Meet Astra, Vega and Orbit: Orhan’s AI agents now have a face
Orhan.ai has launched three AI characters, one for each way it puts agents to work: Astra for customer service, Vega for everyday work and Orbit for back-office processes. Each one runs a live example on the homepage, from the first message to the finished job.
Orhan.ai has launched three AI characters, one for each way it puts agents to work: Astra for customer service, Vega for everyday work and Orbit for back-office processes. Each one runs a live example on the homepage, from the first message to the finished job.
Three agents, three jobs, one way of working
Orhan builds AI agents around three kinds of work: talking to customers, helping employees, and running multi-step processes across systems. Until now those were three lines on a services page. Today each one has a face, a name and a personality.
They are not mascots stuck on a banner. On the homepage, each character runs a live example of its job: what comes in, what it does with it, and what you get at the end. A ring around each one fills step by step as the work moves forward, and its expression changes with what it is doing.
- Astra · AI for Service: answers customers on the phone, WhatsApp and email.
- Vega · AI for Work: answers employees’ questions and finishes routine tasks.
- Orbit · AI for Process: carries documents and approvals across your systems.
Astra answers every customer, at any hour
Astra is the star of the front line. A customer calls in Arabic with a billing question. Astra finds the answer in your own policy and quotes it, checks that the fix sits within the limits you set, updates the billing record, and tells the customer exactly what changed. The whole exchange is kept on file.
Anything unusual goes to a person, with the full context passed along, so nobody has to ask the customer to repeat themselves.
- Voice and chat agents in Arabic and English, including dialects and code-switching.
- Under 30 seconds to first answer, at any hour.
- Supervised: you approve sensitive actions.
- Typical time to production: 6–10 weeks.
Vega reads the fine print, so your team does not have to
Vega wears glasses for a reason: Vega reads everything, but only what you are allowed to see. Someone asks in Teams, in plain words, “Can I buy this laptop?” Vega checks the documents that person has access to, writes an answer with the sources linked, and says plainly when it is unsure.
Then Vega does the work: the purchase request is raised in the system your team already uses, and the result comes back in Teams with a record of every step.
- Every answer is cited to the source record.
- Respects existing permissions; it never reads past them.
- You keep the source code, models and evaluation sets.
- Typical time to production: 4–8 weeks.
Orbit keeps an eye on every step
Orbit has the biggest eyes in the family, because Orbit watches the work move. An invoice arrives as a PDF in a supplier email, with no template needed. Orbit checks it against the purchase order, flags mismatches instead of guessing, and sends it to the owner to approve on one screen with the full context.
Once approved, Orbit updates every system in order (the ERP, then the records, then the team) and leaves a trail that shows who approved what, and when.
- Every action is logged, with an audit trail.
- A person approves before any consequential step.
- Every step is governed and reversible.
- Typical time to production: 8–12 weeks.
Their faces tell you what they are doing
The characters are not decoration. Their expressions follow the work. When an agent is listening or reading, its eyes widen. When it is thinking or checking, its eyelids lower in concentration. When it is acting, it moves; and when the job is done, it settles into a contented squint.
They also notice you: move your cursor near one and its eyes follow. It is a small detail, but it makes the point we care about most: an AI agent should be something you can watch, understand and trust, not a black box.
- Ready, listening, thinking, working and done: each state has its own expression.
- A progress ring fills with every step of the run.
- Each agent glows in its own colour: blue for service, yellow for work, lavender for process.
Why we gave our AI a face
Enterprise AI succeeds when people can see what it is doing and step in when they should. Astra, Vega and Orbit make that idea concrete: each one shows its inputs, its steps, the moment a person approves, and the result. Behind the friendly faces are the same principles we build every system on: your data, your permissions, human approval for consequential actions, and a record of everything.
Say hello on the homepage
Visit orhan.ai and switch between AI for Service, AI for Work and AI for Process to watch each character run its example from start to finish. If one of them looks like the help your team needs, start a discovery and tell us about the work.
Sources
- Orhan.ai — AI for Service (Astra): https://orhan.ai/ai-for-service
- Orhan.ai — AI for Work (Vega): https://orhan.ai/ai-for-work
- Orhan.ai — AI for Process (Orbit): https://orhan.ai/ai-for-process
- Orhan.ai — Start a discovery: https://orhan.ai/contact