Margin is the product
Every per-seat licence for a dialer, a WFM tool and a QA suite comes straight out of programme margin, and you carry them per client, not per platform.
Each client gets their own tenant, knowledge base, branding, compliance posture and reporting, all isolated at the database. You get one platform to run, one team to train, and one bill instead of six.
Illustrative interface. Programme names and scores are sample data, not a performance claim. The screen shows OptiML quality management for an outsourcer: a rail listing six client programmes as separate tenants, an automated scorecard for the selected collections programme covering greeting, disclosure, promise-to-pay capture, tone, calling-window compliance and resolution, and a panel showing that every interaction is scored, that tenant isolation is enforced by row-level security on 231 tables, and which artefacts go into the client audit pack.
An outsourcer buys the same six tools the client could have bought, and then carries them once per programme instead of once across the platform. The economics of that decision show up in every renewal.
Every per-seat licence for a dialer, a WFM tool and a QA suite comes straight out of programme margin, and you carry them per client, not per platform.
New agents take weeks to become useful, and attrition means you are always ramping someone. The first three months of a programme are the least profitable.
You are measured on service level, adherence and quality by clients who check. Sampling 2% of calls is a defensible process right up until it isn't.
Eight things an outsourcer runs on OptiML, each with the mechanism that makes it work.
Routine inbound calls and chats resolved end to end by AI, with complex cases escalating to your agents with the full context bundle attached.
See how voice agents workCollections reminders, upsell, appointment confirmation, win-back and CSAT callbacks through the same dialer and the same agents.
See the dialerEach programme is a tenant with row-level database isolation, its own knowledge, agents, branding, retention and compliance mode.
How isolation is enforcedThe copilot suggests grounded answers with citations, so a new agent performs like a trained one on day two rather than week six.
See the Agent DeskEvery interaction on every programme scored on 15 dimensions, the number your client's audit actually wants.
See quality scoringForecast, schedule, reoptimise intraday and track adherence against each programme's contracted service level.
See workforce managementThe quality/cost router picks the cheapest AI provider clearing your quality floor, and Bhashini's 22 Indian languages carry regional voice at low cost.
See the AI provider layerPer-brand themes, domains and widget styling, so the programme looks like the client rather than like you.
See the channelsSix of them, and none is a separate licence. The dialer, the workforce planner, the quality engine and the desk are part of the platform your programmes already run on.
One tenant per programme, isolated by PostgreSQL row-level security on 231 tables, with its own retention, compliance mode and audit.
Security & compliancePredictive, power and progressive dialing with DNC and consent enforcement applied per programme, not per floor.
Outbound & dialerErlang-C forecasting, scheduling, intraday reoptimisation and adherence tracked against each client's contracted service level.
Workforce managementEvery interaction on every programme scored on 15 dimensions, with calibration, disputes and coaching plans attached to the score.
Quality managementGrounded answer suggestions with citations, customer 360 across every channel, and AI wrap-up so the notes are written before the next call lands.
Agent Desk11 LLM, 8 speech-to-text and 8 text-to-speech providers behind one adapter, with a router that picks the cheapest one clearing your quality floor.
AI governanceSend a call recording, a client scorecard or a policy document from a live programme. We run it on the call rather than showing you a canned demo.
Client audits do not ask whether you have a policy. They ask for the record, the consent, the score and the proof that none of it was edited afterwards.
Tenant isolation enforced by PostgreSQL row-level security on 231 tables, so a defect in application code cannot leak one client's data into another's programme.
Per-tenant retention, compliance mode and audit, so each client contract keeps the rules it was written with instead of inheriting a platform default.
Do-not-call lists, consent state and calling windows enforced per programme, blocking the call rather than logging the breach after it happened.
An immutable audit log with integrity proofs, exportable and signed, for the client audit that arrives with a week's notice.
On Enterprise, the database and recording storage for your OptiML tenancy run in your own cloud account under your keys — one data plane for your operation, not one per client programme. Where a client contract restricts data to the client's own infrastructure, that programme needs its own Enterprise tenancy. Talk to us before you sign it, not after.
Customer-managed data residency is an option on the Enterprise plan, scoped per engagement. The platform itself runs in OptiML Cloud. This option moves your data, not the software. Reference architecture and shared-responsibility matrix available under NDA.
The programme runs as its own tenant. The client's collections policy, dispute matrix and disclosure script are the knowledge base; nothing outside it is answerable. The outbound agent dials from the predictive dialer with the client's calling windows, consent state and do-not-call list attached to the campaign, so an out-of-window number is blocked before the dial instead of appearing in a report afterwards.
A disputed amount escalates to a human on the same floor with the transcript, the entities and the reason for handover already on screen. Every interaction — AI or human — is scored on 15 dimensions against the client's own scorecard, and the monthly audit pack is exported signed. Configuration and mechanism are what this describes. Outcome numbers belong to a named customer, and we do not have written approval for one yet.
We publish outcome figures only with the customer's written approval, so this slot stays empty until one is signed off. In the meantime the mechanism above is the honest version.
Most outsourcers run programmes across several of these at once. The platform underneath does not change; the tenant, the knowledge and the compliance posture do.
The in-house version of the same platform: natural-language IVR replacement, one omnichannel queue and a live supervisor floor.
Contact CentreWhat a BFSI client's risk committee will ask you for: grounded policy answers, immutable audit and data custody.
Banking & financial servicesSelf-service at telecom volume, with cost routing and Bhashini's 22 Indian languages doing the regional heavy lifting.
TelecomEverything on this page is the OptiML CX Platform configured for BPO & Outsourcing, not a separate product. These are the modules it leans on.
Bring a call recording, a client scorecard or a policy document from a programme you run today. We will ground an agent in it and run it live on the call — not a canned demo.
30 minutes, on your own content. No slide deck unless you want one.
Your details stay private. We never share them.