i.e. have you figured out a setup that allows you to fully automate code review, for both code quality (enforcing coding patterns you care about) and finding potential logic issues?

[Andrej Karpathy @karpathy]

Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.

- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.

The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.

So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.

[Robert Wayne @robertjwayne]

@ko1_agmsg @theo I have been grouping my open items by project so that the project lead agents can triage the responses. Even if the session of the lead agent needs to be restarted, or the implementation workers die, it should still be able to recover

[Patrick Collison @patrickc]

Assorted things I've been mulling over regarding agents in the economy:

• Companies are in some sense the original superintelligence. It's often the case that they can use their amortized cognitive surplus to advance their interests against those of certain consumers: relative to the optimization capacity of companies, consumers can be inattentive, underinformed, irrational, etc. (Think, say, of some of the classic insurance upsells at the end of a booking process, which is typically priced substantially above what a regular policy would cost.) Herbert Simon introduced the idea of bounded rationality, and firms often end up benefiting from it. Thanks to personal agents, we will probably get somewhat more rational in how we engage with companies.

• How exactly will outcomes change? Well, certain kinds of price discrimination will work less well (how do coupons work if every agent hunts fastidiously for them?), and certain kinds of customer cross-subsidy will change (today, people who forget to cancel subscriptions subsidize those who don't). Taken across the economy as a whole, I'm not sure what the incidence of all of this will be.

• In corporate strategy, people have long debated the relative importance of distribution and product quality. Perhaps you've made something better, but does anyone know about it? Won't people just continue to buy the ACME Corp Mousetraps? Personal agents will, I think, act as a kind of structural subsidy for product quality. How would markets look different if individuals always spent at least 10 hours researching their purchases? On the margin, focusing on making a product better is, I think, going to become a more effective strategy for companies. This is good!

• It true, a corollary is that information about product quality will become more important and impactful. Today, most people have very limited ability to aggregate and integrate such data, but agents will have insatiable thirst for such signals. (It's often said that data is becoming more valuable because it helps in the training of models. That's obviously true, but I think there's a secondary way in which it's more valuable simply as a complement to cheaper intelligence.)

• A big question is whether agents will increase superstar effects (rational agents all settle on the same product) or increase dispersion (because everyone has slightly different tastes and preferences, which agents are good at eliciting, finally overcoming the artificial clustering we've been subjected to).

• Many internet companies have monetized demand routing, which is in turn often predicated on short attention spans. "I'll click on the first thing I see." This has led to some amount of inefficiency, as companies try to propitiate capricious ranking algorithms and incur various placement fees. It's not all deadweight loss (willingness to pay is itself a kind of signal, which is valuable to buyers), but it's not great. How will the nature and value of this routing change in a world of personal agents? I don't know the answer, but I think that the effects might be pretty big.

• Much of what I'm writing hinges on the assumption that personal agents will be on the consumer's side. I suspect that that will be the winning strategy, but whether it's true, and what it means to be true, involves a lot of other questions.

Taken as a whole, the market is a massively parallelized, decentralized, and highly imperfect value attribution machine, and, taking stock of all of the effects, it seems to me that the quality point may well be the biggest: that agents will change the nature of the backpropagation such that the rewards for better products increase. Maybe this is too optimistic, but I suspect (and hope) that companies that make things that they're proud of will feel that the universe is a little more partisan in their favor.

[Robert Wayne @robertjwayne]

@ko1_agmsg @theo Storing the messages on disk temporarily seems like it would make sense though if T3 isn’t doing that already, I haven’t looked at the code for that part

[Robert Wayne @robertjwayne]

@poteto yeah it’s a good point and the same idea applies to the infrastructure as well, not just code

[Robert Wayne @robertjwayne]

@ko1_agmsg @theo by session restart do you mean when the app itself restarts? if so, I’m not sure how undelivered messages are handled. so far, the messages are reliably getting delivered though, so I haven’t had to deal with a restart while a message is in-flight

[Robert Wayne @robertjwayne]

to provide additional context - I like to have one primary orchestrator so that I can always just communicate with 1 agent. I also find it helpful to group related threads under an orchestrator for that topic.

this way, if I am working on three projects, and each one has 10 agents working on PRs, I don't have to have one orchestrator trying to manage 30 agents working on unrelated things.

instead, the project-specific orchestrators can manage their implementation workers, and those project-specific orchestrators can pass messages up to the main coordinator if something requires my feedback.

in T3, this would show up in the sidebar as 34 sidebar items. in orca, i can collapse it to just the main orchestrator, or show the three project-specific orchestrators, or expand the implementation workers if needed

[Robert Wayne @robertjwayne]

spawning new sessions works fine in orca. the issue is that messaging between sessions is not reliable in orca.

since messaging was unreliable, i wrote custom logic for storing messages locally and making sure that they were successfully delivered.

so far, messaging seems to be working more reliably with T3, which means i don't need to build an intermediate step for storing messages, retrying on failure, keeping track of whether they've been delivered etc.

creating a hierarchy of orchestrators in T3 already works because i just need to keep a track of the parent/child relationships. there's just a UI gap if I want to nest items in the sidebar, not a big issue at all.