AI & Automation
AI Assistants & Chatbots
AI assistants and chatbots grounded in your own documents, policies and systems — answering customers and staff on web chat, WhatsApp and internal tools, not reciting the open internet.
AI Assistants & Chatbots: the overview
A generic chatbot that answers from the open internet is a liability for a business — it doesn’t know your products, your policies or your customers, and it can confidently invent answers. A custom AI assistant is different: it’s grounded in your own data, so it answers about your business specifically, cites its sources, and stays within the guardrails you set.
We build assistants for the jobs that eat your team’s time — customer support, internal help desks, answering questions across your documents and systems — and, where it’s valuable, agents that take actions: drafting a reply, creating a ticket, looking up an order, updating a record. Answers are grounded in your data using retrieval so they can be traced back to a source.
The result is faster, consistent answers that do not depend on someone being at their desk, and staff freed from repetitive questions. Accuracy is measured on a test set of your real questions before launch, and we build with the right controls — human handover, permissions, logging — so an assistant helps without going off the rails.
What AI Assistants & Chatbots includes
Grounded on Your Data
Retrieval over your documents, policies and product data so answers are about your business.
Support & Internal Assistants
Customer support and internal help-desk assistants that take on repetitive questions.
Hand-off to AI Agents
When a request needs action in your systems, the assistant can pass it to an AI agent that works with approval steps and an audit trail.
Guardrails & Accuracy
Source citations, scope limits and fallback to a person to reduce made-up answers.
Channels
Web chat, WhatsApp, internal tools or embedded in your product.
Security & Logging
Access control, data handling and conversation logging for review, with retention you set.
Who AI Assistants & Chatbots is built for
Support-heavy businesses
Answer routine questions accurately and on-brand, so staff handle the cases that need them.
Teams with lots of documents
Make policies, manuals and knowledge answerable in plain language.
Operations wanting agents
Automate lookups and actions across systems with an AI agent.
Product teams
Embed an assistant that helps users get value from your software.
Is AI assistant and chatbot development the right choice?
A good fit when
- Staff or customers ask the same questions repeatedly, and the answers exist in your documents or systems.
- Your source content is reasonably current and someone owns keeping it that way.
- You can say what the assistant must not answer and when it should hand over to a person.
- You want the assistant inside your own website, WhatsApp channel or internal tools.
Consider another option when
- Your help-desk or CRM product already offers an AI assistant over your help articles; trying it first may be enough.
- Answers need actions in your systems, such as posting or updating records; see AI agent development.
- Your documents are out of date or contradict each other. Tidy the source content first, or the assistant will repeat the contradictions.
Usually in a first release
- One assistant for one audience (customers or staff) on one channel
- Retrieval over an agreed set of documents and pages, with source citations
- Handover to a person, and a list of topics the assistant must decline
- A labelled test set of real questions with expected answers, scored before go-live
- Conversation logging and a review screen for your team
Outside the first release unless agreed
- Actions that change data in your systems (that is agent work, scoped separately)
- Additional channels, languages or audiences beyond those agreed
- Writing or cleaning up the source content itself
- Voice or phone channels unless agreed
Anything outside the approved scope is reviewed and agreed before work begins. See how we work.
Data, controls, responsibilities and ownership
Data migration and integrations
- We document which data goes to which model provider for each workflow, and keep personal or sensitive fields out of prompts where the task does not need them.
- Commercial model APIs offer business terms and data-retention settings that vary by provider and account type. We configure the settings you choose and record them; see security and data protection.
- Where your policy requires it, open-weight models hosted in your own cloud account or on your own servers are an option. They usually mean more hosting effort and can be less accurate on hard cases, so we compare them on your test set before you decide.
- WhatsApp runs on the WhatsApp Business Platform, which needs a Meta business account and verification in your organization’s name.
- The index is refreshed when source documents change; stale content gives stale answers, so each source has an owner.
Roles, approvals and audit
- Human handover
- Low-confidence answers, complaints and out-of-scope topics pass to a person with the conversation so far.
- Access by role
- Internal assistants only retrieve content the signed-in user is allowed to see.
- Evaluation before go-live
- Accuracy is measured on a labelled set of real questions from your team, and re-run after every prompt, content or model change.
- Audit logging
- Questions, retrieved sources, answers and handovers are logged for review, with a retention period you set.
What we need from your team
- The documents, FAQs and policies the assistant should use, and a content owner to keep them current.
- A set of real questions (from emails, tickets or chats) with the answers your team would give, to build the test set.
- Decisions on topics the assistant must decline and when it hands over.
- Accounts in your organization’s name for the model provider and channels such as WhatsApp.
- Reviewers who check answers during the pilot.
Ownership, support and running costs
- Project code, prompts, configuration and evaluation sets transfer to you on full payment, and your data and documents are yours throughout. Third-party foundation models remain the provider’s and are used under the provider’s terms; see our IP and ownership policy.
- Model provider, hosting and storage accounts are set up in your organization’s name where possible, so usage, retention settings and terms sit between you and the provider.
- Running costs by category: model or API usage (grows with volume and input length), vector index and storage, application hosting, evaluation upkeep when your content or the model changes, and monitoring. Amounts depend on volume, model choice and hosting.
- Providers update and retire models. A maintenance and support agreement covers re-running your evaluation set and adjusting prompts when that happens.
Related reading for this decision
- AI agent development
For requests that need actions in your systems, with approval steps and an audit trail.
- AI knowledge base and search
The same grounded index can power search for staff who prefer to look things up.
- LLM app development: RAG, agents and what ships
A plain-language explanation of retrieval, guardrails and evaluation.
- Security and data protection
How data access, hosting and retention are handled and agreed in writing.
How we deliver AI Assistants & Chatbots
Custom software built around the way your business works. Five steps, with a free pilot of 2 to 3 key modules before the full build.
Step 1: Understand
We learn how your business works.
Your requirements, workflow, challenges and goals, understood before anything is recommended.
Step 2: Plan
We design the right solution around your workflow.
Modules, workflows, roles, approvals, reports and integrations, agreed before development.
Step 3: Select Technology
Choose the right technical foundation.
Technology options matched to your users, security, budget and growth, not one fixed stack.
- Free pilot
Step 4: Pilot
Test our work before full project development.
Free. You choose 2 to 3 key modules and we build them first, so you can judge our work.
The full project starts only after you approve the pilot.
- Full project
Step 5: Build & Scale
From approved pilot to complete digital system.
Full development, testing, deployment, training and support, built to grow with you.
Guides for this decision
In-depth, practical reading for teams planning this kind of project.
- Guide · 9 min readHow to Evaluate an AI Pilot Before Rollout: Test Sets, Metrics and a Go/No-Go ScorecardAn AI pilot is ready for rollout when it meets acceptance criteria you set in advance, on a test set built from your own data, including the hard cases. This guide covers metrics by task type, human review thresholds, failure-mode testing and a go/no-go scorecard.Read the guide
- Guide · 9 min readEnterprise Knowledge Search with RAG: Permissions, Sources and AccuracyEnterprise knowledge search with RAG lets staff ask questions in plain language and get answers drawn from company documents, with citations, limited to what each person is allowed to see. This guide covers sources, permissions, accuracy and rollout.Read the guide
AI Assistants & Chatbots FAQ
A general chatbot answers from its training data and the open internet and doesn’t know your business. A custom assistant is grounded in your own documents, policies and data using retrieval, so it answers about your specific products and processes, cites sources, and stays within guardrails you control.
We reduce that risk by grounding answers in your data, restricting scope, citing sources, and falling back to a person or a safe response when the assistant is not confident. No system is perfect, so we measure accuracy on a test set of your real questions before launch, review logged conversations, and re-test after every change.
An assistant answers and helps in a conversation; an agent also completes multi-step tasks in your systems — reading a document, looking up an order in the ERP, updating a CRM record — within permissions you define and with a person approving the actions that matter. See our AI agent development service for how agents are built and controlled. We build either, or both, based on the use case.
Web chat, WhatsApp, your internal tools, or embedded inside your product. We build for where your users already are. WhatsApp needs a verified business account in your organization’s name.
We control what data the assistant can reach, document which data goes to which model provider, and configure the retention settings you choose. Where your policy requires it, a self-hosted model is an option. Conversations are logged for review with a retention period you set. We design to support your data-protection obligations; compliance remains your organization’s responsibility and is confirmed by your own assessment.
The pilot shows how accurately the assistant answers a test set of your real questions from your own content, how handover works, and whether your team trusts the answers. It does not prove behaviour at full production volume, on topics missing from the pilot content, or the running cost at scale. Those are measured in a controlled rollout after you approve the full build.
Project code, prompts, configuration and the evaluation set transfer to you on full payment, and your data is yours throughout, so you are not locked into a per-seat AI product you can’t change. Third-party foundation models remain the provider’s and are used under the provider’s terms, and open-source components stay under their own licences, as our IP and ownership policy sets out.
See working software before you commit
Before you commit to the full project, we build 2 to 3 of your key modules as working software, free of charge. Your team tests the pilot, and the full build starts only after you approve it.
See how the free pilot worksUnderstand
We learn your requirements and how your organisation works today.
Select pilot modules
Together we choose 2 to 3 key modules that prove the solution.
Build the working pilot
We build those modules as real, working software, free of charge.
You test it
Your team uses the pilot. The full project starts only after you approve it.
Start a conversation
Tell us how your business works.
Describe what is slowing your team down. We will help you work out what to build, and how a free pilot lets you judge our work before the full project.
Prefer WhatsApp? Start a chatWhat happens next
You send a short brief
The problem, the people involved and any target date. A senior engineer replies within 4 business hours.
We understand your workflow
A first call about how your business works today. An NDA can be signed before you share details.
You test a free pilot
You choose 2 to 3 key modules and we build them first, so you judge real software before the full project.