AI workflow assistants
Assistants connected to approved tools and business systems, with defined actions, access limits and handover to your team.
AI engineering · 1it.pro
Your AI prototype.
A clear path forward.
We review, connect and develop existing AI prototypes. Starting from scratch? We also build assistants and document search around one business workflow, with an agreed scope and human control.
Start where you are
One team. Two starting points.
We already have an AI prototype
Identify what works, what needs testing and how to connect it to your systems before agreeing a deployment scope.
Develop an existing prototypeWe want AI in our business
Choose one repetitive task in enquiries, documents or internal operations. Start with a bounded pilot and clear review criteria.
Explore business workflowsFor teams with a prototype
Keep what works. Resolve what blocks deployment.
We start with your existing code, architecture and constraints. The first deliverable is a technical assessment and a scoped plan, rather than a promise to rebuild everything.
Assess the prototype
Review behaviour, dependencies, failure cases and data access. Separate demonstrated capabilities from assumptions.
Connect the workflow
Integrate your CRM, APIs and, where appropriate, MCP tools. Define permissions, confirmation steps and failure handling.
Agree readiness criteria
Define representative tests, logging and deployment checks. Decide together what is ready and what remains outside the release.
For service businesses
Start with a task your team knows well.
Bring a process, not a list of tools. We will scope a pilot around the inputs, expected output and the decisions your team keeps.
Enquiries and reply drafts
Classify incoming requests and prepare draft replies for staff review. Sending messages and changing records follow agreed rules.
Answers from your documents
Search an agreed collection and show the sources used. Missing or conflicting information should be visible to the person reviewing the answer.
Structured data for your team
Extract proposed fields and summaries from documents, with human verification before they enter the agreed business workflow.
What we build
Three services. A scope built around your task.
Document search with sources
Retrieval-based search (RAG) over selected documents. We agree source access, update rules and how answers will be evaluated.
AI prototype audit and development
Technical review, targeted improvements, integrations and testing, followed by an agreed deployment scope.
How we choose the tools
We favour code-based backends and managed integrations. APIs, MCP and n8n are options where the workflow calls for them. Cloud, self-hosted or hybrid deployment depends on data, operations and budget.
How we work
A defined next step at every stage.
Scope, timing and cost follow assessment. Ongoing support is agreed separately.
Review
Understand the process or prototype, available inputs and current limitations.
Define
Agree deliverables, risks, responsibilities and measurable acceptance criteria.
Pilot
Test a bounded workflow with representative, approved examples.
Integrate and test
Connect the agreed systems and check failures, permissions and human review.
Deploy within scope
Release the agreed functionality, document limitations and arrange handover or support.
Control and data
Make the boundaries part of the system.
Access and actions
Agree which sources and tools are available and which actions require human confirmation.
Evaluation and logs
Use agreed test cases and logging to investigate behaviour. AI answers can still be wrong and need appropriate review.
Data and hosting
Agree where data is stored and sent. Self-hosting alone does not prevent requests to external model providers.
Operations and ownership
Define handover, access roles and any backup or support arrangements within the contracted scope.
See the interfaces
View the product portfolioExplore our product demonstrations.
Our portfolio includes interface examples from our own products. Demonstration screens illustrate workflows; they do not establish a client deployment or guarantee the capabilities of your project.
Working together
Agree the work before the build.
Describe your task or current prototype. We clarify the next step, then agree deliverables, timing, fees and any assessment work. Custom AI development is quoted separately from our website packages.
Your next step
Tell us what you want AI to do.
Describe the process or prototype and what is blocking progress. Please do not include passwords, confidential documents or sensitive customer data.
Questions and answers
Before we start
Yes. We begin by reviewing the code, dependencies, access and intended workflow. The assessment determines what can be retained and which improvements make sense; deployment readiness is not assumed.