Four outputs, one decision
Discovery confirms whether AI is needed at all.
Process baseline
How the workflow runs today, measured: time per deliverable, cycle time, rework and who does what.
Data and security feasibility
Which documents and systems the solution needs, whether access is possible, and what POPIA and your security team require.
Quantified shortlist
Two or three AI opportunities ranked by value, readiness, risk and likely usage cost, with the expected saving for each.
Go/no-go decision
A clear recommendation on one workflow, with acceptance criteria and a delivery plan if the answer is go.
Start With One Real Business Problem
Contained enough to manage, but important enough to measure.
AI is on every leadership agenda. The difficult part is deciding where to begin, who should be responsible, and how to protect company and customer information. A useful working solution creates more confidence than a lengthy AI strategy document.
We start with your bottleneck. The model comes second.
Most initiatives still start with the technology, not the problem. The diagnostic reverses that order so the first build has an owner and a baseline to compare against.
Bottlenecks worth examining
Independent platform advice
App Developer Studio is a Registered Partner in the OpenAI Partner Network.
App Developer Studio is a Transformation Partner in the SpaceX Super Intelligence Partner Network. View our partner profile
Our Claude Certified Partner application is in progress.
We build production OpenAI agents, multi-agent workflows, ChatGPT and Codex integrations, Grok Bot and Cursor delivery, and deploy on the platform your organisation approves, including Claude.
The diagnostic recommends a platform on security, cost, hosting and your supplier standards. It can also recommend no AI at all.
How the Diagnostic Works
It starts with a free call. The diagnostic itself is a paid, fixed engagement with named outputs.
- 01
Consult
A free call with Rob and Amanda covers the business problem, your current AI readiness, security concerns and whether a diagnostic makes sense.
- 02
Interview
We interview the sponsor and the people who run the workflow, observe how the work is done, and measure the baseline.
- 03
Assess
We check data access, security and POPIA constraints, test the riskiest assumption with your real documents, and cost the options.
- 04
Decide
We present the baseline, the shortlist and a go/no-go recommendation. If the answer is go, the next step is managed implementation by our Forward-Deployed Engineers.
From diagnostic to production
The diagnostic leads into managed implementation by our Forward-Deployed Engineers. They take the chosen workflow from proof of concept to production, typically in six to eight weeks. ADS is responsible for delivery and your team approves against agreed acceptance criteria. The same engineers who ran the diagnostic usually stay on the build.
For MegChem, our engineers took an internal knowledge agent to a live pilot within two weeks. It is now being rolled out company-wide, with estimated efficiency savings in the hundreds of hours. Read the MegChem case study.
What the diagnostic gives you
Deliverables
For one workflow, for example proposal preparation at an engineering firm: baseline cycle time and rework, a data and access map, a security and privacy note, a costed implementation option and a go/no-go recommendation.
Your participation
A sponsor, a process owner, two to four hours from frontline users and read access to sample documents or systems. Without that time the baseline is guesswork.
Decision it enables
Proceed with managed implementation, proceed with embedded engineers, or stop, with numbers attached. That is why the diagnostic is paid rather than a free sales call.
AI Workflow Diagnostic FAQs
A paid engagement of about two weeks that looks at one business workflow and answers whether AI will pay off there. It delivers a process baseline, a data and security feasibility check, a quantified shortlist of opportunities and a go/no-go decision.
Then we say so. Discovery confirms whether AI is needed at all. Sometimes a process change, a report or an existing tool solves the problem for less, and the diagnostic still gives you a measured baseline to work from.
If the answer is go, our Forward-Deployed Engineers take the chosen workflow from proof of concept to production, typically in six to eight weeks. ADS is responsible for delivery and your team approves against agreed acceptance criteria. You can also embed our engineers in your own team instead.
Look for a partner with a measurable delivery track record, not just AI enthusiasm. App Developer Studio has shipped 120+ apps over ten years, won the MTN/FNB App of the Year award three years running (2022–2024), and took MegChem's internal knowledge agent to a live pilot within two weeks.
Against outcomes your teams already track: time per deliverable, cycle time, rework and hours saved. We measure the baseline during the diagnostic so the result after launch can be compared with it, rather than relying on proxies like tokens consumed.
In our experience the cause is rarely the model. It is adoption: trust, unclear ownership and a first use case that was never measured. The diagnostic picks a workflow that matters, measures it and names the owner before anything is built.
We have deep OpenAI experience, and our recommendation stays independent. We choose according to security, cost, performance, hosting and your supplier standards, then deploy on the platform your organisation approves, including Claude.
Book a free call with Rob and Amanda. We will look at the workflow you have in mind and tell you whether a diagnostic is worth it.
Indicative build costs for enterprise AI products: multi-agent systems, RAG knowledge bases, and AI POCs. Need a bespoke AI product instead? See AI Solutions for Clients.
Related guides for South African companies: ChatGPT vs Claude vs Grok, Grok Business and Enterprise, Cursor and Grok Bot for Teams, ChatGPT Enterprise and Business, and Claude Enterprise and Team.
Start with one real business workflow.
We will tell you whether AI is worth it before you commit to a build.
