Crypto is creeping again into the startup dialog, however at ETHDenver final week, the excitement was as a lot about Washington because it was about tokens. Coverage shifts are rippling by way of the market as Tether and stablecoins face scrutiny, gamers like Stripe re-enter the dialog, and startups both discover traction or flame out. The hype cycle is over, or at the very least taking a break. So what comes subsequent? On this episode of TechCrunch’s Fairness podcast, Rebecca Bellan sits down with Jacquelyn Melinek, CEO of Token Relations and host of the Speaking Tokens and Crypto in America podcasts, to make sense of how the market has modified and what on the earth of crypto is constructed to final. …
Author: Naveed Ahmad
Within the present AI panorama, we’ve grow to be accustomed to the ‘ephemeral agent’—a superb however forgetful assistant that restarts its cognitive clock with each new chat session. Whereas LLMs have grow to be grasp coders, they lack the persistent state required to perform as true teammates. Nous Analysis staff launched Hermes Agent, an open-source autonomous system designed to unravel the 2 largest bottlenecks in agentic workflows: reminiscence decay and environmental isolation. Constructed on the high-steerability Hermes-3 mannequin household, Hermes Agent is billed because the assistant that ‘grows with you.’ The Reminiscence Hierarchy: Studying through Ability Paperwork For an agent…
U.S. cybersecurity company CISA is reportedly in dire shape, in keeping with bipartisan lawmakers and trade leaders who concern that the company’s capacity to carry out its core mission has been diminished and left it unprepared for a cybersecurity disaster. Information website Cyberscoop’s Tim Starks spoke with sources throughout Congress, the non-public cyber trade, and past, and what got here again mirrored a normal consensus that CISA has suffered below cuts and layoffs through the Trump administration’s first 12 months. Over that point, CISA has misplaced round one-third of its workers, which price it applications, personnel, and experience, together with…
Chip large and world’s most useful firm Nvidia reported record profits in its most up-to-date quarter on Wednesday, as demand for AI compute continues to skyrocket. “The demand for tokens on the earth has gone utterly exponential,” CEO Jensen Huang mentioned on a name with analysts following the outcomes. “I believe we’re all seeing that, to the purpose the place even our six-year-old GPUs within the cloud are utterly consumed and the pricing goes up.” The corporate reported $68 billion in income in the newest quarter, up 73% from the prior 12 months, with $62 billion of that income coming…
For the trendy AI developer productiveness is commonly tied to a bodily location. You seemingly have a ‘Massive Rig’ at house or the workplace—a workstation buzzing with NVIDIA RTX playing cards—and a ‘Journey Rig,’ a modern laptop computer that’s good for espresso retailers however struggles to run even a quantized Llama-3 variant. Till now, bridging that hole meant venturing into the ‘networking darkish arts.’ You both wrestled with brittle SSH tunnels, uncovered personal APIs to the general public web, or paid for cloud GPUs whereas your individual {hardware} sat idle. This week, LM Studio and Tailscale launched LM Hyperlink, a…
As AI-powered search instruments reshape how companies are found on-line, India-founded startup Gushwork helps corporations seize clients from platforms equivalent to ChatGPT, Gemini, and Perplexity — with early traction that’s starting to attract investor help. The 2-year-old startup stated Thursday it had raised $9 million in a seed spherical led by Susquehanna Worldwide Group (SIG) and Lightspeed, with participation from B Capital, Seaborne Capital, Beenext, Sparrow Capital, and a pair of.2 Capital. The spherical values Gushwork at $33 million post-money, up from about $7.5 million following its Lightspeed-led $2.1 million pre-seed in July 2023, an individual acquainted with the matter…
def draw_ring(ring: ConsistentHashRing, dist: Dict[str, int], title: str): node_ids = sorted(ring.nodes.keys()) plt.determine(figsize=(8, 8)) ax = plt.gca() ax.set_title(title) if not node_ids: plt.textual content(0.5, 0.5, “Ring is empty”, ha=”middle”, va=”middle”) plt.axis(“off”) plt.present() return G = nx.Graph() for nid in node_ids: G.add_node(nid) for i in vary(len(node_ids)): G.add_edge(node_ids[i], node_ids[(i + 1) % len(node_ids)]) pos = nx.circular_layout(G) vnode_counts = ring.snapshot() labels = { nid: f”{nid}nkeys={dist.get(nid,0)}nvnodes={vnode_counts.get(nid,0)}” for nid in node_ids } nx.draw_networkx_edges(G, pos, alpha=0.4, width=2) nx.draw_networkx_nodes(G, pos, node_size=2200) nx.draw_networkx_labels(G, pos, labels=labels, font_size=9) plt.axis(“off”) plt.present() Source link
Salesforce pulled out all of the stops to persuade traders that the AI revolution gained’t be its loss of life when it announced fourth-quarter earnings on Wednesday. Salesforce reported a stable quarter of $10.7 billion in income, up 13% year-over-year. For the yr, it reported $41.5 billion in income, up 10% over the earlier yr, with each outcomes boosted by its $8 billion acquisition of knowledge administration firm Informatica final Might. Internet revenue landed at $7.46 billion, and the corporate provided sturdy steering for the yr forward, projecting income of $45.8 billion to $46.2 billion — a ten% to 11%…
Within the high-stakes world of AI, ‘Context Engineering’ has emerged as the newest frontier for squeezing efficiency out of LLMs. Business leaders have touted AGENTS.md (and its cousins like CLAUDE.md) as the last word configuration level for coding brokers—a repository-level ‘North Star’ injected into each dialog to information the AI via complicated codebases. However a current research from researchers at ETH Zurich simply dropped an enormous actuality test. The findings are fairly clear: for those who aren’t deliberate together with your context recordsdata, you might be seemingly sabotaging your agent’s efficiency whereas paying a 20% premium for the privilege. https://arxiv.org/pdf/2602.11988…
Anthropic on Wednesday announced that it has acquired Vercept, an AI startup with deep roots to a few of the greatest names in Seattle’s tech scene. The acquisition marks the newest after Anthropic acquired coding agent engine Bun in December to assist scale Claude Code. Vercept had created instruments for extra advanced agentic duties, together with its product Vy, a computer-use agent within the cloud that would function a distant Apple Macbook. Vercept is likely one of the many startups engaged on re-imagining the non-public laptop for the age of AI brokers. As a part of the deal, Anthropic is…