Dalmut is expanding AI Automation & Intelligent Agents as a priority UAE service. We are not looking for someone who only connects a generic chatbot widget or produces demonstrations that cannot survive real operations. We need a specialist who can identify a valuable workflow, prepare approved knowledge, design permissions and human handoff, connect business tools, test failure cases and explain the system to the people who will own it.
The work you will own
Projects may include website and WhatsApp assistants, lead-qualification agents, internal knowledge search, document processing, appointment automation, support triage and CRM or ERP workflows. Your responsibility begins with understanding the process and ends with a monitored, documented implementation. The interface is only one part of the system.
You will collaborate with strategy, web, software and client-success teams. You should be able to translate a business request such as ‘automate our enquiries’ into specific triggers, data sources, decisions, actions, exceptions and measurable outcomes. You must also identify cases that should remain human.
Core responsibilities
- Run workflow discovery and define the first use case according to frequency, value, risk and data readiness.
- Map approved knowledge sources, data boundaries, permissions, retention and human escalation rules.
- Design prompts, retrieval logic, conversation states, fallback responses and structured outputs.
- Connect websites, messaging providers, CRM, email, scheduling, helpdesk, spreadsheets or APIs through secure supported methods.
- Build test plans covering accuracy, ambiguity, harmful input, missing information, duplicate records and integration failure.
- Create operational dashboards or logs that help teams review conversation quality, handoff and system exceptions.
- Document architecture, credentials ownership, usage costs, model dependencies and change-management procedures.
- Train client teams and establish a process for updating approved knowledge and reviewing performance after launch.
Responsible implementation standard
The specialist must be comfortable saying no to unsafe automation. Sensitive financial, legal, employment, health or contractual decisions require clear boundaries and appropriate review. Customers should know when they are interacting with automation, and they should have a reliable path to a person. Data should be minimized and access should be granted only where the workflow requires it.
We value practical judgment over novelty. A smaller system that consistently qualifies leads and updates a CRM is more valuable than a complex multi-agent demonstration that no one can maintain. You should be able to compare model, platform and integration choices based on reliability, ownership, cost, security and vendor dependency.
Required experience
- Commercial experience implementing LLM-based assistants, retrieval systems or workflow automation.
- Strong understanding of APIs, webhooks, authentication, structured data and error handling.
- Experience with at least one automation or integration platform and the ability to build custom logic when no connector is sufficient.
- Practical knowledge of RAG, embeddings, evaluation, prompt injection risks, data privacy and human-in-the-loop design.
- Ability to communicate technical limitations without exaggeration and to document a solution for non-technical operators.
- Evidence of a working production implementation, including what was measured and how failures were handled.
Useful additional experience
- WhatsApp Business Platform, approved messaging providers or conversational support systems.
- CRM and ERP integrations, especially lead, ticket, task and appointment workflows.
- Document extraction, classification, knowledge management or reporting automation.
- Python, PHP, Node.js or another language for custom integrations and evaluation tooling.
- Security review, data governance, UAE business operations or Arabic-language assistant testing.
How we work together
Assignments begin with a scoped workflow rather than an open-ended request to ‘add AI’. You will participate in discovery, estimate integration effort and define what must be supplied by the client. Delivery is phased through prototype, controlled test, acceptance and monitored launch. A strong collaborator keeps the system understandable and avoids unnecessary dependencies.
The work is remote with regular overlap during Gulf business hours. You will handle confidential processes, customer information and system access, so disciplined credential management and written communication are essential. Long-term collaboration is available for specialists who demonstrate reliability and responsible judgment.
Application requirements
- Describe two production automations and the business workflow each improved.
- Explain the model, tools, integrations, permissions and human handoff used.
- Provide a diagram, sanitized documentation or demonstration that shows your own work.
- State your availability, location, working-hour overlap and commercial model.
- Include the types of use case you would reject or redesign for safety and maintainability.
Working principles
Dalmut expects senior specialists to make work easier to review, maintain and improve. The role is not limited to completing assigned tasks; it includes identifying missing context, documenting decisions and protecting the credibility of the final client experience.
- Document every integration, permission and escalation rule so another specialist can understand the system.
- Test ordinary, ambiguous and high-risk requests rather than only successful demo paths.
- Communicate limitations directly and recommend a human-controlled alternative where automation would create unnecessary risk.
