Software engineering & AI strategy
We help organisations decide where AI belongs, build the software around it and measure how it performs. From the first strategic decision to a system in production.
Discuss your projectGive an agent the context and tools to do useful work, with clear boundaries and a person in control of consequential actions.
Sensitive actions pass through human approval. Tool access stays within defined permissions.
Use hybrid search, knowledge graphs and iterative retrieval as the question requires. Keep access rules and source evidence attached to the answer.
GraphRAG helps connect entities and relationships. Choose the retrieval approach against real questions, and make gaps in the evidence visible.
Compare model and prompt changes on representative tasks. Combine deterministic checks, LLM-as-judge and human review against an agreed rubric.
Validate judge scores against human judgement. Feed production failures back into the evaluation set to check future changes.
Check whether an eligible previous response can answer a similar request. Reuse only when the context, permissions and freshness allow it.
No eligible match? Call the model. Access boundaries and fresh information take priority over a cache hit.
Previous professional experience
The engineering background behind our practice.
Across the AI lifecycle
Useful AI needs more than a working model. The surrounding data, tools, evaluation and operations determine how well it serves the people using it.
Explore all capabilitiesConnect models to the tools and workflows your business relies on. Define what an agent can do, when it needs a person, and how you know it is working.
Agent orchestration, tool integration and human oversight
Explore agentsConnect people to the evidence they need. Combine hybrid search, knowledge graphs and agentic retrieval around the question, the available sources and access permissions.
GraphRAG, hybrid search, agentic retrieval and context engineering
Explore knowledgeDefine what a good result looks like and build the evidence to assess it. Combine LLM-as-judge, human review and automated checks to compare changes and catch regressions.
Evaluation datasets, scoring rubrics and controlled experiments
Explore evaluationSee what happens across model calls, retrieval and tools. Bring quality metrics, latency, cost and reliability into the same conversation, supported by a platform your team can control.
Observability, semantic routing, caching and MLOps
Explore platform operationsSoftware engineering, every day
We design and build web applications, APIs and the services behind them. New products and existing systems receive the same attention to usability, code quality and long-term ownership.
Understand the workflow before defining the interface. Test prototypes with users, build accessible design systems and work through loading, empty and error states. Design and engineering shape the product together.
Choose clear boundaries, sound data models and explicit API contracts. Keep changes small, review code and record important tradeoffs. Refactor deliberately and modernise existing systems in manageable steps.
Define behaviour with test-driven development (TDD), automate checks in CI/CD and prepare releases with rollback in mind. Monitor reliability and performance, maintain dependencies and turn production feedback into the next improvement.
Bring AI into your existing cloud, data and business tools. We design the connections, deployment and operations around your requirements, from managed model services to private infrastructure.
Applied to the work that matters
Give people useful answers from the information your organisation already holds.
Explore the briefConnect reasoning, business tools and human decisions in a controlled workflow.
Explore the briefExtract, classify and validate information from documents, images and complex files.
Explore the briefWorking together
Assess opportunities, data readiness, delivery constraints and build-versus-buy options. Agree the business baseline, ownership and evidence needed to proceed.
What you take forwardPrioritised opportunities, cost assumptions, success measures and a delivery roadmap.
Define acceptance criteria and evaluation cases before implementation, then build in tested increments. Prepare the integrations and operations, with practical training and handover so the team can adopt the workflow.
What you take forwardWorking software, evaluation evidence, operating guidance and team enablement.
Review architecture, quality, security, cost and latency alongside adoption and business outcomes. Agree priorities and a review cadence to turn recurring failures into measurable improvements.
What you take forwardA technical assessment, agreed priorities and a measurable baseline.
Thenesor, part of Alcorac
Technology decisions shape the product, the operating costs and the work of the people who use it. We consider those consequences together.
Thenesor brings software delivery and AI strategy into one practice, owned by Alcorac Ventures, S.L. We work with your team to turn business priorities into systems you can understand, operate and improve.
About the consultancy
