What happens when entrepreneurs stop asking “how can AI help me with this task?” and, instead, start asking – “How can AI run my entire operation?”
One man figured out a solution – using Claude Code and custom agent workflows to build an entire team of multiple departments.
Manthan Jethwani, the founder of Phaze AI, demonstrated his newest creation – an AI-powered operating system, complete with a central ‘AI CEO’ dashboard, helped by specialist agents. He claims that the tool now handles the majority of day-to-day operations in his business with absolute ease.
He shared a glimpse of this system in action on an Instagram video – a single dashboard where AI scans leads, invoices, team performance, and content performance every morning, then recommends precise next actions. With multiple specialist agents taking care of sales, HR, finance, marketing, operations, and more, a single AI tool can essentially function as a multi-employee company – all while keeping the human factor at the core.
AI CEO dashboard explained
In a casual conversation with Financial Express, Manthan describes his creation as a powerful productivity tool where AI agents work together to orchestrate a company.
“It’s just a single dashboard where an AI CEO can scan an entire business every morning — leads, invoices, team, and content — and recommend what to do next,” Manthan told Financial Express. The AI CEO connects to meeting recordings, tasks, and company data, giving it full visibility into operations. Under the CEO sit seven specialist agents that take care of departments like sales, HR, finance, operations, marketing and others.
However, the key element here remains the ‘human approval’, post which a task is executed.
“This isn’t basic automation or a fancy chatbot. The system runs on live data (using tools like Supabase combined with advanced AI models) and operates with structured autonomy. Each department agent performs its role independently but funnels decisions through the central AI CEO for review,” he tells us.
And he built this AI CEO dashboard using Anthropic’s Claude Code. Think about it – a Claude Code-based tool can substitute up to 8 entire departments of people.
Before vs after: Founder finds time
Manthan says that the impact of the AI CEO dashboard on his daily workflow has been dramatic. Traditionally, founders like him spent up to 5 hours buried in administrative work, chasing leads, planning content and reviewing invoices. “With systems like this, that can be reduced to under an hour through centralised briefings and automated execution,” he stated. The time doesn’t disappear; it shifts into higher-value work that only the founder can do,” he added.
He elaborated on how the tool helps him begin the day on a much more productive note. His ‘AI employees’ usually start the day with a priority brief, highlighting risks, opportunities, and next actions.
“For example: identifying revenue at risk from cold leads or suggesting content based on recent performance. It can also summarise overnight industry developments and competitor activity. The goal is to give a complete, actionable snapshot before the workday begins,” he says.
Do tools like these revolutionise single-person businesses?
Manthan believes that for solo founders, one-person startups, and independent creators, this kind of an AI operating system can be truly transformative. Many solo entrepreneurs struggle with wearing too many hats — handling sales, marketing, client servicing, invoicing, and content creation all at once.
Manthan’s setup allows a single person to operate with the efficiency of a small team without hiring full-time staff. It not only automates repetitive tasks, but also catches important details that might otherwise slip through the cracks and frees up mental bandwidth, thus enabling creators and solo entrepreneurs to focus on what they do best – creating value, building relationships, and driving growth.
In essence, it levels the playing field, letting one highly capable individual compete with much larger organisations.
Does that mean no humans anymore?
Despite the impressive level of automation, Manthan is careful to highlight that the system is designed with clear boundaries. When asked if the AI has ever made a decision he had to correct, he explained the philosophy behind the setup:
“These systems are designed with controlled autonomy. Each agent operates within role-based access, but critical decisions like contracts, proposals, or client actions require human approval. The aim isn’t to replace decision-making, but to automate the analysis behind it,” says Manthan. “Human oversight remains central,” he adds.
Hence, it is the “human-in-the-loop” approach that keeps the founder firmly in control of the system, while the AI dashboard offloads repetitive and analytical work.
Vibe coding to make it easy to build systems like this in future
One of the major concerns with an advanced tool like this is tailoring something like this for your business needs. Manthan says that it took him intermediate technical knowledge, APIs, databases, and knowledge about structured prompting to build one that suits his business.
“However, the barrier is falling quickly as tools improve. It’s likely that non-technical founders will be able to implement most of this within a short time in the near future,” he added.
As vibe coding picks up, Manthan says that no-to-low code AI orchestration platforms make such AI operating systems/dashboards more accessible to a much wider range of business owners.
There are things that AI can’t replace
Even though Manthan pushes the boundaries of automation as a software developer, he is clear about the irreplaceable elements of business – relationships and human judgment.
“Relationships and human judgment remain essential. Conversations with clients during high-stakes or emotional moments require trust and nuance. While AI can analyse patterns and suggest actions, creative instinct and leadership decisions still sit with the human. The system supports the business, but doesn’t replace its leadership,” says Manthan.
Manthan’s setup, built heavily using Claude and custom agent workflows, demonstrates