Camlin Connect · No-AI mode

Run Camlin Connect with no AI in the decision path.

No-AI mode is a platform setting, not a promise. Turn it on and the platform refuses AI while your team builds, when a change is saved and when the service runs. Everything a regulated service needs keeps working.

In No-AI mode, for a given published version, input and data, the outcome is the same.

One turn, heard, interpreted and decided, with AI offThree boxes joined left to right. Heard: a speech bubble. Interpreted: a tag with a tick. Decided: a fork where one branch is taken. Under them a switch labeled AI off.HeardInterpretedDecidedAI off

What No-AI mode is

A setting your administrators control

One setting at three levels, so the rule holds whether you use our platform or install it in your own environment. Wherever it is on, a version with an AI step cannot be saved, and an AI step that reached a running service is stopped before it runs.

  • For an organization

    A setting your administrators control and we can lock, with every change recorded in the audit log.

  • For a whole installation

    One switch turns No-AI mode on for every organization on an installation in your environment.

  • In your license

    Installs in your environment can carry No-AI mode in a signed license file verified locally.

What still works

Everything a regulated service needs

No-AI mode removes AI steps, not capability. These all keep working.

  • Intent matching from phrases and patterns, with spelling and mishear correction
  • Menus, answered by keypad or by voice
  • Forms with built-in grammars for phone numbers, dates, numbers and email
  • Conditions, switches and variables
  • API calls with success and error routes
  • Camlin Server data: collections, lookups and counters
  • Knowledge answers read from your sources with a citation, not composed
  • Transfer and handoff to your people, SMS and email

How decisions are made

Heard, interpreted, decided

A fixed pipeline, the same for every turn, with every step recorded.

  1. 01

    Normalise

    Spelling and common mishears are corrected against your vocabulary, and every correction is logged.

  2. 02

    Match in fixed stages

    Exact phrase, then anchored pattern, then containment, then word overlap: the first match wins.

  3. 03

    Report the stage

    Confidence is the stage that matched, not a probability.

  4. 04

    Follow the approved path

    The published release your team approved says what happens for that intent.

What we do not claim

Plain limits

  • Speech recognition and synthesis turn speech into text and back. They do not decide outcomes, and switching them on or off does not change how decisions are made.
  • If the avatar speaks with one of our stock voices, that voice is made by a speech provider outside the United States. That does not change how decisions are made. The privacy policy lists what is kept where.

How you verify it

Check it yourself

You do not have to take our word for it. The evidence is in the product.

  • A trace of every turn

    What was heard, each correction, the intent and the stage that matched, and the path that followed.

  • A test for every path

    Run across channels before you publish, so every route from Start to End has been exercised.

  • Release history

    Immutable releases you can compare, see who approved, and roll back in one step.

  • The audit log

    Who turned No-AI mode on or off, who locked it, and when.

Prefer AI?

Use it where it helps

No-AI mode is one setting. Most services use AI where it helps and rules where they must decide, and you can change your mind later.

  • Let AI draft the service from a plain-language brief, on models you host if you prefer, then review and approve it.
  • Turn AI on per step type, on your own keys, so the relationship with the model provider stays yours.
  • Switching No-AI mode on or off is an administrator change, recorded in the audit log.
  • AI steps go through the same checks, tests and approval before a release goes live.
How Architect works →AI in Camlin Connect →

Questions

Frequently asked

Can we use AI to design a service and keep the running service free of AI steps?

Yes. Design-time AI is separate from the runtime. Your team can let AI draft and review in Architect while the published service runs with AI steps refused. Or turn No-AI mode on for the organization and build with templates and commands only.

Does it work on voice?

Yes. Phone, visual IVR, web chat, SMS and the avatar run the same approved design. Speech recognition turns the caller’s words into text; the decisions after that are made the same way as in chat.

Does the avatar change anything?

The avatar presents and speaks the reply the approved rules chose; it does not decide what happens next. For a No-AI service, set its mood reactions to off.

Can No-AI mode be locked?

Yes. We can lock it for your organization so it cannot be switched off from inside, and the lock is recorded in the audit log.

Is No-AI mode on the Australian platform too?

Yes. The same platform runs in Sydney and in N. Virginia.

Deployment, identity and governance are on the enterprise and public sector page.

Next step

See it run with no AI

We will show No-AI mode on a service like yours, and walk through the trace, the tests and the release history.

Enterprise and public sector →