The Settings page holds the deployment-wide configuration that every bot, call, and campaign relies on. It is not a per-bot screen — what you change here applies across the whole platform.Open it at /dashboard/settings. The page is divided into sections, and each section saves on its own with its own Save button. Changing one section and clicking its Save button does not touch the others.
Settings is a super-admin screen. The credentials, storage, and infrastructure controls on this page affect every call the platform makes. Treat changes here as deployment-wide and coordinate them with your administrator.
Fleet URLs — the list of voice workers calls are distributed across.
Every section follows the same pattern: change the fields, then click that section’s Save button. While a save is in flight the button reads “Saving…” and is disabled; on success a confirmation message appears and the button re-enables.
The display timezone controls how all timestamps are rendered across the console — call logs, the users table, API key dates, and every other date you see. It changes only how times are displayed; it does not change when anything actually happened.On first load the dropdown defaults to your browser’s timezone. Once you save a choice it is remembered and persists across sessions.
Saves the selected timezone to the platform and caches it locally. Shows “Saving…” while in progress. On success a “Timezone updated” message appears and the button re-enables. Disabled while saving.
For Ori operations, set this to Asia/Kolkata (IST) so call-log and report times line up with your working day.
This section stores the credentials the platform uses to reach the speech, language, and voice engines that power your bots. Each field holds the key for one engine. None of these fields is individually required — you only need to fill in keys for the engines your bots actually use — but a bot configured to use an engine whose key is missing will fail.
Once a key is saved, the field no longer shows the value. Instead it displays a masked placeholder with a Replace button beside it, so a saved credential can never be read back off the screen.
Button
What happens when you click
Replace (per key field)
Clears that field and switches it to edit mode so you can paste a new value over the existing one. Only appears when the field already has a saved key.
To enter a key for the first time, the field is an empty input — just paste the value in. To change a key that is already set, click Replace first, then paste the new value.
Each field accepts the credential for a category of voice engine. The platform refers to these by role rather than vendor.
Field
What to enter
Required
Speech-to-Text (STT) engine key
The API key for a speech-to-text engine — the service that transcribes what the caller says. Paste the key provided for the engine your bots transcribe with.
No
Speech-to-Text (STT) engine key (second engine)
The API key for an alternate STT engine, if your deployment uses more than one transcription provider.
No
Language model (LLM) key
The API key for a language-model engine — the service that decides what the bot says next. Paste the key for the LLM your bots run on.
No
Language model (LLM) gateway key
The API key for an LLM gateway, if your bots reach language models through a routing service rather than a single provider directly.
No
Language model (LLM) key (alternate engine)
The API key for an additional LLM engine, if your bots use more than one.
No
Language model (LLM) service-account JSON
The full contents of the cloud service-account JSON key file for an enterprise LLM platform. This field is a multi-line box in a fixed-width font. Paste the entire contents of the .json key file, not just a single key string.
No
Voice / Text-to-Speech (TTS) engine key
The API key for a text-to-speech engine — the service that turns the bot’s words into spoken audio.
The API key for an alternate TTS engine, if your deployment uses more than one voice provider.
No
The service-account field expects the whole JSON file, pasted in full — open the .json key file, copy everything from the opening { to the closing }, and paste it into the box.
Saves every key field to the platform. Shows “Saving…” while in progress. On success an “API keys updated” message appears and the button re-enables. Disabled while saving.
Keys are stored on the backend and are never sent back to the screen in full — once saved, you can only Replace a key, not view it. If you lose a key, you must paste it in again from your own records; the console cannot show it to you.
This section configures answering-machine detection (AMD) — how the dialler listens to a just-answered outbound call to decide whether a real person or a voicemail machine picked up, before it attaches the call to a bot. Getting these values right keeps conversation minutes from being spent on voicemail greetings.These are deployment-wide defaults applied to every outbound call. For what AMD does and how to read its results on a campaign, see Answering-machine detection.
The defaults below are tuned to work out of the box. Only change them if you are deliberately re-tuning AMD and understand the trade-off between catching more machines and accidentally dropping real people.
All time-based fields are in milliseconds (ms). Each field has a minimum and maximum; if you clear a field and click away, it snaps back to its default rather than staying empty.
Field
What to enter
Default
Allowed range
Required
Total Analysis (ms)
Total time spent analysing the greeting and silence before deciding.
2000
500–5000
Yes
Initial Silence (ms)
How long silence is allowed before a greeting is expected.
2000
500–5000
Yes
Greeting Limit (ms)
Maximum time allotted for the greeting message.
2000
500–5000
Yes
After Greeting Silence (ms)
Silence expected after the greeting ends.
1000
100–3000
Yes
Minimum Word (ms)
Shortest sound counted as a word.
120
20–1000
Yes
Between Words Silence (ms)
Silence expected between detected words.
50
20–1000
Yes
Max Words
Most words to detect in the greeting before deciding.
4
1–12
Yes
Silence Threshold
Audio level below which the line is treated as silent.
256
64–2048
Yes
Max Word Length (ms)
Longest a single word can last before it is no longer treated as one word.
Saves every screening field to the backend. Shows “Saving…” while in progress. On success a “Dialler settings updated” message appears and the button re-enables. Disabled while saving.
This section tells the platform where to store call recordings, using S3-compatible object storage. If this is not configured correctly, recordings are not saved and will not be available for playback.
The storage endpoint URL, for example https://minio.example.com. Required for the platform to connect to storage. If cleared, recordings stop being stored.
Yes
Bucket
The storage bucket name recordings are written into, for example recordings. If cleared, recordings stop being stored.
Yes
Region
The cloud region code, for example ap-south-1. Required only for hosted AWS S3. Leave blank for local storage or DigitalOcean Spaces.
No
Access Key
The storage access key ID. If cleared, recordings stop being stored.
Yes
Secret Key
The storage secret access key. After the first save it is shown as a masked value with a Replace button — click Replace to paste a new secret.
The Secret Key works like the voice-engine keys: once saved, it is masked and accompanied by a Replace button. Click Replace to clear it and enter a new secret. This prevents the masked placeholder from being accidentally saved back as the real value.
Saves the storage configuration. If you are about to clear any critical field (Endpoint, Bucket, Access Key, or Secret Key), a confirmation appears first — “You are about to clear: [field list]. Recordings will not be stored until these fields are reconfigured. Continue?” — and you must confirm. On success a “MinIO config updated” message appears and the Secret Key is masked for display. Disabled while saving.
If the Endpoint or Bucket is left blank, the platform skips upload entirely and no recordings are stored — there is no recovering audio for calls made while storage was unconfigured. Confirm storage is set before running any campaign you expect to review later.
This section holds the list of voice-worker addresses that calls are distributed across. The platform uses this list to load-balance outbound calls; adding a worker here lets it take part in handling calls.You can add, edit inline, and remove URLs in the list, but nothing changes until you click Save Fleet URLs.
Adds the address in the New Fleet URL field to the list and clears the input. Only works when the input is not empty. You can also press Enter in the input to add.
Delete URL (X icon, per row)
Removes that worker’s URL from the list immediately.
Save Fleet URLs
Saves the whole list to the platform. Shows “Saving…” while in progress. On success a “Fleet URLs updated” message appears and the button re-enables. Disabled while saving.
Adding, editing, or deleting a URL only changes the on-screen list. None of it takes effect until you click Save Fleet URLs. Avoid listing the same worker twice — duplicates are accepted but serve no purpose.
To check whether the workers in this list are actually healthy and how much call capacity they provide, use the read-only Fleet status page.