Roles and coordination
The lead company, product and technical owners, contacts and the decision-making process.
How independent companies collaborate and deliver shared projects.
Participants agree the terms of each initiative. Each company determines how to organise its own business. Roles, success criteria and quality requirements are agreed before work begins.
The lead company, product and technical owners, contacts and the decision-making process.
Scope, payments, client relationships, change management and the process for resolving disagreements.
Ownership of outputs, confidentiality, licences, data access and rules for component reuse.
Project teams are engaged based on domain expertise, not hierarchy or resource availability. This eliminates the handoff effect between contractors.
Each project has an agreed lead member company and client contact. The contractual model, SLA and reporting are defined around the participating teams and their responsibilities.
UnityBase and other ecosystem platforms have their own architectures. Compatibility is established through agreed APIs and data contracts, identity management and integration checks.
Threat modelling, least privilege, data protection and recovery requirements inform architecture. Regulatory obligations depend on the data, sector, jurisdiction and each project participant’s role.
Define the task, baseline, expected value and success criteria. Compare AI with rules-based automation.
Identify sources, data quality, permissions, integrations, deployment environment and component owners.
Test quality on representative cases, errors, security, latency and the full cost of a correctly completed operation. Use the result to decide whether to scale.
Test load and recovery, configure monitoring, train users and agree support.
Monitor quality, costs and incidents. Changes to models, prompts, data sources or tools require evaluation and a controlled release.
Standards, regulatory requirements and engineering practices serve different purposes. Applicability is documented per project; a company’s certification is established by a valid certificate and its scope.
ISO/IEC 27001:2022 covers information security management, risk assessment and controls for access, incidents and suppliers. Implementing controls and obtaining independent certification are separate steps.
NIS2 applies through EU Member States’ legislation to defined entities and sectors. Financial institutions additionally require assessment of DORA and ICT supplier obligations.
GDPR governs personal data. The Data Act separately addresses access to connected-product data and switching cloud providers. Access rights, portability and contractual terms inform architecture.
Access is evaluated using identity, device posture and request context. Least privilege, segmentation, MFA and workload identity controls complement network protection.
A secure lifecycle informed by NIST SSDF and OWASP ASVS 5.0 verification: code, dependency and secret analysis, SBOMs, artifact provenance and vulnerability remediation.
ISO 9001:2026 is the current quality management edition. Transition from the previous edition is planned with the certification body; each company’s status is checked separately.
AI risk management, accountability, data quality and human oversight. AI Act applicability depends on intended use and organisational role; ISO/IEC 42001 certification does not replace regulatory obligations.
For products within CRA scope: vulnerability handling, updates and assurance evidence. Reporting obligations apply from September 2026 and the main requirements from December 2027.
Security authorisation and integrated information protection systems (KSZI) under current and transitional procedures. Protection, cryptography and assurance requirements follow the information category and system.
Approved tools and data rules, small changes, tests and accountable review. Measure impact through delivery time, defects and release stability.
Evaluation informed by NIST AI RMF and OWASP guidance for LLMs and agents. Tools have limited permissions, spending limits, action logs and defined human approval points.
Scrum for iterative product work and Kanban for continuous flow. Cadence, work-in-progress limits, demonstrations and retrospectives are selected for the team’s context.
Automated builds, tests and security checks; small releases, controlled rollout and rollback. AI-assisted changes undergo the same checks and accountable review.
Model business domains, record decisions in ADRs and verify integration contracts. Select a modular monolith or microservices based on domain boundaries, scaling needs and operational readiness.
Terraform, Ansible and other IaC tools provide versioned environments. GitOps, policies and secrets have explicit controls; recovery is verified through exercises.
OpenTelemetry for metrics, logs and traces; SLIs/SLOs, error budgets, incident reviews and practical RPO/RTO verification.
Versioned models, prompts and evaluation sets; permission-aware RAG, prompt-injection checks, restricted tools, budgets and agent autonomy. Failures have a defined safe fallback.
Team self-service through verified templates, supported versions and transparent costs. Platform value is assessed through developer experience and service outcomes.
Support is provided by the designated member company. Service targets and responsibilities are contractual and specific to the solution.
The SLA defines availability measurement, support hours and the responsible company’s response commitments.
SLOs, RTO and RPO are agreed according to system criticality, architecture and recovery test results.
AI functions have agreed quality thresholds, autonomy limits, budgets and a process for handing work to a person.
Tell us about your company, product and expertise. Let’s explore how we can strengthen each other and identify a first shared task.