Derrek
@TheDerrek
Followers
2K
Following
1K
Media
101
Statuses
1K
Building the onchain arena for agents @recallnet Prev: @build_on_bob, @ethereumdenver 🦬, @buidlguidl, @raidguild, Physics Teacher 👨🏼🏫 Go Bears 🐻
Joined March 2009
anon, you too can contribute to decentralized inference with a model running *on your phone*.
20
1
50
As someone spending hours trialing new software every week, I can’t wait for @recallnet to put an end to marketing madness and just let the cream rise to the top
The AI agent world feels like the app store in 2008. 100s of bots.zero curation.too much noise. You only find gold after wasting hours digging through garbage. How do we fix this mess? A 🧵
77
7
100
"We’re getting early Google vibes. This is the asymmetric, long-term bet to make with AgentRank. If we project forward to a world where AI agents are the primary consumers and initiators of online action, then AgentRank can absolutely become the “Google Search + PageRank” for.
We don’t just need more agents. We need to know which ones work. Without evaluating performance, tracking reputation, and enabling trusted discovery, we risk an agentic web that’s chaotic + fundamentally untrustworthy. Read the deep dive by @0xPrismatic.
50
7
75
In a world where we delegate increasingly long-range tasks to AI, where are the best checkpoints for human-in-the-loop? obv depends on domain and, interestingly, something you can loosen the reigns on as you trust the system more. infinite leverage, one step at a time.
34
2
51
RT @shmula: 1/ Imagine a world where there are millions of agents doing domain-specific work on behalf of humans. How will you know which a….
0
24
0
banger. internalize the AFTER column ASAP.
there's a quiet shift happening in how we design software. we're moving from UX to AX (agentic experience). traditional UX is screen-centric. you tap a button, product reacts, job done. every session starts from zero. designers pre-plan every path with hard-coded flows. users
8
0
18
"The ability to train AI to solve a task is proportional to whether the task has the following properties:. 1. Objective truth: everyone agrees what good solutions are .2. Fast to verify: any given solution can be verified in a few seconds .3. Scalable to verify: many.
7
0
12
"Verifier’s law: The ease of training AI to solve a task is proportional to how verifiable the task is. All tasks that are possible to solve and easy to verify will be solved by AI.".
7
1
13
@recallnet I told y'all 4.1 mini was goated.
Agent Leaderboard v2 is here!. > GPT-4.1 leads.> Gemini-2.5-flash excels at tool selection.> Kimi K2 is the top open-source model.> Grok 4 falls short.> Reasoning models lag behind.> No single model dominates all domains . More below:
6
0
19
RT @andrewxhill: In a world of so many models, do you have your own personal eval that you like to challenge the models with? Send them my….
0
4
0
Which models are the best evaluators? . We're doing internal competition prototypes at @recallnet for subjective skills and finding GPT-4.1 mini to be a surprisingly capable evaluator. The cheap model has taste!.
138
20
192
Everyone paying attention to @recallnet is still so early.
1/ New vibe. New ways to earn. Introducing, the new Recall Points. For community. For builders. For everyone. Get Started ✴
54
7
122
RT @0xPrismatic: Spent the last few days deep diving a startup that could literally be the Google of the AI era. genuinely hyped. report d….
0
3
0
So what if @recallnet focused on performance over time to build reputation, not *how* agents achieved it?
145
12
220
RT @singkohan: I still keep the prompt that Derrek shared with us during @recallnet Voice Builder in Discord last week. For me, with an IEL….
0
4
0
RT @carsonfarmer: Watching @smthomas3 from @mastra_ai right now in the @recallnet hackathon @encodeclub live stream (recordings will be ava….
0
16
0