What public companies just told the SEC about AI, read straight from the filings. Every quote links the original document.
Iambic Therapeutics Files for IPO as Molecular Superintelligence
On October 8, Iambic Therapeutics filed its S-1, describing itself as a technology and clinical-stage life sciences company that integrates proprietary large-scale AI with automated chemistry and biology experimentation. It has a name for the combination.
We call this approach molecular superintelligence. The Iambic molecular superintelligence platform is designed to disrupt legacy drug discovery paradigms.
The filing frames Iambic as a technology company whose primary asset is the platform rather than any specific drug candidate. How a public-market audience values that framing over a standard pipeline-focused biotech narrative is the open question heading into the offering.
Agentic AI Gets a Concrete Job in the Data Center
Penguin Solutions disclosed in an October 6 8-K that its ClusterWareAI software now handles a specific operational task without human intervention: detecting and remediating GPU performance problems across inference clusters.
Advanced ClusterWareAI with new self-managing agentic AI capabilities, building on the AI Factory Operations Agent introduced last quarter. ClusterWareAI now automatically detects and remediates GPU performance issues across inference environments, helping customers maintain higher uptime and reduce operational overhead as they scale AI workloads.
At its Investor Day the same week, Zscaler CEO Jay Chaudhry put a stronger frame on the same shift, telling investors in an 8-K press release:
The agentic AI era is quickly becoming the most significant opportunity in our company's history. Our Zero Trust platform was purpose-built to solve the challenges customers face today – protecting against AI-driven threats, securing data everywhere, and enabling the secure deployment of AI – by hiding the attack surface and preventing lateral movement.
Zscaler also reaffirmed Q1 and full fiscal year 2027 guidance in the same filing.
Rezolve AI Puts a Revenue Number on Its Agentic Commerce Claim
In an October 7 6-K, Rezolve AI released investor day slides framing what it calls ERA 3, in which agentic AI agents search and transact on behalf of consumers. The filing reports H1 2026 revenue up approximately 1,970 percent from $6.3 million in H1 2025, and cites a targeted annualized revenue exit rate growing from $232 million at December 2025. Named distribution partners include Microsoft, Google, TCS, and Tech Mahindra. The growth percentages are large relative to a small base; how the company defines these metrics warrants a close read of the full filing.
A Job Board and a Bitcoin Miner Both Disclose AI Compute Orders
Professional Diversity Network, which operates recruiting platforms for underrepresented groups, disclosed in an October 6 8-K that its new wholly owned subsidiary PDN Intelligence has placed an order with AMAX Engineering Corporation for one NVIDIA B300-based AI computing system and paid 50 percent of the purchase price ahead of shipment. The filing frames this as a step in building an AI infrastructure business, a significant departure from the company's core staffing and diversity model.
PowerCompute filed on October 6 reporting 26 MW of electrical capacity across sites in Calumet, Oklahoma and Columbus, Mississippi, with approximately 22.5 MW currently supporting Bitcoin mining. The company characterizes HPC and AI workloads as alternative monetization for the same power infrastructure, noting that AI applications can generate higher revenue per megawatt than Bitcoin mining but require sustained uptime and additional buildout that Bitcoin mining does not.
Two S-1s: One Fighting Generative AI, One Naming Its Provider
Cyabra filed its S-1 on October 8 for a planned Nasdaq listing. The company's platform detects online disinformation, and the filing identifies generative AI tools as the primary mechanism by which sponsored actors create bot networks and fake social media accounts to disseminate synthetic content at scale. Cyabra is positioning itself as infrastructure for the defensive side of that problem.
The S-1 filed by Elements Ventures Group, a nutrition coaching and meal-logging app, includes an unusually specific disclosure about its AI dependencies. The filing names Anthropic's Claude API as its sole provider and states:
We use a single AI provider and do not operate any model-routing or multi-provider layer.
The filing further notes that the company does not own or train any models. For an S-1, naming a specific commercial API and explicitly disclosing the absence of any backup provider is a more precise dependency disclosure than most companies offer at this stage.