Agentic Systems DeveloperTaking on new projects

I build AI systems that do the work.

I design and build AI agents, automations, and custom software that connect your data, tools, and workflows — so your team spends less time on repetitive work.

For startups, SMEs, agencies and SaaS teams with work that should run itself.

Example run/inbound-lead.qualify
running
  1. TriggerNew inquiry received from website form
  2. RetrievePulled company context from CRM and docs
  3. ReasonMatched request against your ideal-client criteria
  4. ActDrafted a tailored reply and created the CRM deal
  5. HandoffSent to sales for one-click approval

1 / 5

Proof, not promises

  • Production SaaS

    Building Sociaboost — a multi-client social media platform for brands and agencies — at Audasoft Technologies since 2024.

  • AI in production

    Shipped OpenAI-powered content generation inside a live product, not just a demo.

  • Full-stack delivery

    Interfaces, APIs, backends and mobile apps — React, Next.js, Go, Node.js, Spring Boot, Flutter.

  • 3 companies since 2023

    Software delivered for SaaS, real-estate and data-operations teams.

The real problem

Your business doesn't need more software. It needs less manual work.

  • Repetitive operations

    Your team spends hours every week on work that follows the same steps every time.

  • Information everywhere

    Data is spread across documents, databases, emails, CRMs and tools nobody fully connects.

  • Too many tools

    People switch between disconnected systems and copy the same data from one to the next.

  • Manual decision-making

    Someone analyzes the same kind of information again and again before anything happens.

  • Software that stops at the dashboard

    Your tools show you what needs to happen — then wait for a human to actually do it.

  • →

    That's where agentic systems come in.

    An agentic system doesn't just display information. It understands the request, finds the context it needs, decides what to do within rules you set, and executes the work across your tools — with a human in the loop where it matters.

What I build

Systems that understand, decide and execute.

Every engagement starts from a workflow and an outcome. The technology is chosen to fit — never the other way around.

  • AI Agents

    LangGraphLLM APIsMCP
    The problem
    Some work needs judgment, context and several tools — so simple rules or Zapier-style triggers can't handle it.
    What I build
    Agents that reason through the task, retrieve the right information, call your APIs and execute steps within boundaries you define.
    What you get
    Work that needed a person at every step moves forward on its own — people approve what matters.
    • Reason through tasks
    • Use tools & APIs
    • Retrieve information
    • Execute workflows
    • Decide within boundaries

  • Business Automation

    n8nWebhooksREST APIs
    The problem
    Data gets copied between email, spreadsheets, CRMs and internal tools by hand — slowly, and with mistakes.
    What I build
    Event-driven automations that connect your tools through n8n, APIs, webhooks, MCP, databases and external services.
    What you get
    Fewer copy-paste tasks, fewer errors, and processes that run the same way every time.
    • n8n workflows
    • APIs & webhooks
    • Database sync
    • MCP
    • External services

  • Custom AI Applications

    Next.jsReactTypeScriptGo
    The problem
    Off-the-shelf tools don't fit how your business actually works, and bolting AI onto them only goes so far.
    What I build
    Complete web applications with AI built in — interface, backend, integrations and deployment, delivered as one system.
    What you get
    A product your team or customers actually use, built around your workflow instead of a generic template.

  • Knowledge & RAG Systems

    RAGEmbeddingsVector search
    The problem
    Answers live in documents, databases and people's heads — so the same questions get asked and researched over and over.
    What I build
    Retrieval pipelines that index your documents, databases and structured data, so AI answers from your own sources.
    What you get
    Faster, grounded answers from your company knowledge — instead of generic AI guesses.

  • AI-Powered SaaS

    LLM APIsNext.jsGo
    The problem
    Your SaaS shows users their data — but they still have to do the work themselves.
    What I build
    Agents, AI assistants and intelligent workflows embedded directly into your existing product.
    What you get
    A product that does more of the job for your users — and a stronger reason to choose it.

Show the system

How an agentic system actually works.

Pick a scenario and click any layer. Business explanation first — the technical detail is right underneath.

Scenario

  1. What it does for you

    Everything starts with an outcome you care about — a lead answered, a ticket resolved, a report delivered.

    In this scenario

    Every inbound lead gets a fast, relevant answer — and sales only spends time on good fits.

    Under the hood

    We define the trigger, the success criteria, and the boundaries the system must respect.

    Workflow mappingSuccess criteria

Try it live

Describe a workflow. See the system.

Tell the AI about a task your team repeats. It will sketch the agentic system I could build for it — mapped onto the same architecture as above.

0 / 1500 characters

Or try an example:

  1. Business Goal
  2. AI Agent
  3. Reasoning & Planning
  4. Knowledge / RAG
  5. Tools / MCP
  6. APIs & Databases
  7. Automation
  8. Business Result

Your system will appear here, mapped onto these eight layers.

Work

Case studies.

Real systems, described honestly: the problem, what was built, how it works, and what changed.

FeaturedAudasoft Technologies — Sociaboost · 2024 — present

AI-powered social media platform

A SaaS where brands and agencies plan, generate and schedule social content across many clients and accounts — with AI built into the workflow.

  • AI content generation
  • Scheduling
  • Multi-client SaaS
  • Social integrations
View case study
  1. 1WorkspaceAn agency picks a client and its connected social accounts.
  2. 2CreateA post is drafted — AI generates or rewrites content on request.
  3. 3PreviewContent is previewed per platform before it's scheduled.
  4. 4SchedulePosts land on the content calendar for each account.
  5. 5PublishThe platform publishes to the connected networks on time.

How it works

From idea to working AI system.

A simple, predictable process. You always know what's happening and what comes next.

  1. 01

    Discover

    We identify the workflow that's costing you time or money — and whether it's worth automating.

  2. 02

    Design

    I map the process and decide where AI, automation, APIs and traditional software each belong.

  3. 03

    Build

    I build the system: integrations, agents, interfaces and workflows — with regular demos.

  4. 04

    Deploy

    The system goes live, connected to your existing tools and infrastructure.

  5. 05

    Improve

    We measure results and keep improving reliability, cost and performance.

What should we automate?

Is your workflow worth automating?

Answer five quick questions about one repetitive task. You'll get a rough estimate of the time it costs — and how much could be handed to a system.

How many people do this task?3
How often does each person do it?
2 times
How long does it take each time?20 min
How many manual steps are involved?6 steps
Which tools are involved?

Time spent today

42 h / month

Potentially saved

20–31h / month

≈ 4 working days per month

Automation fit

Strong candidate — this is worth a closer look.

Estimate only. Assumes roughly 48–73% of the task time could be handled by automation, based on the number of steps and tools. Real results depend on your workflow.

Ways to work together

Start where you are.

Three engagement levels. Each one is scoped and priced after a first conversation — no surprises.

  • Most common start

    01

    AI Workflow Audit

    For businesses that know they have repetitive work but don't know what to automate first.

    Includes

    • Workflow analysis
    • Automation opportunities
    • AI opportunities
    • Technical architecture
    • Implementation roadmap

  • 02

    AI Automation

    For businesses ready to automate one specific workflow, end to end.

    Examples

    • Lead qualification
    • Customer support
    • Content workflows
    • Internal operations
    • Reporting
    • Data processing
    • Document workflows

  • 03

    Custom Agentic System

    For companies that need a complete AI-powered application, built and deployed.

    Includes

    • Agent architecture
    • RAG
    • MCP
    • Integrations
    • Custom frontend
    • Backend
    • Deployment
    • Monitoring

Who I work with

Built for teams that want work done, not more dashboards.

  • Startups

    Turn an idea into a working AI product you can put in front of users.

  • SMEs

    Automate the repetitive operational workflows that eat your team's week.

  • Agencies

    Build AI capabilities you can deliver to your own clients.

  • SaaS companies

    Add agents and intelligent automation to the product you already have.

  • Teams with manual processes

    Replace repetitive human workflows with reliable software.

Portrait of Mohamed Besha Frikha

Based in Tunisia · Working remotely

About

One developer. The complete system.

I'm Mohamed Besha Frikha, a full-stack developer based in Tunisia, focused on agentic systems — AI that connects to real tools and data and actually executes work.

Since 2024 I've been building Sociaboost at Audasoft Technologies, a SaaS used by brands and agencies, where I've shipped production features in React, Next.js and Go — including OpenAI-powered content generation. Before that, I built a property-management platform with Spring Boot and Flutter, and data-management tooling with automated cleanup.

Today I help businesses turn manual workflows into AI systems: the agents, the automations, and all the software around them.

Experience

  • Audasoft Technologies — Full-Stack Developer · Sociaboost2024 — present
  • ASM All Soft Multimedia — Full-Stack Developer2024
  • FRS — For Right Solutions — Full-Stack Developer2023

Bachelor's in Big Data & Data Analytics — Collège de Paris International

Why work with me

  • 01

    Full-stack engineering

    I build the complete system — not just the AI layer that someone else then has to integrate.

  • 02

    Agentic architecture

    Agents are designed around tools, context, workflows and execution — not a chat box on top of your app.

  • 03

    Business first

    Technology is chosen for the workflow and the outcome. Sometimes the right answer is a simple automation.

  • 04

    Production mindset

    Reliability, maintainability, security, cost and deployment are part of the design from day one.

  • 05

    One developer, complete system

    Frontend, backend, AI, integrations and automation — one point of contact, no hand-off gaps.

FAQ

Questions founders usually ask.

Start a project

Tell me about the workflow.

Two minutes. The more specific you are, the more useful my reply will be.

What happens next

  1. 1I review your workflow and goals.
  2. 2I reply with questions or a time for a short call.
  3. 3We agree on a first, clearly scoped step.

Prefer to talk directly?

What are you trying to build?
How many people are involved? (optional)
Timeline (optional)
Budget (optional)