Support Portal Communication platform

AI agents on every channel, in every language

The communication layer your agents are deployed on.

Book a demo See it work
  • 10+ Channels customer-facing and internal, in one place
  • 95% Response accuracy grounded in your own content, with the source cited
    24/7 Coverage answered in seconds, with no shift limits
  • 50% Workload reduction your team left on the work that needs a person
    EU or on-prem Hosting your cloud, your datacentre, or air-gapped
  • 400+ Systems integrated CRMs, ERPs, helpdesks and your own APIs
    Any Languages detected and answered automatically

How it works

An AI agent handles the request, end to end

It replies on any channel, in any language, and answers from your own documentation with the source cited. It stays inside the rules you set, calls your systems to complete the action, and hands to a person with full context when one is needed. Pick a sector and a capability below.

Book a demo

Sources

Website
Help Center
Documentation
CRM
Billing system
Network status
SIM portal
Conversation history

Does

Proves

  • Instant replies on 10+ channels
  • Handles any language automatically
  • Consistent tone across every inbox
  • Around the clock, with no shift limits

Speaks on

Live chat
WhatsApp
Email
Voice
SMS
Instagram
Facebook Messenger
Telegram
LINE
TikTok

Operations

The AI answers. The platform runs the operation.

Around the conversation sits the part nobody demonstrates and everybody depends on: a clock every reply is measured against, rules that move work without a person touching it, and a record of what actually happened.

  • SLA tracking

    First response, next response and resolution times, each with a live status: due, missed, breached. Set per inbox, per team or per customer.

    metduebreached

  • Automation engine

    Five trigger types, more than fifteen conditions and more than fifteen actions. Route, assign, label, escalate, snooze or call a webhook, automatically.

    Trigger Condition Action

  • Satisfaction and quality

    A satisfaction survey fires after resolution, and responses can be reviewed internally, so quality is a number rather than an impression.

Reported on

Every report exports to CSV, over any date range from seven days to a custom window.

  • Bot volume handled by AI, resolution rate, handoff rate
  • CSAT score, response rate, rating distribution
  • SLA hit rate and misses, by policy, agent or inbox
  • Agent messages, response time, resolutions, waiting time
  • Conversation volume, first response and resolution times
  • Live overview open, unattended and unassigned, in real time

Coworker

The same AI, working for your team

Coworker sits in the conversation sidebar and answers to the agent instead of the customer. It reads the thread, drafts, rates, and pulls what the agent would otherwise go and look for. It does not replace anyone: it removes the part of the job that is retrieval.

  • In the conversation, not in another tab

    It works in the panel beside the thread, with the thread already loaded.

  • It shows its reasoning

    Show steps opens the chain behind a suggestion, so an agent can judge it before sending it.

  • Threaded sessions

    Each session is independent, so a new question never costs the context of the last one.

  • Your rules, not its own

    It follows the same guidelines and scenarios configured for the customer-facing agent.

More than implementation

Engineers audit your operation, then build the agent around it

Nobody here hands you a login and calls it onboarding. Real engineers study how your requests actually arrive and where they stall, then configure the platform against that: your channels, your content, your rules, your systems. What goes live is built for your use case, not a template with your name on it.

Talk to an engineer

  1. Audit

    We map how requests reach you today, what volume each channel carries, and where the time is actually lost.

  2. Channel integration

    Every channel connected into your infrastructure: WhatsApp embedding, email routing, the chat widget.

  3. Knowledge base construction

    Your knowledge groups built and structured: docs scraped, PDFs loaded, Help Center connected, retrieval tuned.

  4. AI configuration

    Guardrails, scenarios, tone, custom tools, handoff triggers and escalation logic, set for your use case.

  5. Testing and validation

    Every flow run end to end, replies, handoffs, SLA rules and automations, before a customer sees any of it.

  6. Team training

    Hands-on sessions with your agents and admins: the inbox, Coworker, automations and reporting.

Connections

It plugs into what you already run

Every channel your customers and your teams already use, and the systems the answers have to come from and go back into.

Channels in 10
  • Live chat Embeddable website widget
  • WhatsApp Meta Cloud API, Twilio, 360Dialog
  • Email SMTP and IMAP, Microsoft and Google OAuth
  • Voice Incoming and outgoing
  • SMS Twilio or Bandwidth
  • Instagram DMs, story replies, mentions
  • Facebook Messenger Human agent mode included
  • Telegram Full bot integration
  • LINE Full LINE API integration
  • TikTok TikTok DMs
Systems out 7
  • Model endpoint Point it at any OpenAI-compatible endpoint, on your key
  • Slack Sync conversations, reply from Slack
  • Single sign-on SAML SSO and two-factor authentication
  • Shopify View and manage orders inline
  • Stripe Billing and credit top-ups
  • Webhooks 10 event types, any HTTP endpoint
  • Dashboard Apps Embed internal tools with full context

Anything else connects through Webhooks or a Dashboard App, which embeds an internal tool in the conversation with its full context.

Where to start

Start on one workflow. Widen when it holds.

Nobody hands an AI agent their whole operation on day one, and we would not ask them to. Each step below is the same product, opened wider.

  • Pilot

    One workflow, on one channel, measured against what it replaces.

    • An audit of the operation before anything is built
    • One channel and one knowledge source, live
    • Altores runs the whole setup, no technical team needed on your side
    Scope a pilot

  • Operate

    Every channel your organisation runs, customer-facing and internal.

    • The ten channels, in every language your customers write in
    • The agent reads and writes in your systems of record
    • Handover to a person with the history, the sources and the tone
    • The same agent on internal queues, from IT to HR
    Talk to an engineer

  • Sovereign

    The same platform, inside a perimeter you define.

    • Your own cloud, the European Union, or fully air-gapped
    • The model endpoint of your choice, on your key
    • Audit logs, retention policies, SAML single sign-on
    Plan a deployment

No step is a lock-in: the pilot runs on the platform the other two run on, so widening is a configuration and not a migration.

Questions

What buyers ask before they commit

The seven that come up in every evaluation, answered without the marketing layer.

A pilot starts on one workflow, on one channel, with one knowledge source. It begins with an audit of how requests reach you today, which is also what gives you the before-and-after: without it, nobody can say whether the agent improved anything. We build, test and train on that scope before opening a second channel.
Altores Support Portal is the communication layer where your AI agents are deployed. It centralises every channel your organisation runs, customer-facing and internal, and gives each AI agent what it needs to work: retrieve answers from your own systems and documentation, update records, route a request to the right team, execute the action, and hand over to a person with full context. Complete AI customer service is the first use, not the limit.
It writes as well as reads. It updates records in your CRM and systems of record, and executes actions such as booking an appointment, adjusting a subscription or re-running a provisioning step. Every write is logged in the audit trail with the conversation that caused it.
When a request is outside automated scope, or when a person should clearly take it, the agent hands over with a summary of what happened, the sources it consulted and a sentiment read. The receiving team opens the conversation knowing the history and the tone, so the customer is never asked to explain again.
Yes, and that is the point of it being a communication platform rather than a support tool. The same recognition, routing and execution apply to internal requests arriving on internal channels, from an IT or HR queue to a deploy request in Slack.
Ten: live chat widget, WhatsApp, email, voice, SMS, Instagram, Facebook Messenger, Telegram, LINE and TikTok DMs.
No. Altores runs the full setup: channel integration, connecting your systems of record, knowledge base, agent configuration, SLA policies, automation rules, testing and team training. Your team goes live ready to operate.
In Europe by default, with optional deployment in your own cloud or fully air-gapped on the sovereign AI platform. Full audit logs, data retention policies, contact deletion, and 2FA or SAML single sign-on for access control.

Full product documentation, from deployment to governance, is in the Support Portal guide.

Next step

See it on your own channels

Tell us how your requests arrive today. We will show you what an AI agent does with them, on the channels you already run.

Book a demo Read the docs