Agentic Economy Weekly Thoughts #1
This week I'm watching: marketplace experimentation around offchain tasks, the emerging "wallet wars" for agentic capital, discovery solutions that still feel undifferentiated, and identity standards struggling with spam and legitimacy. Plus a look at what actually makes sense to build right now.
After ~5 months of posting x402 (nearly every week), the recent explosion of new protocols, standards, and ideas has led me to switch to a broader focus on the agentic economy.
Here are some of the latest trends, thoughts, and areas of growth I'm watching in the space.
No. 1 | The Marketplace Revival and Tradeoffs
Interestingly enough, the concept of a marketplace has come in two waves for the crypto industry: NFTs and now agentic tasks. The resurgence of the marketplace has been led by the @daydreamsagents and @virtuals_io teams (in my opinion).
The unique part about agentic marketplaces is the singular trust assumption that underpins much of blockchain -- the reliance on "something" or someone to actually make the call. Why is the concept of an AMM (Automated Market Maker) so powerful? There is no reliance on someone else or something to make the decision.
The cool part about an NFT marketplace is that it can operate similarly to an AMM, where there are two assets being swapped in an escrow-like manner. If the conditions are met, the swap is handled and secured by the underlying blockchain.
So what happens now when an agent is asking for a task that does not occur onchain?
The answer: a lot of experimentation around smart contracts, agentic judging, and human arbitration. I don't think this is a negative -- I actually think it's an area ripe for growth.
One trend I see: @Executi0nMarket is leading the surge in task markets and agentic marketplaces building on top of @SkaleNetwork.
No. 2 | The Agentic Wallet Wars
Agentic wallets are a really interesting idea because they have the same base layer and therefore tradeoffs to keep in mind for developers. A few of the mistakes I'm noticing agentic wallet teams making:
Chain Specific
Only supporting a single chain directly limits your adoption. There are SO many good builders experimenting in their own corners of the industry that by only choosing the "most valuable" chains today doesn't mean you'll be the winner tomorrow. @OpenWallet from @moonpay is currently the best from a support perspective that I have found.
Cross-chain Capital
A problem that persists today for many wallets and their underlying blockchains is cross-chain capital. Users having funds on XYZ exchange or wanting to use ABC onramp is great until they want to go do something (or more relevantly, the agent does) elsewhere. Nearly every action at this point incurs a fee of some level. The onramps and wallets who figure out how to better abstract this for agents have a chance to corner the "non-popular" chains. The @TrailsHQ by @0xPolygon is one such team; they've enabled one-click bridging to SKALE from nearly any chain, tying directly into the onchain elements that power SKALE.
The Human-in-the-Loop Experience
The most nuanced part of AI today: the human. Everyone wants to build agents BUT the agents have to do something for someone. Whether you are building trading agents like @Cod3xOrg (checkout @SkaleNetwork School live stream here on X on 3/31/26 to hear from @uv founder @0xBebis_ on this). This is where the harness and the interactions for agents really come into play. The experience to onboard an agent is very much so "dev-focused" today, and most of the non-developer focused tooling that isn't a CLI winds up being custodial, which then in some cases can limit how developers want to actually build or how users want to consume.
No. 3 | Discovery (Feels Like) It Is Lacking a Leader
A number of teams have expanded into discovery, such as @agentcashdev from @merit_systems, @dexteraisol, @corbits_dev, and 402index from @RyanTheGentry with the largest aggregation.
Natural language discovery tends to be the key (with the behind the scenes being custom per provider) where agents can do something like:
Query: I need to generate an image of a cat
Response: { json data list of services }
Agent -> pays via x402/MPPNote, this design tends to be efficient as shown by Codemode from @CloudflareDev.
However, it feels like agentic builders tend to all converge onto a subset of solutions (often driven by what the LLMs seem to use best, i.e., Next.js, Resend, Postgres, etc.), but the adoption of AI x Crypto crossover is a drop in the bucket compared to that of AI and more broadly development with agentic coding tools.
Time will tell who becomes the market leader for agentic discovery, but for me, it feels like the winner will bring us something a bit more unique than just search and execute tools.
From a technical perspective, I think all these teams are pushing in the right direction, especially as the basic "Bazaar" just doesn't really scale.
No. 4 | Agentic Identity Stagnation
The creation of new agents is occurring at a rapid pace, which is great, but I see one critical flaw with ERC-8004: the ability to curate and create strong feedback is complicated, and spam is hard to avoid.
I'm ideating on this actively, so if you have ideas and want to chat, let me know. I think ERC-8004 and associative standards will continue to evolve and get stronger, but right now, definitely seeing a struggle around how to know what is legitimate vs. not.
Maybe private feedback with public approval is one option? I.e., feedback stays private and can be verified by multiple private parties who are rotated (government style) and that they post public scoring based on that?
No. 5 | What to Build?
If you're looking to build SOMETHING and aren't really sure what? My recommendation is to checkout Conditional Transactions from SKALE, where you can bring automated execution of encrypted solidity functionality right into the protocol. Examples include (but are not limited to):
- Price of token hits X, execute Y
- Liquidation level hit on lending protocol, auto liquidate small portions
- Auction ends -> handle payouts
- Prediction Market closes -> handle payouts
More coming in the next week from SKALE.
A reminder that the above is purely focused on technicals and not legal or financial advice. It is not an endorsement for any tokens. DYOR.