AI agents for customer service, every call reviewed, every trend traced to its accounts.
Hand Duale AI your tickets, calls and escalations. Long cases return as sourced case files, each with a proposed resolution. The work runs in the background, and your team decides on every result.Customer analysis, themes ranked by growth across tickets and calls
Each theme comes with its volume, its change from last month and quoted examples.You hand over: “Why did complaints rise in September, and for which accounts?”
How the Task splits it: one branch per theme or key account reads the month’s tickets and call transcripts through Tools. Each branch then compares them with last month’s analysis, kept in your Library. Code execution computes each theme’s volume and change.
You get back: one table of themes, sorted by growth, for your next service review.
Illustrative example: the September analysis, with fictional accounts.
- Theme
- Billing errors after the price change
- Tickets and calls in September
- 236
- Change from August
- +61%
- Quoted example
- “We were charged the old price and the new one.” Ticket 58213
- Accounts most affected
- Linden Home, Hazel Sports
- Theme
- Late deliveries
- Tickets and calls in September
- 412
- Change from August
- +38%
- Quoted example
- “This is our third late order this month.” Call on September 18
- Accounts most affected
- Linden Home
- Theme
- Password resets
- Tickets and calls in September
- 180
- Change from August
- Stable
- Quoted example
- “I have been locked out since the update.” Ticket 58377
- Accounts most affected
- Spread across accounts
Your team decides: the fixes, who owns each one, and which customers to contact first.
After-call review, every call checked against your quality grid
Your team leads see the checks each call met or missed, and coach from them.When it starts: your code starts one Task when each call ends.
How the Task splits it: focused branches classify the reason for the call, check the call against the quality grid kept in your Library, and draft the wrap-up.
You get back: one review per call, with its reason, its outcome, each quality check met or missed, and a wrap-up ready to file.
Illustrative example: the review of a call about a damaged delivery.
- Field
- Reason
- For this call
- Refund request for a damaged delivery
- Field
- Outcome
- For this call
- Refund agreed once the customer sends a photo of the damage
- Field
- Quality checks
- For this call
- Identity confirmed: met. Next step confirmed with the customer: missed
- Field
- Wrap-up
- For this call
- A three-line summary of the call
Your team decides: coaching and escalations. Here, the team lead sees the missed check and follows up with the adviser.
Escalation case files, each dispute ready for a decision
Advisers settle multi-order disputes from one file, with a resolution proposed under your refund policy.When it starts: an adviser asks your team’s assistant about a case too long for a chat. The assistant is an application your company built on the Platform, and it hands the case to a Task.
How the Task splits it: one branch per order reads the order, its delivery records and the related tickets through Tools. Each branch also checks the order against your refund policy, kept in your Library. The branches then converge on one case file.
You get back: one case file that cites each order, delivery record and ticket it relies on. It proposes a resolution and names the policy passage behind it.
Illustrative example: a refund dispute that spans several orders, their deliveries and past tickets. The proposed resolution reads: “Refund order 4417 in full; the carrier’s record confirms the damage on delivery. Refund policy, section 3.2.”
Your team decides: the resolution and the reply to the customer.
The capabilities behind your service work
Each one is set up once on the Platform and serves your other use cases too.Tools
Your agents reach your ticketing, telephony and order systems through your own code or MCP servers. Code execution does the counting.
Libraries
Keep your quality grid, refund policy and past analyses in a Library. Each result quotes the passage it relies on.
Usage analytics
See how many reviews your agents complete each day, and follow each branch of a Task in the Dashboard Task view.
Before you hand over your first calls
The facts your security, finance and procurement colleagues check.One fixed price per Task
A standard Task costs €0.10 and a fast Task €0.20, excluding VAT, whatever the number of branches or the duration. See pricing
Your providers, your contracts
Your model providers bill you directly for inference.
Your choice of hosting
Use the managed service, hosted in the European Union, or run the Platform in your own infrastructure.
Encrypted storage
All stored data is encrypted, from Library files to database backups.
Unlimited agents
Add agents as your teams and use cases grow.
Your content remains yours
Duale AI does not reuse your content to train a model. See subprocessors
Customer service questions
No. Your advisers keep their phone system, tools and routine; your own code starts each review when a call ends. See how IT and security teams start one Task per event, or read how your code starts a Task.
Yes. Each review applies the grid of the team that took the call. See how Libraries hold each team’s rules.
Yes. Each case file lands on the right ticket, so the adviser opens it where they already work. See how Tools connect your ticketing tool.
Yes. Tickets, call transcripts, emails and chat logs all count, so each theme shows its volume across every channel. See how Tools reach the systems that hold them.