A real deployment, made public in 2025. The client is not named when the source does not name it — only PHNIX Card appears by name, because that partnership is public.
Industries
Where Hal-AI agents are already working
Aviation, pharmacy, financial services, logistics, retail, healthcare, HR and education. The foundation is the same in every one of them: agents that hold context, query your own systems and hand the conversation to a person when the case calls for it.
This is not a chatbot running a flow. It is an operational agent trained to do the work.
Three cases published in 2025 — aviation, pharmacy and financial services — plus the platform's other applications.
Service by sector
The same reading for every operation that is live
Publish report
Volume by channel — 7 days
source: api_get_conversations
| Sector | Handled | Resolved | Escalated | State |
|---|---|---|---|---|
| Retail | 1,284 | 1,061 | 223 | Live |
| Healthcare | 962 | 823 | 139 | Live |
| Logistics | 748 | 596 | 152 | Peak at 8 am |
| Education | 506 | 448 | 58 | Live |
| Period total | 3,500 | 2,928 | 572 |
The screen reads the same period two ways: in the chart, the 3,500 conversations split by channel (1,842 WhatsApp, 830 webchat, 552 voice, 276 email); in the table, those same 3,500 split by sector, with what the agent resolved on its own (2,928) and what went up to a person (572). The advantage is not needing one dashboard per operation: switch the vertical filter and the indicators stay the same, whether it is a store, a clinic, a carrier or a school. Names and numbers are fictitious — what the platform delivers is the reading, not the result.
How to read this page
Two labels, no gray area
Hal-AI keeps what has already gone public separate from what the platform can do. The first three sectors are cases published in 2025; the five that follow describe what the platform does in that sector, using the same product pieces — and nothing beyond that.
What Hal-AI agents, Squads and integrations do in that sector. It is not a published case and does not represent an identified client.
Published case · 2025
Aviation: check-in and travel disruptions resolved over WhatsApp
A major Brazilian airline serves passengers over WhatsApp with Hal-AI agents: check-in, travel disruption handling and multilingual support. This is the sector where the cost of a late answer shows up in minutes — a gate closing, a connection missed, a bag left behind.
- Check-in on the channel the passenger already uses — no app to install, no password to remember.
- Disruptions handled inside the conversation — the agent queries the airline's systems and comes back with the alternative, instead of telling the passenger to call.
- Multilingual support — the same agent replies in whatever language it was approached in, with no need for a separate team per language.
- Escalation to a person — when a case goes beyond what the agent should resolve, the conversation moves to an agent with the full history attached.
+40
languages supported over WhatsApp
24/7
no waiting for business hours
Official WhatsApp
Multilingual
Human takeover
Published case · 2025
Pharmacy: the agent inside the system the chain already runs
One of Brazil's largest pharmacy chains integrated Hal-AI Agentic with Alpha 7, the management system used at the counter. The integration has a potential reach of more than 8,000 pharmacies — and that is exactly the point of the case: the agent goes where the operation already happens, not into a parallel channel.
- Integrated with the management system — the agent looks the answer up in Alpha 7 instead of answering from memory.
- Item availability — the most common question at the counter now gets answered over WhatsApp, and verified.
- Scale per location — one agent serves many stores, each with its own context.
- Sensitive matters go to a person — anything requiring professional guidance is routed, never improvised.
+8,000
pharmacies in the potential reach
Alpha 7
management system integrated
The chain is not named — the public source for this case does not name it either.
Published case · 2025
Financial services: payroll-deductible credit with a full audit trail, at PHNIX Card
In partnership with PHNIX Card, Hal-AI brings payroll-deductible credit service to the WhatsApp Business API with a traceable customer journey. Credit is a sector where the conversation is the process: every question asked and every answer given has to stay on the record.
- Official Meta channel — the conversation runs on the WhatsApp Business API, not on an improvised number.
- Traceability — the history stays on the customer record, turn by turn, available for internal audit.
- Context that is not lost — the customer comes back days later and the agent knows where the conversation stopped.
- Handoff to the team — a human agent takes over with the entire history in view.
- Public partnership with PHNIX Card
- Official WhatsApp Business API
- Turn-by-turn record in the CRM
Platform application
Logistics: query the system, bring the answer back
In logistics the question is almost never answered by text: it is answered by the system. The Hal-AI agent takes your operational APIs as tools, checks the load, the route and the status, and brings the answer back into the same conversation — including when the answer is "not yet."
- Queries against your own system — the operation's REST endpoints become tools the agent can call.
- Asynchronous replies — when a lookup takes time, the agent says so and comes back with the result.
- A Squad that follows up on its own — the orchestrating agent wakes up at the agreed time, checks what changed and decides whether anyone needs to be alerted.
- A report published by the agent — the operational summary comes out as a document with tables and charts, delivered by email.
Platform application
Retail: a virtual salesperson who also talks to the owner
In retail the agent serves the customer and reports to the manager. It answers product questions, presents options, checks stock in the connected systems and follows up with whoever stalled mid-purchase — and, on the other side, sums up the day's activity and flags what deserves attention without the owner having to open a dashboard.
- With the customer — product questions, options, availability and follow-up for anyone who dropped off midway.
- With the manager — the day in summary, recurring complaint patterns and customers who almost bought.
- Text, audio and image — the customer sends a photo of the product and the agent understands it.
- Human takeover — a salesperson steps into the conversation whenever they want, without starting over.
Platform application
Healthcare: the clinic's virtual receptionist
Clinics and private practices lose appointments to messages nobody answered. The agent responds on official WhatsApp, explains preparation and paperwork, organizes the appointment request and confirms attendance — and reports back to the manager on what happened to the day's schedule.
Scheduling and confirmation
The agent takes the request, confirms attendance and records everything on the patient's record, with the conversation history attached.
Preparation and paperwork
Preparation instructions, required documents and directions all come from the clinic's own Knowledge Vault.
A clear boundary
Clinical questions are never improvised: the agent routes them to the responsible team along with the context of the conversation.
Platform application
HR: screening and internal answers drawn from company data
In HR the agent works both sides of the desk: it answers the questions employees always ask — policy, benefits, procedure — from the material the company has indexed, and it helps the team screen candidates by querying the connected recruiting system.
- Internal answers with a source — the Knowledge Vault holds the documents the agent is allowed to cite.
- Screening against your own system — profile searches run through the recruiting APIs registered as tools.
- Case-by-case analysis — subagents handle one case at a time, in parallel, instead of returning a statistical summary of the whole set.
- People still decide — the agent organizes and presents; the HR team makes the call.
Platform application
Education: an enrollment assistant that keeps the interest warm
Schools and training providers concentrate the decision into a few weeks a year. The agent answers questions about programs, schedules, paperwork and enrollment steps over WhatsApp, follows up with anyone who stalled midway, and gives the coordinator a picture of who is asking what.
- Program and schedule questions — answered from the institution's official material, not invented.
- Enrollment steps — documents, deadlines and next steps, in the right order.
- Follow-up with prospects — whoever stopped midway gets a nudge from the agent.
- A view for the coordination team — what was asked most and where people gave up.
What stays the same
The sector changes. The foundation does not.
Every industry brings its own vocabulary, its own systems and its own rules. What Hal-AI brings to all of them is the same set: memory and context, an official channel, and a clear path to a human.
Memory and context
Vectorized history, episodic facts, customer traits and a sense of time. The agent picks up where it left off days later, without asking the customer to repeat what they already said. This is the subject of the patent filed with the USPTO in 2025.
Official channel
Hal-AI is a Meta Business Partner and runs on the Official WhatsApp Business Cloud API — and the same agent answers in the webchat on your site and over the phone by SIP, with the same prompt and the same memory.
Escalation to a human
The agent recognizes what it should not resolve alone and passes the conversation to a person, history included. Human takeover is part of the product, not an integration workaround.
Your own systems
ERP, medical records, order management or recruiting systems become agent tools over a REST API or an MCP server. The agent looks things up instead of assuming.
Control over what is autonomous
Parameterized queries instead of open database access, an allowlist of what the agent may do, and a closed recipient list. Security designed for an agent that decides.
Accountability
What the agent did becomes a document: sections, tables with data pulled straight from the API and charts drawn by the server, delivered by email.
The sector defines the vocabulary. Memory defines the service.
What's your sector?
Bring the real scenario — the systems your operation runs on, the volume arriving over WhatsApp, what gets stuck in a queue today. We'll show you how the agent fits into that flow.