Engagements run in four stages. Every engagement starts with a 30-minute call with a founder. No SDR layer. No proposal-team relay.
We work embedded with your team. Joint architecture decisions, code review, deploy-day support. The engagement runs through your team's standards, not parallel to them.
Engagements come in three shapes. Shape is decided after the intro call, when we understand the scope — not before, because pre-scope shape commitments lead to the wrong shape for the work.
For engagements where the scope is well-defined and the shipping criteria are clear. 4–6 month focused builds — an indexer, a verification primitive, a stablecoin module, a chain-tooling buildout. Named engineers, named deliverables, named timeline. Fixed price.
For engagements where the protocol's roadmap moves faster than fixed-scope deliverables can capture. Outstaff engineering under MSDA-style terms — Zpoken engineers integrate with your team and ship at the protocol's roadmap velocity, not at contract-amendment cadence.
For engagements that span multi-year chain partnerships, ongoing protocol maintenance, and joint architectural ownership. Less common; takes the most senior architectural commitment from both sides.
Before a contract is signed, discovery typically runs 1–2 weeks at no charge: we read the project documentation, hold working calls with your team, take a first pass over code and architecture, run demos where they help, and negotiate scope and terms. Billing starts when the contract is signed, against the deliverables and terms written into it; you can end the engagement per those terms.
One boundary. When research is the deliverable — a feasibility study, a verification-primitive design, an architecture investigation — that is a billed engagement like any other. Unbilled discovery is for scoping a clear need, not a channel for free research.
The contracting entity is Zpoken OÜ, registered in Estonia. Standard NDA, full IP transfer on engagement deliverables, auditable code-handling — we work in the client's repos by default. Where an engagement routes work through our self-hosted models, that is inference-only and agreed in the code-handling terms — nothing is retained outside the client's environment. Detailed terms on request.
30 senior engineers across L1, ZK, DeFi, and DeAI. Distributed across six European countries — UK, Portugal, Poland, Estonia, Ukraine, Italy. Every engineer works from one of these six countries. If you need to know where a specific engagement's bench sits — for compliance, timezone, or preference — ask on the intro call and we'll tell you, per engineer. No engineer photos on the site (privacy constraint from the team).
AI-native, by construction. AI agents are built into the engineering workflow, not bolted on. Engagements run against an internal arsenal of them — code generation, verification harnesses, chain-specific tooling — built and maintained by the same engineers who ship the work. The model layer is deliberately mixed: Claude — as an Anthropic partner — for frontier models; open-source models self-hosted on GPUs we own, used when an engagement's code-handling terms rule out third-party APIs — inference-only, nothing retained; our own orchestration across both. Agents draft and check. Engineers answer for what ships. What a client sees: the tooling is already built on day one, and live protocols stay maintained without the bench growing around them. The engagement clock doesn't pay for tooling we'd build anyway.
If you're considering an engagement, the first step is a 30-minute call with one of the founders. We discuss commercials, scope, and shape after we understand the problem.