tkxel’s FDE engagement models are a staged path from a fixed-price assessment to production and beyond. You start with a two to four week discovery, prove value in a pilot, ship with an FDE pod, then choose managed AI operations, training or embedded engineers. Each stage stands alone, so you commit one step at a time.
How tkxel’s engagement models fit together
tkxel’s engagement models follow one path: define the outcome, prove it, ship it, then run or hand it over. Most clients enter at the assessment, but you can start wherever your project actually is.
The stages are deliberately separate contracts. That gives you a clear decision point after each one, with working evidence in hand, rather than signing up to a long programme on the strength of a proposal.
The pricing basis changes as uncertainty falls. Early stages are fixed price because the scope is small and well bounded. Build stages are priced per phase or per pod month around a defined outcome, and operations move to a recurring fee tied to a service level.
- Assessment and discovery: 2–4 weeks, fixed price.
- Pilot or proof of concept: one workflow, real data, agreed success metric.
- FDE pod project or programme: one or more outcomes shipped to production.
- Managed AI operations: ongoing monitoring, tuning and support.
- Training package: skills for the team that will own the system.
- Embedded FDE engineers: individuals added to your team.
Engagement models compared
The table summarises each model so you can see what you are buying at each stage. Durations are typical ranges; we confirm them during scoping, once we understand your data, systems and approval steps.
| Model | Duration | Pricing basis | Best for | What you get |
|---|---|---|---|---|
| Assessment / discovery | 2–4 weeks | Fixed price | Defining one outcome and testing feasibility | Prioritised use cases, data and access review, delivery plan and estimate |
| Pilot / POC | Typically 4–8 weeks | Fixed price with agreed success metric | Proving value on one workflow before scaling | Working pilot on real data, results against the metric, go/no-go recommendation |
| FDE pod project or programme | Typically 2–6 months per outcome | Fixed fee per phase or monthly pod fee | Shipping a defined outcome to production | A 3–5 engineer pod, production system, documentation and handover |
| Managed AI operations | Ongoing, monthly | Recurring monthly fee scoped to service level | Keeping AI systems accurate, secure and running | Monitoring, model and prompt tuning, incident response, regular reporting |
| Training package | Days to weeks | Fixed price per package | Teams taking ownership of AI systems | Role-based sessions, hands-on labs, playbooks |
| Embedded FDE engineers | Flexible, monthly | Per engineer per month | Adding FDE skills to a team you already lead | Individual engineers working under your direction |
Assessment and discovery
The assessment is a two to four week, fixed-price engagement that turns a broad ambition into one outcome a pod can deliver. We review your use cases, data, systems, security constraints and stakeholders, then recommend what to build first and what it will take.
You leave with a delivery plan and an estimate you can take to budget holders, whether or not you continue with us. If the assessment shows the outcome is not ready, because the data is missing or the process is not stable, we say so. That is a useful result too, and much cheaper to learn now than halfway through a build.
- Workshops with business owners and the people who do the work today.
- A review of data sources, quality, access and security constraints.
- A ranked shortlist of use cases with effort and value estimates.
- A recommended first outcome, team shape, timeline and fixed estimate.
Pilot or proof of concept
A pilot proves that the chosen outcome works on your real data before you commit to a full build. We agree a success metric in writing up front, such as processing time, accuracy or adoption, and build against one workflow.
Pilots are worth running when the risk is technical or the business case needs evidence. They are less useful when the outcome is already well understood, and in that case we will usually suggest going straight to a pod so you do not pay to prove something twice.
A pilot ends with a clear go, adjust or stop recommendation. Because it runs on real data with real users, the code is built to be extended into production rather than thrown away, which shortens the pod phase if you continue.
FDE pod project or programme
An FDE pod project puts three to five tkxel engineers inside your environment to ship one outcome to production and hand it over. The pod brings its own technical lead, works in your tools and repositories, and reports against the production goal every week.
When you have several related outcomes, a programme runs pods in sequence or in parallel under shared governance. For work built around AI agents, our agentic FDE pod adds specialists in agent design, evaluation and guardrails.
Pods are priced as a fixed fee per phase or a monthly pod fee for a set team, so you know what each month costs. The scope flexes within the outcome as the pod learns, but the outcome itself does not change without your agreement.
- One named outcome per pod, agreed before work starts.
- Weekly demos of working software.
- Handover with code, runbooks, documentation and paired sessions.
Managed AI operations
Managed AI operations keep a production AI system accurate, secure and available after launch, for a recurring monthly fee. AI systems drift as data, models and user behaviour change, so someone needs to watch quality, retune prompts and models, and respond to incidents.
This model suits companies that want the result without building an in-house AI operations team. The trade-off is ongoing dependency on a partner, so we document everything as we go and can transition operations to your team whenever you are ready.
The monthly fee is scoped to a service level: what we monitor, how quickly we respond, how often we report and how much improvement work is included. That keeps the cost predictable and makes it easy to compare against hiring your own team.
Training package and embedded FDE engineers
The training package prepares the people who will own and use the system, through role-based sessions, hands-on labs and playbooks. It pairs well with a pod handover, and it also works on its own for teams starting their own AI work. Sessions are built around your own systems and data where possible, so engineers practise on the code they will maintain and business users practise on the workflows they will run.
Embedded FDE engineers are individual tkxel engineers who join a team you already lead. You direct the work and keep delivery accountability, so this is closer to staff augmentation than to a pod. It is the right option when you have strong leadership in place and need specific FDE skills, such as data integration or applied AI, rather than a self-directed team.
What every tkxel engagement includes
Every tkxel engagement, whatever its size, includes the same working basics. They exist so you can see progress, keep control of decisions and own what we build.
Some of these add a little overhead to short engagements. We keep them anyway, because skipping documentation or security reviews to save a week tends to cost far more when someone has to run the system later.
- A written outcome or deliverable agreed before work starts.
- A named lead on our side and a named owner on yours.
- Regular demos and status against the agreed goal.
- Work inside your security perimeter, under our ISO 27001 controls.
- Full ownership of code and documentation by you.
- A clear exit point at the end of each stage.
How to choose your starting point
Choose your starting point by asking how clear the outcome is and who will lead the work. If you are unsure what to build, start with the assessment. If you know the outcome but need evidence, run a pilot. If the outcome is clear and you want it live, go straight to a pod.
If you already lead a capable team and only need specific skills, embedded engineers are the lighter option. You keep direction and accountability, and you can scale the number of engineers up or down month by month as your roadmap changes.
Think too about what happens after launch. If you have no plans to staff AI operations internally, factor managed operations into the budget from the start, and if you do, add a training package so your team is ready before the pod leaves.
Every engagement runs under tkxel’s ISO 27001, ISO 20000 and ISO 9001 certified management systems and CMMI DEV Level 3 practices, with teams across the USA, Saudi Arabia, Portugal and Pakistan. If you are still comparing delivery approaches, our guide to FDE vs staff augmentation vs consulting lays out the trade-offs.