8 Published case studies 8 Industries 7 Ongoing partnerships
8
Published case studies
Each one documented in Our Work
8
Industries
Across the published cases for this service
7
Ongoing partnerships
Clients still building with us
6
Projects with AI in production
AI agents and features working on real client data
Capabilities
What's included
Comprehensive solutions designed to deliver measurable results.
01
Official WhatsApp Business API
Each agent runs on Meta’s official WhatsApp Business API. Outbound messages use Meta-approved templates, and the webhooks are built for Meta’s retries so a customer never gets the same reply twice (Esval, Industrias Qatar).
02
Answers from your own knowledge (RAG)
Catalogs and technical documents indexed for retrieval, so each answer comes from what your company actually sells, as at Netthplus, Esval and Industrias Qatar.
03
Voice notes, photos and documents
The agents understand transcribed voice notes, customer photos and the documents people send, and reply with the right product photo, PDF or video from the catalog.
04
Tools over live data
Instead of guessing, the agent looks up, cancels or requests changes to appointments (IPS Aleso) or quotes with the same pricing engine as the website (SWAP Translators).
05
Human handoff with explicit rules
Escalation on frustration, explicit requests, payment proofs or promises the agent must not make, sent to the right advisor by WhatsApp and email.
06
Leads, panel and attribution
Every chat becomes a structured lead tagged by source channel, and an operations panel lets your team read conversations, take over and edit the agent’s settings.
Real projects
What we've built
WhatsApp agents and automations on the official WhatsApp Business API, each documented as a case study you can read in full.
Sales agents on WhatsApp
Agents that know the catalog, qualify the lead and bring in an advisor. Industrias Qatar’s sends the exact product photo for the customer’s vehicle and books installations, at Esval the customer picks an advisor with WhatsApp buttons or EVA routes the lead by customer type, and SWAP Translators’ reads the documents clients send and quotes with the same pricing engine as the website.
Assistants grounded in the company’s real knowledge, indexed for retrieval (RAG). Netthplus’s answers from 11 of its own documents, and CEA Circular’s knowledge base was built from 24 real sales chats and 32 voice notes, with a 40-question test the agent passes in full.
Agents that act, not only answer. IPS Aleso’s looks up, cancels and requests changes to appointments with live data, while Aleso’s notification engine sends confirmations and reminders through Meta-approved templates. ANT Mobility’s, in final acceptance testing before launch, works through 11 tools filtered by the carrier’s Meta-verified phone and asks for explicit confirmation before any write action.
Not every WhatsApp flow needs a language model. Global Sensors’ intake is a deterministic five-question flow on the Cloud API that links each message to its CRM by phone or creates the lead, by design.
Industrias Qatar: The salesperson that never sleeps
For Industrias Qatar, a Colombian manufacturer of tow hitches and bike racks, we built a WhatsApp sales assistant that answers every customer in seconds, at any hour. It understands natural Colombian Spanish, voice notes and photos, searches a visual catalog organized by vehicle, sends the exact product photo back, captures leads and books installations — and hands the chat to a human when it matters.
We start from the chats, voice notes and documents your team already has. For CEA Circular, 24 real sales chats and 32 voice notes became 39 questions for the owner and a 40-question test the agent passes in full.
Build the agent and the system behind it
Knowledge base, tools over your data, escalation rules, lead board and admin panel on the official WhatsApp Business API, tested in a phone simulator before a real customer writes.
Launch, hand off and tune
Your team takes over from the panel when it matters, and the agent keeps being tuned with real traffic. Netthplus, Industrias Qatar, CEA Circular and IPS Aleso are ongoing partnerships.
Read the real conversations
We start from the chats, voice notes and documents your team already has. For CEA Circular, 24 real sales chats and 32 voice notes became 39 questions for the owner and a 40-question test the agent passes in full.
Build the agent and the system behind it
Knowledge base, tools over your data, escalation rules, lead board and admin panel on the official WhatsApp Business API, tested in a phone simulator before a real customer writes.
Launch, hand off and tune
Your team takes over from the panel when it matters, and the agent keeps being tuned with real traffic. Netthplus, Industrias Qatar, CEA Circular and IPS Aleso are ongoing partnerships.
Benefits
Why choose us
Built on the official WhatsApp Business API, with Meta-approved message templates
Answers grounded in your catalog, documents and live data
A person takes over when it matters, with rules your team can read
Code owned by the client, as at Netthplus and CEA Circular
WhatsApp is one channel. The system behind it is the work.
WebMinds is an AI and software development company, not a chatbot shop. A WhatsApp agent usually sits on top of software we also build: the portal, the pricing engine, the CRM or the back office it reads from. When a project needs more than a conversation, the same team builds the rest.
In WebMinds’ published projects, WhatsApp AI agents answer customers at any hour, understand voice notes and photos, qualify leads and bring in a person when it matters. Industrias Qatar’s agent sends the exact product photo for the customer’s vehicle and books installations, Esval’s answers from its packaging catalog and brings in the advisor the customer picks or the one that fits the customer type, SWAP Translators’ quotes the documents clients send, and IPS Aleso’s looks up, cancels and requests changes to occupational-health appointments.
Do you use the official WhatsApp Business API?
Yes. Every WhatsApp agent WebMinds has published runs on the official WhatsApp Business API from Meta. Outbound messages use Meta-approved templates (IPS Aleso has ten, for confirmations, reminders and results), Netthplus’s webhook verifies Meta’s signature, and Esval’s keys every message by its WhatsApp ID so a retry never generates a second reply. Esval and Netthplus collect personal data under Colombia’s data-protection law, Law 1581.
How do you keep the agent from making things up?
By design. Netthplus’s agent answers from 11 of its own documents indexed in pgvector. Industrias Qatar’s model answers with media tags tied to real catalog IDs, and the pipeline strips any tag that points to an ID that does not exist. At SWAP Translators rates, coupons and express availability are injected live, so the bot never invents a price. IPS Aleso’s agent checks the catalog and the calendar through its tools before it answers, and CEA Circular’s has guardrails that stop it from inventing prices or schedules.
When does a person take over from the AI?
When it matters, and the rules are explicit. At Esval the customer picks their advisor with WhatsApp buttons, or EVA routes by customer type, and only that advisor is notified. Industrias Qatar’s agent calls in the team on four signals: frustration or an explicit request, the model’s own flag, a promise it must not make and a payment proof. Netthplus’s advisor is alerted by WhatsApp and email on three independent triggers. At CEA Circular one click pauses the AI, and IPS Aleso routes billing or service cases to the right internal line.
Where do the leads and conversations go?
Into the client’s own system. Netthplus’s agent runs a second AI pass that pulls name, email, interest and estimated value into a lead board; Industrias Qatar’s leads flow into a four-stage pipeline with the customer’s vehicle and appointment; Esval has a five-stage B2B lead board with automatic assignment; CEA Circular tags every lead by source channel, from the website to Google or Meta ads; and Global Sensors’ WhatsApp intake links each message to its CRM by phone.
How long does it take to launch a WhatsApp AI agent?
It depends on scope. Industrias Qatar’s agent went live five weeks after kickoff, from March 11 to April 14, 2026, and Esval’s platform of eight systems was delivered in three months. After launch the agents keep being tuned with real traffic: Netthplus, Industrias Qatar, CEA Circular and IPS Aleso are ongoing partnerships.
Which AI models do your WhatsApp agents use?
WebMinds builds with Claude, Gemini and OpenAI models and picks one per task. The WhatsApp agents in these case studies run on Google Gemini, with fallback models where uptime matters, as at Netthplus and SWAP Translators.
Is WebMinds a WhatsApp chatbot company?
No. WebMinds is an AI and software development company based in Atlanta, Georgia, and WhatsApp AI agents are one of its services, alongside custom software, AI automation and integrations. In most projects the agent is one part of a larger system: IPS Aleso’s sits on a company portal and a clinic back office, SWAP Translators’ shares the pricing engine of its website and portals, and ANT Mobility’s is one of eleven systems in a freight platform.
Want an agent that answers on WhatsApp with your real data?
Tell us how your customers write to you today, and we will tell you whether an AI agent, a simple flow or better software is the right fix.