AI Agents on the Way to “Discover New Knowledge”: Are AI Companies Well-Prepared?

One practical step toward readiness is ensuring AI agents can freely gather the online information they need.

Recently, OpenAI CEO Sam Altman has noted that people already treat AI agents akin to junior employees – assigning them work and refining their output – and he predicts that soon such agents will even help humans discover new knowledge on their own. In Altman’s words, “I would bet next year that … we will start to see agents that can help us discover new knowledge…”.

This bold forecast raises a question: are AI companies ready for AI that not only follows orders but autonomously finds insights? Being “well-prepared” means having the right technology infrastructure and efficient processes to support these agents.

Rotating Residential Proxies: Enhancing AI Agents’ Workflow

One practical step toward readiness is ensuring AI agents can freely gather the online information they need. An agent crawling the web at high volume often triggers anti-bot roadblocks – from IP bans and CAPTCHAs to geo-restricted content. Such interruptions severely hinder an AI that’s supposed to continuously learn from data. Rotating residential proxy servers have emerged as a key solution to keep data pipelines flowing. These proxies route an agent’s web traffic through a pool of real user IP addresses and periodically switch to a new IP. In effect, the agent’s requests look like ordinary users’ browsing, not a single bot, which greatly reduces the chance of detection or blocking.

By using rotating residential proxies, AI companies can supercharge their agents’ data-gathering workflow in several ways:

  • Avoiding IP bans: Spreading an agent’s requests across many residential IPs mimics the traffic of many users, helping avoid rate limits or outright bans on any single IP.

  • Bypassing CAPTCHAs: Because the agent’s requests appear to come from legitimate user devices, websites are less likely to flag them as bots. The agent encounters far fewer CAPTCHAs and anti-scraping blocks, allowing uninterrupted data collection.

  • Accessing geo-specific data: Proxies let an AI agent “appear” to be in whatever region is needed. This means it can fetch local information (news, prices, etc.) that would otherwise be off-limits due to geographic restrictions.

Crucially, using real residential IPs gives the agent’s traffic a high degree of anonymity and legitimacy. Experts usually note that rotating residential proxies often provide the best blend of stealth, speed, and flexibility for these AI use cases. By using proxies, AI companies ensure their agents maintain continuous, unrestricted access to vital data sources without getting blocked. In short, this kind of networking strategy helps keep an AI agent’s workflow efficient – enabling it to gather the knowledge needed for novel insights.

Unprecedented Scale and Widespread Adoption

On a broader level, AI companies are ramping up their capabilities to support more powerful, autonomous AI. AI adoption in industry is already mainstream – by 2024, 78% of organizations were using AI (up from 55% a year earlier), and this surge was fueled by massive investment (U.S. companies poured $109 billion into AI in 2024 alone). Major tech players are racing to build ever-larger models and faster AI infrastructure. Nearly 90% of top-tier AI models in 2024 came from industry labs (vs 60% a year earlier). Keeping up with these models is challenging – the computational power required to train cutting-edge AI is now doubling roughly every five months.

To meet these demands, companies are building dedicated AI supercomputers (often via cloud partnerships) and optimizing their software to be more efficient. That’s crucial because a more autonomous AI agent that constantly learns could otherwise rack up enormous computing costs. Preparedness isn’t just about hardware, though – it’s also about organization. While 92% of businesses plan to increase AI investments in the next few years, only about 1% of leaders today feel their firms are truly “AI mature” (fully utilizing AI’s potential). In other words, everyone is investing in AI, but few feel fully ready.

Being well-prepared also means putting guardrails in place. Leading AI firms are establishing ethics teams, safety tests, and other safeguards to ensure that smarter AI systems remain aligned with human goals. If AI agents soon start discovering knowledge or solutions we never imagined, companies will need proper practices to harness those breakthroughs responsibly. The race is on: businesses that wisely prepare their technology, infrastructure, and policies stand to benefit the most from the coming generation of AI “teammates” that can drive true innovation.