WHAT WE BUILT
Real projects we've built. Each one started with a problem a business couldn't solve with their existing tools.
Gym WhatsApp Bot
A gym owner was growing fast but managing everything by hand. Member onboarding was spread across Google Sheets, a booking system, accounting software, and WhatsApp groups. Daily tasks like checking leads, new sign-ups, electricity meters, and water meters were tracked in someone's head. Trainers were independent contractors with no proper onboarding process. The small things that make the experience personal, like remembering birthdays and injuries, were getting lost.
We built a WhatsApp bot because that's where the members already are. They book classes, buy supplements, and check schedules themselves. Trainers check in with GPS so the owner knows who's actually there. AI reads photos of meters and equipment gauges so nobody has to type numbers into a spreadsheet. The owner gets one place to see what's happening instead of checking six different apps.
KEY OUTCOMES
- Members book their own classes, no phone calls needed
- GPS-verified trainer attendance, no more buddy check-ins
- One dashboard instead of six different apps and spreadsheets
Dala
A consulting firm had staff using ChatGPT on their own with no structure. Everyone was prompting differently, getting wildly different quality. There was no record of what anyone asked or what the AI said back. And sensitive client data was going straight to third-party servers with no control over how it was stored or used.
We built a desktop app that gives each team member AI chat with standardised prompts and skills for their role. The AI searches company documents directly, runs tasks, and remembers previous conversations. Everything runs on the user's own machine. No data leaves. No third-party training on your information.
KEY OUTCOMES
- Works with Claude, GPT, and Gemini, all in one place
- AI searches your company docs and runs tasks without copy-pasting
- Runs locally, your data never leaves your machine
Client Portal
An agency was managing clients across email chains, WhatsApp messages, spreadsheets, and a shared drive. Clients kept asking "where's my project at?" and the team spent hours each week writing status updates. Contracts were sent by email and chased for weeks before anyone signed. Invoices were created by hand from time logged in a different system.
We built one platform where the team manages everything: clients, projects, tasks, documents, time tracking, and invoicing. Clients log in and see their own progress, download documents, and sign contracts right there. No more "just checking in" emails. Time tracking feeds straight into invoices so nothing gets missed.
KEY OUTCOMES
- Single platform replaces email chains, spreadsheets, and shared drives
- Clients check their own progress, download documents, and pay invoices
- Contracts and proposals get signed inside the platform, no chasing
Deployment Pipeline
A small software team was deploying by hand. Someone would log into the server, pull the latest code, restart services, and hope nothing broke. Every deploy took 20 to 30 minutes and was stressful. If something went wrong, rolling back meant doing the whole thing again. Different projects lived on different servers with no consistency.
We built a system where the team clicks one button and the code goes live. Each project gets its own web address automatically. Health checks confirm everything is working. If something breaks, they roll back in seconds. The whole deploy takes under two minutes now, and nobody has to touch a server.
KEY OUTCOMES
- One click to deploy, no logging into servers
- Each project gets its own web address automatically
- Version tracking and health checks, roll back if anything breaks
Site Attendance
A construction company was paying workers weekly based on daily attendance, but every week the same arguments came up. Workers said they were on site. Foremen couldn't prove otherwise. Paper sign-in sheets went missing or got filled in after the fact. It was costing money and causing friction between the crews and the office.
We built a web app that works like a native app on the foreman's phone. They open it in the browser, pick a worker, and snap a photo. The photo gets compared against the worker's face on file, so there's no faking it. If the site has no signal, it stores everything on the phone and syncs when they're back in range. At the end of the week, there's a clear record of who was where and when.
KEY OUTCOMES
- Every check-in verified with facial recognition, no more disputes
- Works offline on remote sites, syncs when back in range
- Clear daily records for payroll, no paper sign-in sheets
Restaurant Operations
A restaurant owner running multiple fast-food brands across multiple locations had no central view of the business. Managers did nightly cashups and emailed slips to the owner. Weekly review meetings happened with no reliable data. A burger supposed to cost R10 was actually costing R12 because of wastage, and nobody caught it. Five different suppliers were charging different prices for the same ingredients with no easy way to compare.
We built a web app that pulls data straight from the existing POS system. Managers upload their reports, AI extracts the numbers from CSVs and PDFs, and everything lands in one place. The owner sees every store and brand. Managers only see their own. Variance reports flag anything over 3% so the owner knows exactly where money is leaking before it adds up.
KEY OUTCOMES
- All stores and brands in one dashboard, no more emailed slips
- Variance alerts catch over-costing before it adds up
- Managers see their stores, owner sees everything, no data leaking across brands
AI Tutor & Knowledge Agent
A company had policy documents that nobody read. New hires got a PDF dump and were expected to figure it out. When someone had a question about a specific policy, they'd ask a colleague who might be wrong, or dig through hundreds of pages. L&D had no way to know who actually understood the material and who just clicked through. Knowledge gaps only showed up when something went wrong.
We built a system that turns policy documents into structured courses automatically. An AI tutor walks each learner through scenario-based questions. It doesn't give answers. It guides them there, adapting based on what they get right and wrong. After completion, the AI analyses every conversation to find knowledge gaps and generates follow-up material. A separate knowledge agent sits on the company site so staff can ask policy questions any time. When something can't be answered, the question gets flagged to the policy owner with the full conversation for context.
KEY OUTCOMES
- Policy documents turn into structured courses automatically
- AI finds knowledge gaps and generates follow-up material
- Staff get instant answers to policy questions, unanswered ones flagged to owners
- Works alongside existing LMS, full audit trail for L&D reporting
Proposal Builder
A consulting firm was writing proposals from scratch every time. Someone would dig through old emails for pricing, copy sections from past documents, and manually build wireframes in PowerPoint. A single proposal took days. Half the time it went out with inconsistent branding or outdated rates.
We built a pipeline inside the client portal that takes a lead from first conversation to finished proposal. The team captures requirements, generates wireframes, selects service packages, and produces a branded proposal document. Everything stays in one place. A proposal that took days now takes hours.
KEY OUTCOMES
- Lead to wireframes to proposal in one flow
- Consistent branding and pricing on every proposal
- Proposals go out in hours, not days
Contract Analysis Agent
A legal team was reviewing contracts line by line. Every new agreement meant hours of reading, cross-referencing against company policy, and flagging risky clauses. Junior staff missed things. Senior lawyers spent their time on routine checks instead of negotiation.
We built an AI agent that reads contracts and flags problematic clauses automatically. The client or their lawyer defines what "problematic" means. Liability caps too low, unfavourable payment terms, missing termination rights. The agent highlights every issue, explains why it flagged it, and references the rule it broke. Lawyers review flagged sections instead of reading everything.
KEY OUTCOMES
- Client defines the rules. The agent enforces them consistently
- Every flag links back to the specific rule it violates
- Lawyers focus on negotiation, not routine clause-checking
HAVE A SIMILAR CHALLENGE?
Let's talk about how we can build something like this for your business.
Not ready to talk yet? Take our free 5-minute assessment and get a report showing what your business needs to do next.