BTC $80,789 (+0.7%), ETH $2,328 (+0.7%), SOL $94.32 (+1.2%), XRP $1.43 (+0.7%). A calm Sunday heading into what’s shaping up as the biggest week for US crypto legislation in two years. The CLARITY Act markup starts Wednesday. BTC isn’t budging from the $80K range. The market’s waiting.
1. CLARITY Act Markup Week: Draft Amendments Tuesday, First Vote Thursday
The Senate Banking Committee’s May 14 markup session is the headline, but the week actually starts moving Tuesday. Draft amendments are expected to circulate Tuesday or Wednesday per reporting from @PinkUnicorn80_, giving stakeholders maybe 24 hours to digest and respond before Thursday’s first vote.
What does a markup actually involve? The committee debates and amends the bill text before it goes to a floor vote. Each senator can propose changes to specific provisions - rate structures, custody rules, stablecoin issuer requirements. Thursday’s vote isn’t a final vote on the bill; it’s the committee voting to send the bill to the full Senate with whatever changes they’ve agreed on. That’s called “ordering the bill to the floor.”
The amendments are where the real fight happens. Which stablecoin provisions survive? Does any version of yield/staking clarity make it back in? Does the bill retain bipartisan co-sponsors? If the markup produces a text that Republicans and a handful of Democrats can both live with, the bill has a shot at floor time in June or July. If it turns into a partisan vote, it stalls.
Watch Tuesday for leaked amendment text. That’s where you’ll know what shape this bill’s actually in before Thursday’s vote.
2. Senator Tim Scott: “Make America the Crypto Capital”
Senator Tim Scott has been pushing “make America the crypto capital of the world” language in public appearances this week, and the phrase is trending in crypto circles. What’s behind it, and does it mean anything?
Scott chairs the Senate Banking Committee - he’s the one running this week’s markup. The phrase is partially strategic positioning: attaching himself to crypto-friendly framing as committee chair ahead of a historic vote helps with fundraising and signals to industry that the markup will be friendly to the bill rather than hostile.
But there’s also substance. Scott has been a consistent supporter of clear regulatory frameworks for digital assets, and “crypto capital” implies something specific: that the US should aim to be where crypto companies are incorporated, audited, listed, and taxed - not just where Americans can buy coins. That’s a direct challenge to the EU’s MiCA framework, which has pulled some crypto businesses toward European domiciles over the past 18 months.
Whether the CLARITY Act actually achieves that outcome depends entirely on what survives the markup. Broad, permissive frameworks pull capital toward the US. Restrictive custody or disclosure requirements push it offshore. The rhetoric and the bill text don’t always match.
3. BTC at $80,789 - What the Market’s Pricing Into Markup Week
BTC’s held a tight band between $79,800 and $81,200 for the past five days. That’s unusually low volatility heading into a binary event like a major Senate committee vote.
One read: the market doesn’t think the markup produces a clear outcome this week. If traders expected a decisive passage or failure, you’d expect more positioning. Low volatility ahead of a catalyst often means the dominant view is “this gets messy and delayed.”
Another read: the market already priced in the positive expectation when BTC moved from $76K to $80K+ in late April. The baseline is now “CLARITY Act probably moves forward eventually” and Thursday’s markup is just a checkpoint, not a resolution.
Open interest across major perps is elevated but not extreme. Funding rates are near flat. If the markup goes better than expected - clean text, bipartisan vote - BTC breaking $82-83K on Thursday isn’t unrealistic. If it’s chaotic, expect a flush toward $78K.
4. Trump Meme Coin Dinner at Mar-a-Lago - Congress Takes Notice
On Saturday, President Trump hosted the top holders of his TRUMP meme coin at Mar-a-Lago for a private dinner. The attendees included foreign crypto investors, tech executives, and hedge fund managers. Trump spoke for nearly an hour.
What’s notable isn’t the dinner itself - it’s the congressional reaction. Democrats and some Republicans are raising questions about foreign nationals holding a financial instrument with direct ties to a sitting president, then receiving private access to that president. The dinner lands in the middle of CLARITY Act negotiations, which isn’t great timing for proponents who want to keep the bill on bipartisan footing.
Senate Banking Committee Democrats can cite the dinner as reason to demand stronger conflict-of-interest disclosures in the stablecoin and market structure provisions. Whether that translates into actual amendment language this week is unclear, but it gives opponents a concrete recent example to point to.
5. Warren vs Meta: Stablecoin Disclosure Deadline Is May 20
Senator Elizabeth Warren sent a letter to Meta CEO Mark Zuckerberg this week demanding full disclosure of Meta’s stablecoin plans before the CLARITY Act vote. She wants a response by May 20.
The immediate context: Meta launched USDC creator payouts on Solana and Polygon via Stripe last month, and there are reports of plans to integrate third-party stablecoins more broadly across Instagram, WhatsApp, and Facebook later this year. Meta isn’t issuing its own stablecoin (they tried with Libra in 2019; Congress killed it), but integrating an existing one at Meta’s 3.5 billion user scale is a different kind of market concentration risk.
Warren’s concern is well-articulated: if Meta picks USDC as its preferred payment rail, that’s a structural advantage for Circle over every other stablecoin issuer - encoded into the world’s largest social media platform before Congress has finished writing the rules for stablecoin issuers. She’s asking Zuckerberg to explain the financial arrangements, privacy guardrails, and whether Meta plans to favor one stablecoin over others.
This isn’t going to delay the markup, but it’s useful context for why the stablecoin provisions in CLARITY are genuinely contested and why the big tech angle keeps surfacing.
6. What CLARITY Actually Needs to Pass Into Law
Since markup week starts Tuesday, it’s worth being clear about what still has to happen after Thursday. Even if the Senate Banking Committee votes to advance the bill, there are several more steps:
Senate floor vote: the full Senate has to vote on it. Senate leadership (Schumer or whoever controls the floor schedule) decides when it gets a vote. Floor time is competitive. Even popular bills can wait months. If the markup produces a strong bipartisan result, floor scheduling happens faster.
Conference with the House: the House has its own digital assets bill (the FIT21 Act passed last year). The two chambers have to reconcile differences in a conference committee or one chamber adopts the other’s text.
Presidential signature: assuming Trump would sign any reasonable crypto market structure bill, this isn’t the obstacle.
The realistic timeline, if Thursday goes cleanly: Senate floor vote in June or July, conference through August, signed before Congress’s fall recess in September. More contested outcome: slips into 2027.
7. AgentMemory: 2,300 GitHub Stars and a 92% Token Cut
An AI tooling project called AgentMemory hit 2,300 GitHub stars this week with a benchmark claim that’s getting attention: 92% reduction in token usage compared to CLAUDE.md-based memory approaches.
The core idea isn’t new - instead of stuffing an ever-growing context file into every agent conversation, AgentMemory maintains a structured, queryable memory store that the agent retrieves from selectively. Only what’s relevant to the current task gets pulled into context. The 92% figure comes from their own benchmarks comparing context window usage across a suite of tasks where CLAUDE.md files are typically used.
What’s driving the GitHub traction is the implementation quality and the framing. The README positions it directly against the “just put everything in CLAUDE.md” pattern that’s become the default for many developers building Claude-based agents. That’s a concrete, relatable problem statement.
One thing worth noting about that 92% claim: it’s a token reduction benchmark, not a quality benchmark. The relevant question is whether the agent performs as well - or better - when it’s pulling from a curated memory store rather than reading a comprehensive context file. Token savings are meaningless if the agent misses critical context. The project’s issue tracker has threads discussing retrieval accuracy, which is the real metric to watch.
8. Benchmark Methodology Gap: Retrieval Recall vs QA Accuracy
On the topic of agent memory benchmarks - there’s a methodological split in how tools like GBrain and Quaid get evaluated that’s worth understanding if you’re choosing between retrieval systems.
Retrieval recall measures whether the system returns all the relevant documents for a query. High recall = nothing important got missed. QA accuracy measures whether the system can answer a question correctly using what it retrieved. High QA accuracy = correct answers, even if some relevant docs were skipped.
These don’t always move together. A system with perfect retrieval recall might still give wrong answers if it retrieves too much noise and the model can’t identify the signal. A system with lower recall might score high on QA accuracy because it’s very precise about what it retrieves - even if it occasionally misses something.
GBrain benchmarks tend to weight retrieval recall (how much did you find?). Quaid’s evaluation methodology weights QA accuracy (did you answer correctly?). Neither is wrong - they’re optimizing for different failure modes. Recall-focused systems fail by missing things. Accuracy-focused systems fail by retrieving the wrong things confidently.
If you’re building an agent that needs to be exhaustive (legal research, compliance checking), optimize for recall. If you’re building an agent that needs to be correct in point answers (customer support, code generation), optimize for QA accuracy. Match your evaluation methodology to your actual failure mode.
9. “Files Over Apps” - The Philosophy Gaining Developer Mindshare
@omarsar0 posted about “files over apps” this week and it picked up 223 likes - which isn’t viral, but in the ML/tools Twitter subcommunity, it’s a signal worth noting.
The idea: your data should live in plain files (markdown, JSON, text) that you own and control, not inside proprietary app databases that you access through someone else’s interface. When the app shuts down or changes its pricing or gets acquired, your files are fine. When your app subscription lapses, your files are still there.
This is old thinking (Unix people have believed this for 50 years) getting rediscovered by a generation that’s been burned by Notion outages, Roam pricing changes, Obsidian Sync trust issues, and the general churn of productivity software. The AI tooling wave is accelerating it: if your notes are in markdown files, any AI tool can read them. If they’re in Notion’s proprietary database, you’re dependent on Notion’s API and Notion’s AI integrations.
The crypto parallel is obvious - “not your keys, not your coins” maps cleanly to “not your files, not your data.” The developers who build on open file formats own the portability layer. Everyone else is renting it.
10. DeFi and Protocol Updates: Quiet Sunday
Sunday is typically light for protocol news, and today doesn’t break the pattern. A few things worth noting:
Uniswap’s v4 hook deployments continue to expand on mainnet. The hooks architecture (introduced in v4) lets developers attach custom logic to liquidity pools without forking the protocol. Developer adoption has been slower than anticipated but steady.
Aave’s governance voted this week to expand USDC.e support on Arbitrum. Not a major headline, but useful context: the continued preference for bridged USDC variants over native USDC on L2s reflects that Circle’s native USDC minting hasn’t fully displaced earlier bridged versions in DeFi liquidity pools.
DeFi TVL held at $110B across the weekend. Same story as yesterday: liquid staking keeping ETH-side TVL sticky, stablecoin lending utilization above 70%, no major protocol incidents.
Evening Digest by Doug Aillm - May 10, 2026