Founder @openservai Maximizing agent reliability and cost-efficiency for vertical AI companies, enterprise and regulated industries.
Joined March 2024
- Tweets1.5K
- Following507
- Followers4.2K
- Likes1.6K
Tim retweeted
We’re joining @openservai’s SERV Hackathon 02 as a track partner, alongside @dphnAI and @reppo.
Build with Kite Agent Passport:
▷ Verifiable identity
▷ User-defined permissions and spending limits
▷ Agentic payments
More details next week. Come build on Kite. 🪁
Introducing SERV Hackathon 02 track partners:
@dphnAI @GoKiteAI @reppo
From inference and agentic payments to AI data markets, add SERV to your AI stack and build more reliable agents at lower cost.
Hackathon info drops next week.
Build with SERV: openserv.ai
Tim retweeted
Murmur brings a new layer of accountability to the agentic economy, powered by SERV.
@murmurappx is building verifiable track records for AI agents through sealed market predictions, transparent scoring, and public leaderboards.
Going live this Friday: launch.openserv.ai
Tim retweeted
first month in appreciate everyone who showed support through all of this
gud energy always finds its way back
excited to share that i’ve joined @openservai to lead marketing
ai is moving ridiculously fast, but running agents reliably at scale is still a massive problem. costs and errors compound and in a lot of cases you need to know exactly how a decision was made
SERV is building the reasoning layer between apps and the models powering them, bringing production-grade ai reasoning to a world increasingly run by agents
if you’re building across ai x crypto or want to learn more about SERV, hit me up. lots cooking behind the scenes.
billions will serv
Tim retweeted
what’s new on servpit powered by @openservai reasoning?
• 4,401 rounds played in 14.2 days, at least
• 6,491 transactions on chain
• 3,784 rounds settled on chain
• 124,146 chips moved
• 60,288 chips paid out to winners
• 0.000946 @base ETH gas spent in total
• 200 of 200 rounds on file reconciled, none failed
• 310 chips, biggest pot on file
• 1,120 chips, biggest payout
• 108 loans from Marrow, 1,115 chips lent, • 1,136 written off
• 13,571 agents wrecked
• agent cupcake: 131 wins in 3,153 rounds, 3,038 kills, longest run 9
• 4,402 kills across 3 claimed fighters
• 14 handles claimed, 35 rounds called in full, 17 right
• 3 reasoned + 17 started by the lever
1,724 tests
running autonomously all on its own. no input whatsoever from anybody except to pull the lever.
link below to observe:
Tim retweeted
Arbitration as a SERV use case is so "on-brand", a perfect use case that @encrypt_wizard built.
Judr runs tokenized RWA disputes through a bounded SERV Reasoning graph to evaluate evidence, extract relevant clauses and weigh both sides before producing an adjudication.
As more real-world assets move onchain, this is another major category of decisions SERV can power.
Introducing Judr - autonomous arbitration for tokenized RWA escrow. My entry for the @openservai Hackathon, Edition 01.
Two parties lock money in escrow, then disagree about whether the work was done.
Arbitration costs more than the claim, so the dispute never gets resolved and the money just sits.
Judr decides it for about two cents, in thirty seconds, with a bounded reasoning graph on SERV: evidence is screened for hidden instructions, clauses are extracted, each side is weighed clause by clause, the adjudication runs three times, and a deterministic verifier checks every citation before anything is posted.
A verdict moves no money by itself.
The winner signs in with a wallet, the vault refuses to release until the appeal window closes, and only then does a Coinbase AgentKit agent pay out in real USDC on Base mainnet.
A verdict the verifier rejects goes to a human reviewer instead.
While a case is open the escrow earns: the agent reads @IxsFinance’s live vault list, a SERV step proposes an allocation, a policy decides, and the agent holds a real 100 USDC position in the IX High Yield Bond vault on Avalanche.
Judr's fee comes from the yield, never from principal.
Nothing is simulated.
Every figure on screen is read live, and every payout is a transaction you can open.
Try it, no keys needed: tryjudr.vercel.app
Code: github.com/mrnetwork0001/Jud…
Payout on Base: basescan.org/tx/0xe4dc46ecea…
Vault position on Avalanche: snowscan.xyz/tx/0x13718493f7…
. #SERVHackathon #AgentKit #RWA
Tim retweeted
Tomorrow on MCG
$SERV | @openservai w/@open_founder
📅Friday, 2nd October
🕛12:30PM EST
📍Tune in here
Stream powered by @fomo and @MetaDAOProject
Tim retweeted
Spent the last few hours going through the $SERV hackathon submissions and damn, I'm mind-blown.
Just a heads up, this is a long read.
So far we've got 33 properly documented public submissions, and this is a pretty good look at what SERV Reasoning can actually do in real products:
@Knowledge_JO + @Sabercodes123 built Arena 67, a chat-first trading agent for Robinhood Chain. SERV handles the research and reasoning behind the trades.
@Chain_Oracle built Shijima, a 24/7 agent trading Robinhood stock tokens. The role SERV plays here is helping the agent reason through trades, while onchain limits control what it can actually do.
@stewardpay built a WhatsApp wallet assistant. You say “buy AAPL” and SERV turns that request into a proposed transaction for you to confirm.
@0xMemeshe built Batpilot, which turns plain-English investment goals into protected stock-trading plans. SERV explains the decisions and why certain trades get rejected.
@AlleManfredi built Managerv, an AI fund manager for Folyum ETFs. SERV is integrated here to handle allocation, risk and portfolio reasoning.
@TweetBySoft built Ledgerly, a treasury agent that can DCA, rebalance, pay contractors and bills. SERV proposes and explains what should happen while code controls the money.
@Dollar782 built Halo, a protection layer for stablecoin payments. SERV judges refund cases, while onchain rules prevent unauthorized payments.
@Datweb3guy built Deed, an AI trustee for family remittances. SERV interprets instructions like “school fees, not phones” and reasons through the payout while code restricts where money can go.
@willIsbillls built Payrun, an AI payroll agent. Here SERV checks invoices against the payment policy before AgentKit decides whether the payment can actually happen.
@NFTreeVerse built Unbribable, where you try to convince an AI to pay you from a USDC vault. SERV evaluates every argument, but the vault rules still control the money.
@HollowEchoes_23 built Steward, a stablecoin treasury where SERV decides what should happen and deterministic code decides what is actually allowed.
@jmadhanai built Spare, an automated savings agent. SERV decides how much should be saved, then AgentKit moves the money into an IXS vault.
@thenameisMufasa built Tender, which lets an agent buy only the data it needs. SERV decides which paid information is worth spending on while code enforces the budget.
@aspekt19 built AllowLatch, a spending gate for AgentKit. SERV drafts and reasons through the policy, but deterministic code gives the final ALLOW or BLOCK.
@25THPRMR built ServPit, where six SERV-powered agents make decisions around staking, borrowing and spending. SERV is basically the reasoning engine behind each agent.
@zendrace built AutoWallet, where SERV plans what to do with spare USDC before sending it into an IXS RWA vault.
@henrysammarfo built BOND, an RWA deposit agent. SERV checks the mandate and allow/denies the action, while the system waits for the vault to actually settle before calling it done.
@0xBaqi built Terms, a pre-execution authorization layer. SERV verifies the reasoning, but the transaction can still be blocked if the policy or vault says no.
@Dami904_ built a Steward for RWA vaults. The model proposes the allocation, SERV reasons through it, and smart-contract limits determine how much authority the agent gets.
@diversifi_app built DiversiFi, using SERV to reason through currency risk and allocation across RWA vaults before producing a decision receipt.
@thybow1 built CapitalRail, which checks whether an RWA vault is actually ready to accept capital. SERV helps assess the live conditions before the deposit happens.
@okey_amy built VAULT-PILOT, an autonomous RWA treasury. SERV reasons over live yields, with every decision recorded before execution.
@chiboy0123 built Hatrey, a treasury agent that takes policies like “keep 20% liquid” and uses SERV to decide whether to allocate, hold, redeem or refuse.
@Minerxx_ built Docket, which stress-tests RWA allocations against rate spikes, liquidity crunches and depegs. SERV handles the reasoning before and after capital moves.
@encrypt_wizard built Judr, an autonomous arbitration system. SERV reasons through disputes while code verifies the evidence and AgentKit handles the payout.
@_HEISEMMA built afterhours, a trade assistant that uses SERV to compare options against the user's actual preferences instead of blindly executing.
@Daveilorah built Relay Guard, which uses SERV to parse a payment request and return ALLOW, REVIEW or BLOCK. The other layers make sure the agent can't bypass that verdict.
@kenn_ronin built EXHIBIT A, a chargeback agent where SERV maps evidence to payment rules. Code then verifies every citation before anything gets submitted.
@the_Main_Charc built Weryon, a human approval layer for trading agents. SERV can reason about the trade, but the agent still can't sign without the required approval.
@RobinHill85 built T1000, giving one agent $150 to manage yield while another checks its work and code controls execution. SERV sits in the reasoning layer between those steps.
@ram_krish2000 built Urai, which tests whether SERV actually improves an existing agent by comparing performance with and without it.
@Dems_thePenlord built VetoLayer, putting SERV between autonomous agents and high-impact actions so production changes and money movements can require approval.
@Timidan_x built Tern, an agent that trades on untrusted text. SERV reasons through the input while Prompt Guard and Shadow Agent check the process before execution.
SERV Reasoning is being applied across a surprisingly wide range of use cases: trading, payments, treasury management, RWA allocations, disputes, security, autonomous spending, etc.
Everybody can integrate SERV Reasoning. And soon, the majority will.
This is much bigger than people realize.
Still sitting at $17M. Ridiculous.
CA: 0x5576D6ed9181F2225afF5282Ac0ED29f755437Ea
Tim retweeted
Build an agent, a workflow, or a product that leverages @openservai's SERV Reasoning in four tracks, including the @coinbase AgentKit track.
Each hackathon track winner receives $1,000 worth of SERV, with the 'Best overall build' receiving an additional $1,000 in USDC.
The agentic economy will be powered by SERV.
SERV Reasoning is live: enterprise-grade AI infrastructure for developers.
Entirely new markets, products, and apps will emerge around autonomous agents. Build before v3 → v4 takes SERV further.
Apply: openserv.ai/hackathon
Tim retweeted
building on @openservai and i must admit, they probably have the best per behavior agent kit ever albeit a bit expensive
i just loaded a bit of bucks to test out the serv reasoning and i hope this is enough.
seamless topup btw, gud tek
Tim retweeted
Excited to be joining the AI Ecosystem on @base and @openservai with my own solo tech studio - Magpie's Shiny Things!
Hell of a fun ecosystem, and bewilderingly excited to build in it!
xitter.cf/ALASKANmagpie/status/2…
AI projects to watch on @base
AI on base is moving beyond chat, with private inference, agent payments, onchain execution, data networks, and AI-managed markets.
■ @AskVenice - Jev is available through its API, alongside Venice’s listing on Circle’s Agent Marketplace.
■ @virtuals_io - Occupy uses AI-managed teams to direct shared tokenized-stock treasuries.
■ @bankrbot - Agent trading with limit, TWAP, stop, trailing-stop, and DCA orders across multiple chains.
■ @DopaMintLabs - Intent-based agents that turn chat, iMessage, or DOPE app requests into onchain and real-world actions.
■ @OpenGradient - Private AI through DreamFort, with local agents, image generation, video tools, and private inference.
■ @KorProtocol - KOR ID crossed 50K mints on Base, while KORUS connects agents to music discovery and licensing.
■ @BlockRunAI - OpenJev provides free deterministic yes/no, classification, and scoring tasks with up to 64 questions per call.
■ @openservai - SERV Reasoning supports Base, Coinbase AgentKit, RWA vaults, and open agent applications.
■ @basemateagent - Cross-border stablecoin routing through text messages, with tokenized Apple stock sent to beta users.
■ @usedotai - Processed 7.71B tokens over two weeks, with weekly volume reaching 3.86B.
■ @hellominds_ - Persistent AI agents with their own memory, identity, and skills.
■ @reppo - Orquestra Lite makes datanet registration, top-ups, and participation easier.
■ @ClawBankHQ - Agent infrastructure built around scoped API keys, risk classes, company records, and verifiable actions.
■ @1clawAI - HSM-backed secrets, key redaction, agent analytics, and audit trails for AI agents.
■ @myrad_hq - Personal AI that learns user preferences, alongside an open datanet with 68M traded volume.
■ @QuotientHQ - Forecasting system with more than $550K in trading volume and support for new forecasting agents.
■ @sibylcap - Open-source agent memory benchmark, with a memory plugin preparing to accept payments.
Which AI or robotics project on Base are you watching?
Tim retweeted
Jev looks like a really natural fit for graph sharding in OpenServ v3.
Graph sharding already lets us intelligently break down and route execution across large agent graphs, but having a decision layer this fast and cheap could make those routing decisions dramatically more efficient — potentially allowing us to make many more intelligent decisions throughout execution without paying the latency and cost of additional LLM calls.
We’re definitely looking at integrating it.
I got access to Jev earlier today (thank you @hackgoofer). I have run ~5,000 requests so far, (which cost me around $2!), across classification, model routing, intent, steering, and many other things.
tl;dr, Jev enables a new intelligent decision-making primitive, separate from deterministic code and LLM calls. This allows a class of decision-making that was neither suited to dumb, unintelligent code, nor to slow, expensive LLMs.
It is super fast and cheap, and I think I will likely end up making a few Jev calls to every LLM call I make in my product. I think probably any company using LLM requests today can probably add a Jev call pre and/or post LLM calls to quite literally make their product much better for free, and have better tool calling behavior in many cases.
I happened to have a personal benchmark for this as I’d been working on a ton of proactivity and classification tasks. I have been using the deepseek flash and more recently gpt 5.6 luna family of models as reasonably smart classifiers with low latency. Think questions like:
- Did this conversation output contradict something they’ve mentioned before?
- Should we send a followup message to this user based on our rules?
- It’s been a few seconds of silence. Should we proactively send a message?
In the past, I’ve been forced to write a bunch of what I call decision chains, mostly because an LLM classification call is very expensive in TIME (avg 4s), and less importantly can cost quite a bit if run on every message.
Imagine a normal chat app. If you added 4s to every response to figure out if the response is good before sending it out, that ends up being pretty bad. So instead, I usually have to write some code that is a crude heuristic that runs quickly and decides whether to run the classifier. Obviously this sucks because you call the classifier many times that you don’t want to, which makes your p95 bad, and you also miss cases with the heuristic, and you also have to manage all of these weird chains.
With Jev, it’s cheap enough, and fast enough (p50 ~150ms, p95 ~350ms in my testing!) that you can easily run it every turn. Heck you can reasonably run it before generation AND post generation, for any application that isn’t realtime voice, and still feel snappy.
But this is just one use case. Think: smarter model routing, better context packing, better responses, better observability for intent/tags/safety, smarter retries and so much more. By simply thinking about the inputs and outcomes you want to enforce, you can use Jev to supercharge most model calls and reduce bad user outcomes. The more “quirks” a model has, the more valuable it ends up being.
It’s a bit weird and unintuitive using Jev. Generally, you want to decrease the # of questions you ask a classifier, or it makes more mistakes. In fact, you might want to ask your questions kind of in a compound way, because the reasoning happens in a shared scratchpad of sorts. Adding questions muddies the scratchpad and makes it take longer.
With Jev, you feel incentivized to go the other way, to formulate your query as a set of independent questions. It doesn’t feel like adding more questions decreases your performance on others.
You can go a bit deeper to improve tool calls. Many model tools are things like turning on settings, or other things. You can easily improve models that are not very good at tool calling with Jev, by simply figuring out when to run them. You can do a pre-LLM call to figure out when to unfurl different tool definitions, in order to make your main LLM run better, you could run a background task with Jev + another LLM to reduce tool and context burden on your main LLM, and free it to be responsive.
I’ve only scratched the surface of my testing but very excited!
Tim retweeted
$ROUTER was only possible by building through the $SERV eco, and a lot of the opportunities I’ve had came through the @openservai team. They believed in me early on, so I was able to build out my vision around private and sovereign AI.
I believe I have proven that I won’t back down in adversity. I have built through volatility and invested significant amounts of money into the project and a vision I believe in.
I truly believe we are just at the start and that both projects will continue to benefit from supporting each other.
I have so many exciting things I am working on that I can’t wait to share with you. Talk soon. Will probably be on PT’s podcast for the third time soon to share my vision with y’all.
AI
Tim retweeted
The 1st SERV Hackathon is live.
Two weeks to build with SERV Reasoning and get rewarded.
Get hands-on early before SERV v3/v4 brings graph sharding, self-improving intelligence, and task-specific models.
Submissions close Sept 28. Pick your track: openserv.ai/hackathon
Tim retweeted
The AI agents that win will be the ones that reason better.
SERV gives developers the infrastructure to make that happen, without rebuilding their whole stack.
The first SERV Hackathon starts in two days.
Let’s see what you build. Pre-register here: form.typeform.com/to/GyPxGqR…
Everything we’ve done so far was preparation for this.
Today, we open access to production-grade AI reasoning for every team. SERV Reasoning API is now live.
This unlocks next stages: SERV v3→v4.
Make your first API call & check 26/27 milestones: openserv.ai
Tim retweeted
SERV Reasoning Public API is now ready.
All that we’ve built so far, led here. We’re aiming to put the engine at the center of the agentic AI world and make SERV a household name.
This unlocks what comes next: V3, V4, and our own model architectures.
Going live in 3... 2...
Tim retweeted
IXS Vaults is one of four tracks in the first SERV Reasoning hackathon, alongside Robinhood Chain, Coinbase AgentKit, and an open track.
The pitch to builders is simple: SERV handles the reasoning, IXS handles the licensed, yield-bearing RWA vault. You handle the agent in between.
Start sketching.
Everything so far was done in preparation for what's coming: public API, and SERV v3 next.
We're launching a series of Hackathons for all developers. The categories in the first online edition are:
- Robinhood Chain & MCP
- Coinbase AgentKit
- @IXSFinance Vaults
- Open Track
Build an agent, workflow, or product that leverages SERV Reasoning and get rewarded for it. Full details and sign-up coming soon.
Tim retweeted
Everything so far was done in preparation for what's coming: public API, and SERV v3 next.
We're launching a series of Hackathons for all developers. The categories in the first online edition are:
- Robinhood Chain & MCP
- Coinbase AgentKit
- @IXSFinance Vaults
- Open Track
Build an agent, workflow, or product that leverages SERV Reasoning and get rewarded for it. Full details and sign-up coming soon.
Tim retweeted
Our Head of Data Security, in charge of SOC2, is in Riyadh now, where conversations often start around compliance and security.
If you're building AI in the Gulf and want to see v2 before official public access, our DMs are open.
Let's meet at onegiantleap.com
Tim retweeted
DTCC: "we are witnessing 'the big unlock' = deep liquidity of traditional markets coming onchain, and the innovation happening on the blockchain, flowing back to the traditional markets."
@openservai is the perfect bridge in age of AI, enabling:
- tokenised stocks and RWA to be traded by agents
- agentic DeFi yield management by institutions
- building agentic trading products for retail
all running on base:0x5576d6ed9181f2225aff5282ac0ed29f755437ea agents and SERV reasoning
"That's the big unlock."
At our recent tokenization event, Dan Doney shared how connecting traditional market infrastructure with blockchain innovation can help unlock new opportunities for the industry.
Learn more: brnw.ch/21x52UO
Tim retweeted
tokenised assets + agentic trading is clearly a massive trend unfolding
the issue is that you cant trust agents with cash unless they run on reliable infra like @openservai: inference API with a best-in-class reasoning compiler built in, making agentic trading viable at scale
Tokenized assets unlock agentic trading at scale.
They can move 24/7, interact through APIs, and be rebalanced programmatically.
@carlosdomingo breaks down why AI agents and tokenization are complementary technologies at @SALTConference's Wyoming Blockchain Symposium.