Our work

Case studies

One real deployment, described in full with the client redacted. Below it, three earlier engagements and a set of clearly labelled illustrative scenarios with no figures.

Case study · In production

Multi-agent email coordinator for a US financial services firm

Client
US financial services firm, lead-driven, under licensing restrictions
Channel
Email, live. Voice on the same engine, in pilot.
Runs on
Salesforce, Velocify, Microsoft 365, Webex, Freshservice

Before

Inbound leads and client email arrived into a shared inbox. A coordinator read every message, decided who it belonged to, and routed it by hand. Senior reps received leads under a weighted rotation that existed only in a spreadsheet. Follow-ups were tracked by memory. The firm operates under restrictions on what unlicensed staff may discuss with customers, so every outbound message carried regulatory exposure.

What we built

An agent system that classifies inbound email, routes it under the firm's actual rotation rules, including weighted allocation and the distinction between original lead owner and reassigned opener, and tracks follow-up to completion. Suspicious messages become IT tickets for human review rather than AI decisions. Every routing and messaging decision is logged with its reasoning.

The hard part

Compliance cannot live in the prompt. Instructions to a model degrade over a long conversation, so a separate validation layer inspects every output before it reaches a customer and blocks anything touching restricted topics. Prompt-level guardrails weaken. An output check does not.

Integrations

Salesforce, Velocify, Microsoft 365, Webex, Freshservice, hosted on the client's existing cloud.

Status

In production. Engagement expanded beyond original scope at the client's request, including a new module for their Salesforce and Webex workflow.

“Great work from Salah! He's really dedicated to his work and provides amazing solutions for AI development and connections. I would highly recommend him for any of your online development needs.”

Verified Upwork client · 5.0 · Top Rated Plus

Want the same for your inbox?

A 30-minute call about how inbound work moves through your team today. Written plan and fixed price within 3 days.

Book a call

Other completed engagements

Workflow Automation2025

AI Content Workflow

Srishna · Visual news platform · India and Australia

Srishna publishes daily news as visual cartoons across India and Australia. We built an AI workflow system to support their editorial and content production pipeline, reducing the manual effort required to produce, format, and distribute content at their daily publishing cadence.

On-Premise AIFebruary 2026

Local LLM Infrastructure

Private deployment · Technology company

A company needed a large language model running entirely on their own hardware - no cloud connection, all processing local. We optimised and deployed the full infrastructure, tuning model performance for their specific use cases while keeping all data on-premise.

“Results were nothing short of fantastic.”
AI Product2025

Property Analysis AI

Real estate operator · United States

Built an AI system for a US real estate operator that evaluates properties against market data and produces structured investment assessments, replacing a manual analysis process that previously took days per report.

Illustrative scenarios

Representative accounts of what each type of project involves. These are not client projects, and they carry no figures.

Workflow Automation · Illustrative

Legal Research at a Law Firm

Research briefs are the firm's quiet bottleneck. Preparing a single brief means a partner or senior associate spending four or more hours searching case law, reading precedents, and assembling the argument structure - supported by paralegals working on research full time. Client work queues behind it, and the most expensive hours in the firm go into document retrieval rather than legal judgment.

What is at stakePartners bill at the highest rate in the firm. Every hour spent locating precedents is an hour not spent on strategy, client counsel, or new matters. At a typical firm's volume, research alone can consume a large amount of senior time every year.

What we build
  • A research system with access to the firm's full case-law library that reads a research question in plain language, locates the relevant precedents, and drafts a structured brief with each citation linked to its source.
  • A mandatory partner review step: nothing the system produces reaches a client or a court without a named lawyer reading and approving it. The system prepares; the lawyer decides.
  • A complete audit trail. Every brief records which sources were consulted and why, so the firm can demonstrate the basis for any document it produces.

How it rolls outWeeks one and two cover the document library and security review - the firm's case files never leave its own environment. Weeks three to five test drafts against briefs already produced manually, with partners scoring the output blind. By week six the system is in daily use for live matters, with the partner review step as the permanent quality gate.

What you ownThe firm owns the entire system and runs it on its own infrastructure. No per-seat licences, no usage fees, no dependency on Verel Systems for daily operation.

Knowledge System · Illustrative

Policy Knowledge Base at a Financial Firm

The compliance team is the firm's single source of truth for policy questions - and its single point of congestion. Staff email questions about internal policy, regulatory requirements, and procedure; the compliance team searches thousands of documents manually and answers hours later. Regulation prohibits any of this material from leaving the firm's premises, which rules out every cloud product evaluated.

What is at stakeSlow answers delay client onboarding, transaction approvals, and audit responses across the whole firm. The deeper risk is inconsistency: under time pressure, different compliance officers occasionally give different answers to the same question.

What we build
  • A private knowledge system installed entirely inside the firm's own building, on its own hardware. Nothing - documents, questions, or answers - ever touches an external network.
  • Coverage of all policy and regulatory documents the firm holds. Staff ask questions in plain language, in Arabic or English, and receive a precise answer citing the exact document and page it came from.
  • A review dashboard for the compliance team showing what is being asked and answered, so they supervise the system rather than answer each question by hand.

How it rolls outThe first two weeks cover installation and document ingestion inside the firm's premises, with their IT security team observing every step. Weeks three and four are a supervised trial: the compliance team compares the system's answers to their own before any staff access. General rollout comes in week five, department by department, with the compliance team retaining authority to correct or annotate any answer.

What you ownThe firm owns the system outright and its own staff maintain the document library. We have no ongoing access to their premises or their data.

AI Reception · Illustrative

AI Receptionist at a Medical Clinic

One front-desk employee handles every inbound call alongside walk-in patients. During consultations, calls ring out; after hours and on weekends, they go entirely unanswered. Thirty to forty percent of after-hours calls are lost - and each lost call is a patient who books with the clinic next door.

What is at stakeFor a private clinic, a missed call is a missed booking, and a missed booking is revenue that does not return. The clinic is effectively closed to new business for two-thirds of every day, while paying for advertising that generates calls nobody answers.

What we build
  • An AI receptionist that answers every inbound call immediately, day or night, and speaks naturally with patients in Arabic or English - whichever the caller uses.
  • Direct connection to the clinic's appointment calendar: it offers available times, books the appointment during the call, and sends the patient an SMS confirmation before they hang up.
  • Judgment about its own limits: medical questions, emergencies, and unusual requests are passed to staff with a recorded summary, never improvised.

How it rolls outWeek one maps the clinic's booking rules - consultation types, durations, physician schedules, insurance questions. Weeks two and three run the system in parallel: it answers, but staff monitor every call and can step in. Full operation begins in week four, starting with after-hours calls only, then extending to daytime overflow once the clinic is confident in it.

What you ownThe clinic owns the system. Call-line running costs are paid directly to the telephony providers at usage rates - a small fraction of a staffed line - with no margin added by us.

Workflow Automation · Illustrative

Invoice Processing in a Finance Department

Every supplier invoice - PDFs, scans, photographed paper - is read by a person, checked against purchase orders, and typed into the accounting system. The finance team loses hours each day to this, and the error rate runs around eight percent: wrong amounts, wrong codes, duplicates. Errors surface weeks later as payment disputes and reconciliation work.

What is at stakeBeyond the daily lost hours, errors have a compounding cost: supplier disputes, late-payment penalties, and month-end closes that drag. The team is due to grow with invoice volume - automation is the alternative to a new hire.

What we build
  • A processing system that receives invoices in any format, reads them, matches each against its purchase order, and enters verified invoices directly into the company's accounting system.
  • A confidence rule: invoices the system is certain about flow straight through; anything unusual - a mismatched amount, an unknown supplier, a possible duplicate - is held for a person to review, with the discrepancy highlighted.
  • A complete audit log: every invoice records what was read, what it was matched against, and who approved it.

How it rolls outWeeks one and two connect the system to the accounting software and train it on historical invoices. Weeks three and four run in shadow mode - the system processes every invoice but a person verifies each entry before posting. Automatic posting is enabled in week five for invoice types where the system has proven itself, expanding from there.

What you ownThe company owns the system, integrated into its own accounting environment, with documentation its IT staff use to adjust matching rules themselves.

Knowledge System · Illustrative

Customer Support at an E-Commerce Company

A small support team handles hundreds of tickets a week, and the same questions - order status, returns, product specifications, delivery times - make up the great majority. Average resolution time is 48 hours, customers escalate while waiting, and the team burns out on repetition while difficult cases sit in the same queue as everything else.

What is at stakeSlow support in online retail translates directly into refund requests, chargebacks, and lost repeat customers. The company faces a choice between growing the support team or changing how routine questions are handled.

What we build
  • A support system grounded in the company's actual material: product documents, returns and warranty policies, and live order data - so it answers from facts, not guesses.
  • Automatic resolution of routine enquiries: where is my order, how do I return this, does this product fit my use - answered immediately, around the clock, with the relevant policy or order detail cited.
  • A clean handover for everything else: when a case needs a person, the agent receives the full history and the customer's details already assembled, instead of starting from scratch.

How it rolls outWeeks one and two ingest the product library and connect order data. Weeks three and four, the system drafts replies that agents review and send - building an accuracy record before any customer receives an unreviewed answer. From week five it handles routine categories autonomously, with the team monitoring a daily quality sample.

What you ownThe company owns the system and updates the product knowledge itself as the catalogue changes. Support tooling and help-desk integration run inside its own accounts.

AI Product · Illustrative

Property Intelligence Platform

The founder has proven demand for property investment analysis - clients pay for reports assembled manually over days. What they do not have is a technical co-founder, a development team, or eighteen months to learn to build software. Investors want to see a product, not a spreadsheet service.

What is at stakeManual delivery caps the business at a handful of clients. Every month without a product is a month competitors can close the gap, and the founder's window with early customers and investors is finite.

What we build
  • A complete commercial software product: customer sign-up and accounts, subscription billing, and the analysis engine that turns property data into the investment reports clients already pay for.
  • The analysis capability at its core - the system evaluates properties against market data and produces the structured assessment previously built by hand over days, in minutes.
  • An operations view for the founder: customers, subscriptions, usage, and revenue in one place, manageable by one person without technical staff.

How it rolls outWeeks one and two lock the product scope to what customers have already proven they will pay for - nothing speculative. The build runs with the founder reviewing a working version every week and adjusting based on customer feedback. Final weeks cover billing, onboarding, and launch to the existing client list.

What you ownThe founder owns everything: the product, the code, the customer relationships, and the revenue. No equity taken, no licence fees, no lock-in to Verel Systems.

Find out whether this fits your company

A 30-minute call. We look at how inbound work moves through your team today. If it does not fit, we say so.

Book a call