Home / Forward deployed engineering
A forward deployed engineer works inside your team, on your data and your systems, and ships the working solution instead of a recommendation. It is one of the four ways we work with South African businesses, and the one to choose when your own team should own the build.
Book a call if you want one, or just send the form. We reply within one business day. POPIA-compliant.
A forward deployed engineer (FDE) is a software engineer who works inside the client's business rather than from a vendor's office. They sit with the people who run the process, build on the client's real data and systems, and stay until the solution is live and the team can run it.
Forward deployed consulting applies the same idea to advice. The consultant works the problem with your team and leaves behind a running system, not a slide deck.
The role was popularised by Palantir and is now a job title at OpenAI, Anthropic, AWS, Databricks and Salesforce. In August 2026 AWS confirmed at its Johannesburg Summit that its forward deployed engineering programme is coming to South Africa. Salesforce, PwC and Palantir delivery partners are already hiring FDEs in Johannesburg and Cape Town. Those programmes are built for enterprise accounts. We bring the same way of working to businesses with 10 to 1,000 people, priced in rands.
New to the term? Read the plain-language guide: What is a forward deployed engineer, and does your business need one?
Forward deployed consulting is advisory work done inside the client's business, by people who then build what they recommend. The consultant maps the process with the team that runs it, decides what to configure and what to build, and stays through delivery. The deliverable is a working system, not a report.
It exists because the usual split fails: a strategy firm recommends, a development shop builds from the recommendation, and nobody owns the gap between them. Our directors spent years on each side of that gap at Deloitte, Accenture and SAP programmes. Forward deployed consulting closes it by making the same people responsible for the advice and the build, on a fixed scope, in rands.
The same four-step cycle we run on every project, with one difference: the engineer doing the work is in your business, not reporting on it from somewhere else.
We map the process with the people who run it, as it works today. We put a rand figure on the friction and agree in writing what done looks like. No tools are chosen yet.
Often the right answer is a platform you already pay for that nobody set up. Sometimes it is custom software. Usually it is both. We tell you which, fix the scope and the price, and only then start.
We build, configure, integrate and test against your live systems and your real edge cases. Every fortnight you see working software, not a status report. If we have misunderstood something you find out in week four, not week twelve.
Your team is trained, the documentation is written for whoever maintains it next, and every account and credential is in your name. We support it for 30 days. After that you can carry on alone, keep us on, or call when something changes.
Most businesses have tried at least one of the other three. This is where the money usually goes wrong, and why.
| Management consultant | Outsourced dev shop | Contractor via agency | Forward deployed engineer | |
|---|---|---|---|---|
| Where they work | Your boardroom, then their office | Their office, from your spec | Your desk, your instructions | Inside your team, on your systems |
| What you get at the end | A recommendation to implement | Software that matches the spec, whether or not the spec was right | Hours worked | A live system your team runs |
| Who decides what to build | They advise, you decide, someone else builds | You, up front, in a document | You, day to day | We do, with you, after mapping the real process |
| How scope is managed | Phases and change requests | Change requests and invoices | Open-ended | Fixed scope and price before the build; changes are a conversation first |
| Who owns the result | You own the report | Depends on the contract | You, if it was documented | You: code, accounts and credentials in your name |
| Where it fails | Nothing gets implemented | The spec was wrong and nobody noticed until delivery | Knowledge leaves when they do | If the process is not worth automating. We say so in week one. |
South African businesses with 10 to 1,000 people that have outgrown spreadsheets and WhatsApp but cannot justify an enterprise programme. Our directors ran transformation programmes at Deloitte, Accenture and NYSE-listed SaaS firms; you get the same scoping discipline without the programme office, priced in rands.
Accounting, CRM, inventory and email all exist. Information still travels between them on a person. We connect what you already pay for before we build anything new.
You have people who could run a better system. They do not have the weeks to build it. We deploy alongside them, using your tools and your stand-ups, and leave the system in their hands.
Document processing, customer queries, reporting, internal knowledge. We find the one use case that pays, build it against real work, and give you a one-page policy so it is safe to use.
We recommend the cheapest one that solves the problem. All three are quoted in rands before any building starts.
One defined process, one agreed price, one delivery date. The default for a first engagement, because it caps your risk while you find out whether you want to work with us.
A set number of days a month for businesses digitising several processes in sequence, or who want the systems we built kept current as the business changes.
An engineer inside your team for a defined period, on your tools and your stand-ups. For businesses with in-house capacity that needs implementation experience alongside it.
We deploy with
A forward deployed engineer (FDE) is a software engineer who works inside the client's business rather than from a vendor's office. They sit with the people who run the process, build on the client's real data and systems, and stay until the solution is live and the team can run it. The role was popularised by Palantir and is now used by OpenAI, Anthropic, AWS, Databricks and Salesforce.
Advisory work done inside the client's business by people who then build what they recommend. The consultant maps the process with the team that runs it, decides what to configure and what to build, and stays through delivery. The deliverable is a working system, not a report. It is how we run every engagement, whether the label is engineering or consulting.
A consultant studies the problem and hands you a recommendation to implement. A forward deployed engineer implements it: they write the code, configure the tools, connect the systems and run it against your real data, then hand over a working system with your team trained to own it. Forward deployed consulting is the same idea applied to advice: the advice comes with the build.
An outsourced team works from a specification you write and delivers at the end. A contractor fills a seat and does what they are told. A forward deployed engineer owns the outcome: they find the process worth fixing, decide what to build against what to configure, ship it in fortnightly demos, and leave it running in accounts you control.
There is no list price because it depends on the process. What we commit to: fixed-scope work is quoted in rands and agreed before any building starts; retainer and in-team deployments are a set number of days a month at a rand day rate. Vendor forward deployed programmes are usually attached to a platform commitment; ours is not, and first engagements are deliberately scoped small.
Both. We are based in Gauteng and work in South African hours. On-site days are normal for discovery and go-live in Gauteng; the rest of the country is served remotely with fortnightly working demos. Everything we build is designed to survive interrupted connectivity, because loadshedding is a constraint, not an excuse.
You do. Code sits in a repository you own, deployed to infrastructure in your name, with documentation written for whoever maintains it next. There is no licence to keep paying us and no lock-in by design.
Four questions are enough for us to say whether it is worth automating, how we would approach it, roughly what it costs in rands, and whether we are the right people to do it.
Book a call if you want one, or just send the form. We reply within one business day. POPIA-compliant.