There was a time when artificial intelligence was a niche topic that drew little mainstream attention. That changed with a wave of disruption.
When OpenAI launched ChatGPT, it ignited intense competition across the tech industry. Companies quickly recognized the massive potential of large language models (LLMs) and rushed to build, refine, and deploy their own AI systems at unprecedented speed. Amid this evolution emerged Manus, an ambitious general AI agent that stood out by emphasizing action over mere conversation.
Unlike traditional chatbots, Manus could independently handle complex workflows, from research and data analysis to building websites or generating reports, often asynchronously in the cloud.
Now, as a company acquired by Meta Platforms, it can tap into the latter's massive resources.
And with that in mind, Manus's ongoing refinement as a practical, user-centric tool continues with one significant enhancement.
And that is allowing users to select specific agent modes when configuring tasks.
Product update: You can now select which agent to run your scheduled tasks for better credit management!
Settings > scheduled tasks > advanced settings
Try it now! pic.twitter.com/cWLkUDfb4S— Manus (@ManusAI) February 2, 2026
For example, users can choose between the more versatile Manus 1.6, the high-performance Manus 1.6 Max for complex tasks, or the lightweight Manus 1.6 Lite for everyday needs when configuring scheduled tasks.
This feature, accessible via settings under scheduled tasks and advanced options, optimizes credit management by matching the agent's capabilities to the task's demands, preventing overuse of premium resources on routine automations.
For instance, users can now set recurring workflows like weekly AI news summaries or Singapore event finders to run at preferred times, with options for repetition, expiration dates, notifications via Gmail or GitHub, and precise agent preferences.
Before this, Manus made its first major data partnership, which is with SimilarWeb.
With the partnership, Manus can embed SimilarWeb's trusted digital intelligence, and route it directly into its workflows. Using SimilarWeb's Model Context Protocol, the AI agent pulls verified web traffic, engagement, and market metrics on-demand via simple prompts, no API keys or logins needed. It draws from users' Manus credits, with a late-January optimization cutting usage roughly in half for many queries, delivering about twice the analyses per credit.
Accessible data covers monthly visits and unique visitors, bounce rates, visit duration, traffic sources (organic, paid, direct, referrals, social), device splits (mobile/desktop), geographic breakdowns by country/region, SEO signals, and e-commerce/app insights where available, often including up to 12 months of historical trends for year-over-year or seasonal analysis.
Manus can then turn those raw numbers into ready-to-use outputs: competitor benchmarks, channel deep dives, market overviews, spreadsheets, charts, slide decks with takeaways, marketing plans, or even micro-websites, all grounded in real SimilarWeb data to minimize hallucinations.
A single prompt can produce a full competitive report, regional traffic visuals, or strategic recommendations in minutes.
Turning complex digital data into actionable insights is why we built Manus.
Together with @Similarweb, we'll be unlocking so much more value for our users.
Huge thanks to @kippbodnar, @searchbrat for the deep dive!
This is just the start of unlocking better decisions with… https://t.co/mXYqNVojGS— Manus (@ManusAI) February 2, 2026
Taken together, these developments suggest Manus is evolving into something closer to a cloud-based AI operator:
Being able to runs on schedules, giving users the ability to choose the computational "muscle" based on task type, pulling live and structured intelligence, and deliver finished work and not drafts, Manus is moving towards a different paradigm from conversational AI as an assistant.
Manus is becoming a system users configure once and let run in the background, which is more or less, an equivalent to hiring a junior analyst who never sleeps, rather than opening a chat window.
And under Meta’s umbrella, the implications are that, the company can have access to vast infrastructure, distribution channels, and ecosystem integrations that could push Manus beyond standalone workflows into deeply embedded enterprise or creator tools.
If early AI was about answers, this phase is about outcomes.
While Manus's enhancements (like agent selection for cost efficiency) are meaningful for practical workflows, the broader class of autonomous AI agents still faces real challenges around reliability, security, transparency, governance, cost, and ethical use.
Regardless, Manus’s trajectory shows the industry inching toward agents that don't just respond, but also operate.