Your tier-1 support team, run by AI and backed by people.
We take over your email and chat queue. Grounded AI resolves what it can and never guesses. Our trained agents handle the rest. You pay per resolved ticket, with an SLA on response time and quality.
Built for support teams that are stretched
Typically 10 to 100 agents, in-house or outsourced, with a large share of repeatable email and chat tickets.
You are paying a BPO by the hour
Costs rise with every ticket and quality is sampled, not measured. Move a queue to per-resolution pricing and QA on every conversation.
You are hiring to keep up
Volume is growing faster than you can recruit and train. Absorb the growth without adding headcount for every spike, launch, or season.
You bought AI and nobody runs it
The chatbot is live but the tuning, reviews, and edge cases land on an already busy team. We run it and own the result.
How it works
We learn your support
We connect to your helpdesk and learn from your past tickets, knowledge base, macros, and procedures.
We agree the scope
Together we pick the queue and ticket types we take over, and set the SLA for response time and quality.
AI first, people behind it
The AI resolves what it can with cited answers. Everything else goes to our trained agents, with an AI-drafted reply ready.
You see every outcome
Weekly reporting on resolutions, response times, and QA scores, plus the knowledge gaps we found and fixed.
AI that knows when not to answer
Speed is easy. Being right is the hard part. IrisAgent is built so your customers never get a confident wrong answer.
Compared with the usual options
| Hiring in-house | Traditional BPO | AI software alone | IrisAgent Managed Support | |
|---|---|---|---|---|
| Time to live | Months to recruit and train | Weeks to months of onboarding | Weeks to configure, then ongoing tuning | About two weeks for a pilot queue |
| How you pay | Salaries, benefits, tools | Per hour or per seat, with minimums | Seats or platform fee, plus your staff | Per resolved ticket |
| Cost as volume grows | Rises with headcount | Rises with hours | Flat fee, but staffing still rises | Falls per ticket as AI resolves more |
| Quality checks | Sampled by a team lead | Sampled, usually a few percent | Up to you | Every conversation, AI and human |
| Risk of made-up answers | Depends on training | Depends on training | Depends on the vendor | Answers only from your sources, or hands off |
| Who runs it | You | Them, on your playbook | You | We do, against an SLA |
Pay for resolved tickets, not hours
One monthly bill based on outcomes. As the AI takes on more of your queue, your cost per ticket goes down.
Where it fits best
Start with one queue
Pick one email or chat queue. We take it over for 30 days, report every outcome, and you only pay for resolved tickets. If it does not beat what you have today, you walk away.
The same grounded AI already resolves real customer conversations for teams like Dropbox, Tekion, and InvoiceCloud.
Questions
What exactly do you take over?
A tier-1 email and chat queue that we agree on together, for example billing questions, account access, order status, or how-to requests. We answer those customers end to end, inside your helpdesk, under your brand. Anything outside the agreed scope goes back to your team.
Who answers when the AI cannot?
Trained human agents working inside your helpdesk, on your procedures and tone of voice. The AI drafts a cited reply for them, so they resolve faster and stay consistent. Every conversation, AI or human, is quality checked.
How is this different from a BPO?
A traditional BPO bills for hours and seats, so cost grows with your volume. We bill per resolved ticket and use AI to resolve as much as it safely can, so your cost per ticket falls over time instead of rising. You also get QA on 100% of conversations, not a small sample.
How is this different from buying AI software?
With software, your team still has to set it up, tune it, review it, and staff the tickets it cannot handle. With Managed Support we own the outcome: the AI, the humans behind it, and the SLA.
Will the AI make things up?
No. IrisAgent answers only from your knowledge base, past tickets, and systems, and cites its sources. When it is not confident, it does not guess. It hands the ticket to a human agent.
How do you price it?
Per resolved ticket. There are no seat minimums and no platform fee. Tickets resolved by AI and tickets resolved by our agents are priced separately, and both are on one monthly bill. We agree the rates and the SLA before the pilot starts.
Which helpdesks do you work with?
Zendesk, Salesforce Service Cloud, Intercom, Freshdesk, and Gorgias. We work inside your existing instance, so your data, macros, and reporting stay where they are.
How fast can we start?
A pilot on one queue typically goes live within two weeks: we connect to your helpdesk, learn from your past tickets and articles, agree the scope and SLA, and start answering.
Prefer to keep your own team on the queue? See managed resolution, where you pay only for AI-resolved conversations, or read the levels of autonomy framework for AI customer support.