Building a custom AI chatbot development takes 4-8 weeks for most UK businesses. A simple single-channel FAQ chatbot - answering common questions on your website - takes 4-6 weeks from discovery call to go-live. A multi-channel enterprise chatbot with CRM integration, WhatsApp Business API, voice escalation and a full GDPR audit takes 8-16 weeks. The timeline breaks into four phases: discovery and scoping (week 1), knowledge base setup and model configuration (weeks 2-3), integration and testing (weeks 4-5), and GDPR review with go-live (weeks 6-8). In our experience at Softomate, 80% of project delays are caused by slow data provision from the client side, not development speed. When clients arrive at the discovery call with their documents ready, we routinely go live 5-7 days ahead of the standard schedule.
Last updated: 18 May 2026
Published 18 May 2026Five factors drive AI chatbot build time: the number of channels the bot must operate on, how many existing systems it needs to integrate with, the size and quality of your knowledge base, the complexity of your GDPR obligations, and the number of distinct conversation flows required. Get these right in scoping and your timeline becomes predictable within a few days.
When a prospect asks us how long their chatbot will take, we always ask five questions before quoting a number. Those questions map directly to the five factors below. A solicitor in East London with one website channel, no CRM integration, a 40-page FAQ document and straightforward client data handling might be live in 4 weeks. A property management company needing WhatsApp, email, a CRM sync, a 500-document knowledge base and ICO-grade data handling might take 14 weeks. Both are normal - they are just different projects.
Here is how we classify projects at Softomate. We have delivered over 60 chatbot builds since 2021 and these categories hold up consistently across sectors including financial services, healthcare, property and professional services.
| Factor | Simple chatbot (4-6 weeks) | Standard chatbot (6-8 weeks) | Complex chatbot (8-16 weeks) |
|---|---|---|---|
| Number of channels | 1 (website widget only) | 2 (website + email or SMS) | 3+ (website, WhatsApp, voice, Teams) |
| CRM integrations | None or read-only webhook | 1 CRM (read and write, e.g. HubSpot) | 2+ systems (CRM, ERP, booking platform) |
| Knowledge base size | Under 50 documents or pages | 50-200 documents | 200+ documents, live data feeds |
| GDPR complexity | Standard privacy notice update | DPA with third-party processors | ICO audit trail, sector-specific obligations (FCA, CQC, SRA) |
| Conversation flows | 3-5 flows (FAQ, contact, escalation) | 6-12 flows including booking or quoting | 13+ flows, conditional logic, multi-language |
The model powering your chatbot matters less to the timeline than people assume. Whether we deploy GPT-5.4 or Claude 4 as the underlying language model, configuration and integration work takes the same amount of time. The model choice affects response quality and cost per query, not the build schedule. What does add days to a project is switching models mid-build after testing reveals a quality gap - which is why we lock this decision in discovery.
One point worth emphasising: the timeline above assumes an active, communicative client. When clients go quiet during testing or take two weeks to review documents, every week of silence adds a week to delivery. We build buffer into our standard timeline for this, but the fastest projects we have ever delivered have all had one thing in common - a client who treated the build like a sprint, not a task to delegate and forget.
Discovery is the single most important week of any chatbot project. We spend it understanding your business, your customers, your data and your systems well enough to write a scoping document that becomes the contract for everything that follows. Get this week right and the rest of the build is largely mechanical.
At Softomate, discovery is included in every project at no extra charge. It is not a sales call - it is a working session. By the end of week 1, we will have mapped every conversation the chatbot needs to handle, identified every system it needs to touch, and documented every piece of data it will process. If GDPR issues are going to block the project, we find them now, not in week 7.
Here is what our discovery week looks like in practice. We typically run two or three calls: a stakeholder kick-off (60 minutes), a technical access call with your IT team or agency (45 minutes), and a knowledge base review call (30-45 minutes). Between calls, our team reviews whatever documentation the client has provided and drafts conversation flow diagrams in Miro for client review.
The stakeholder kick-off covers your business objectives - what does success look like at 90 days? Which customer questions cost your team the most time? Where do conversations currently break down and result in a phone call? We also establish tone of voice here. A chatbot for a barristers chambers sounds very different from one for a fast-moving e-commerce brand, and that difference shows up in the system prompt and fallback language.
The technical access call is where we learn what systems you run. We need to know if you are on Salesforce, HubSpot, Zoho or a bespoke CRM. We need to know whether your website is on WordPress, Webflow or a custom stack. We need to understand your hosting environment and whether your IT policy allows outbound webhook calls. These questions sound administrative but they directly affect build time. A CRM with a well-documented REST API takes 2-3 days to integrate. A legacy system with no API and only CSV export can add 2-3 weeks to the project.
The knowledge base review is about understanding the raw material. We look at what documents you have, in what formats, and how current they are. PDFs, Word documents, Google Docs, website pages, Notion wikis - we can ingest all of these. The question is whether they are accurate, up to date, and comprehensive enough to answer the questions your customers actually ask.
By end of week 1, we deliver a scoping document covering: conversation flow diagrams, a confirmed channel list, a system integration specification, a data processing register draft, an itemised fixed-price quote, and a week-by-week delivery plan.
What the client needs to provide in week 1:
When clients arrive at the discovery call with these items ready and organised in a shared Google Drive or SharePoint folder, we save between 3 and 5 working days on the overall project. That is not an estimate - it is what we observe consistently across projects. The discovery phase becomes a genuine sprint rather than an information-gathering exercise spread across multiple follow-up emails.
Weeks 2 and 3 are the most technically intensive part of any chatbot build. We are turning your raw documents and business knowledge into a structured, queryable knowledge base that the language model can draw on accurately - and we are configuring the model itself to behave exactly as your business needs it to.
This phase is where the quality of a chatbot is really determined. A poorly built knowledge base produces a chatbot that hallucinates, gives outdated answers or fails to find information that is clearly in the documents. A well-built one produces a chatbot that answers correctly 94-97% of the time on in-scope questions from day one. We aim for the latter on every project.
The process follows a consistent sequence, though the time spent on each step varies by project size:
One thing we are rigorous about in this phase: we never move to integration until the knowledge base QA passes. It is tempting to start integration work in parallel to save time, but a chatbot wired into your CRM with a poor knowledge base creates confusing test results that are slow to diagnose. The sequence matters.
For larger knowledge bases - 200+ documents - this phase expands to 3 weeks. For a simple 30-page FAQ site, it can compress to 5 working days. The client review step in the middle is the most variable element. When clients prioritise this and return detailed feedback within 48 hours, we can close all gaps and move to integration on schedule.
Integration is where the chatbot connects to the real world - your website, your CRM, your WhatsApp number, your booking system. Weeks 4 and 5 are about making those connections reliable, then testing the whole system end to end under conditions that mirror actual customer use. This phase surfaces unexpected edge cases that no amount of isolated testing can catch.
Every integration we build uses webhook-based architecture by default. Rather than embedding the chatbot in a monolithic application, we expose a clean API that your website, WhatsApp Business API or any other channel can call. This means the chatbot logic is decoupled from your website stack - a WordPress update or a site migration cannot break the bot. It also makes adding new channels later a matter of days rather than weeks.
The most common integrations we build and their typical delivery times are shown below. Note that these times assume we have API access from week 1 - waiting for credentials or test environment access is the most common reason integration takes longer than planned.
| Integration type | Typical build time | Key dependencies |
|---|---|---|
| Website chat widget (custom JavaScript) | 1-2 days | Website platform access, brand CSS, CSP header rules |
| HubSpot CRM (contact creation, deal update) | 2-3 days | HubSpot API key, field mapping document from client |
| Salesforce CRM (read and write) | 3-5 days | Connected App credentials, sandbox access, Salesforce admin sign-off |
| WhatsApp Business API (via Twilio or Meta Cloud API) | 3-5 days (plus 3-7 days Meta approval) | Verified Meta Business account, approved phone number, message template pre-approval |
| Email escalation (automated handoff to human agent) | 1-2 days | SMTP credentials or SendGrid/Mailgun API key |
| Calendly or Acuity booking | 1-2 days | Calendly API key, event type configuration |
| Custom REST API (bespoke CRM or ERP) | 5-10 days | Full API documentation, test environment, authenticated access |
| Zoho CRM | 2-4 days | OAuth credentials, module mapping |
WhatsApp deserves special mention here because it has a dependency we cannot control: Meta's approval process. To use the WhatsApp Business API - whether via Twilio or Meta's own Cloud API - you need an approved Meta Business account and a verified phone number. This approval process typically takes 3-7 business days but can extend to 10-14 days for new business accounts or businesses in certain sectors. We start this process in week 1 of discovery as soon as a client confirms WhatsApp is in scope. Even so, Meta's approval timeline means WhatsApp projects rarely go live in under 5 weeks from contract signature, regardless of how fast we build.
The testing sequence in weeks 4-5 follows a structured path: unit testing of each integration in isolation, regression testing of the full conversation flow end to end, load testing at 10x expected peak concurrent users, and finally user acceptance testing (UAT) with real members of the client's team. UAT is where we consistently find the most valuable feedback. Actual users ask questions in ways that QA engineers never anticipate. We typically discover 3-8 edge cases in UAT that we close before go-live.
We ask clients to assign 2-3 team members to UAT for a structured 3-day window. We provide a test script covering the most important flows, but we also encourage free-form testing. Those unscripted sessions produce the most useful data. When clients invest properly in UAT, the hypercare period after go-live is dramatically quieter.
GDPR review is not a box-ticking exercise at Softomate - it is a technical and legal process that determines whether your chatbot can legally process the personal data your customers share with it. In weeks 6-8, we complete a structured compliance review, update your privacy documentation, and then move through a staged go-live with a monitoring period before full launch.
The ICO's AI and data protection guidance (updated in 2025) is explicit: if your chatbot processes personal data - which virtually every customer-facing chatbot does - you need a lawful basis for that processing, a DPIA (Data Protection Impact Assessment) if the processing is high risk, and clear disclosure in your privacy notice. For businesses in regulated sectors, the requirements go further: FCA-regulated firms need to consider Consumer Duty obligations around AI-assisted communications, CQC-regulated services need to consider access-to-care implications, and SRA-regulated solicitors need to consider legal professional privilege.
Our GDPR review in weeks 6-8 covers the following steps. First, we finalise the data processing register we drafted in discovery - documenting every category of personal data the chatbot collects, the lawful basis for processing it, the retention period, and the processors involved (the AI model provider, the vector database host, any CRM the data flows into). Second, we draft or update your privacy notice to include the chatbot's data practices in plain English. Third, we put a Data Processing Agreement (DPA) in place with Softomate as a data processor and with any sub-processors we have introduced. Fourth, for high-risk deployments, we complete a DPIA and share it with your DPO or legal team for sign-off.
Soft launch comes before full public deployment. We go live with a restricted URL or a non-indexed page, or we enable the chatbot only for a specific segment of traffic (e.g. returning logged-in users). This soft launch window typically runs for 5-10 working days. During this period, we monitor every conversation in real time, review fallback triggers (moments the bot could not answer and escalated to a human), and track response latency. If anything looks wrong, we can fix it without the pressure of full public visibility.
Full go-live is followed by a 30-day hypercare period. During hypercare, we hold a weekly 30-minute call with the client to review performance metrics, triage any issues that have emerged, and make knowledge base updates as needed. After hypercare, projects typically transition to a monthly retainer starting from £350 per month, which covers knowledge base maintenance, model updates and monthly performance reporting.
The fastest chatbot projects we have ever delivered moved quickly because the client came prepared, stayed engaged and made decisions promptly. Development speed is rarely the bottleneck - client-side readiness is. Here are the six things that consistently accelerate delivery, with specific time savings based on our project data.
We are transparent with every client about where the schedule risk sits. If you want an aggressive timeline, we will tell you exactly what you need to do on your side to achieve it. We have gone from contract signature to go-live in 18 working days for a simple single-channel FAQ chatbot with a fully prepared client. We have also had 14-week projects extend to 22 weeks because of slow document provision and delayed system access. The build quality in both cases was identical - the difference was entirely client-side readiness.
Yes, for a very simple single-topic chatbot - for example, a bot that answers 15-20 standard questions about one service with no CRM integration. However, if WhatsApp is in scope, Meta's phone number approval process alone takes 3-7 business days and cannot be expedited. For a website-only FAQ chatbot with documents ready from day 1, our minimum viable timeline is 3 weeks. Rushing below this risks a knowledge base that has not been properly tested and a GDPR position that has not been reviewed - both create problems later that cost more to fix than the time saved.
Client-side delays, consistently. The three most frequent causes are: slow document provision (clients taking 2-3 weeks to gather knowledge base materials after contract signature), late system access (CRM or website credentials arriving in week 3 instead of week 1), and unclear escalation rules (no confirmed decision on who the chatbot should hand conversations to when it cannot help). Development delays are rare when scoping has been done correctly. Our internal build stages have run late on fewer than 5% of projects in the past two years.
Not automatically. Build time is driven primarily by the complexity of integrations and the size of the knowledge base - not by chatbot quality. A well-scoped, well-tested 4-week chatbot will consistently outperform a poorly-scoped 12-week one. The quality determinants are knowledge base accuracy, system prompt engineering and the rigour of QA testing - all of which we apply at the same standard regardless of project length. Complexity adds time; rigour adds quality.
Most heavily in week 1 (discovery calls and document provision), lightly in weeks 2-3 (reviewing knowledge base gaps when flagged), actively in weeks 4-5 for user acceptance testing, and then at go-live approval in weeks 6-8. Total client time commitment across a standard 6-week project is typically 8-12 hours. We design the process this way deliberately - your time is valuable and we want focused input at the moments it matters, not endless status calls throughout the build.
Every project includes a 30-day hypercare period with weekly review calls, real-time conversation monitoring and priority fixes for any issues that emerge. After hypercare, we offer a monthly maintenance retainer from £350 per month. This covers knowledge base updates when your products or services change, model updates as GPT-5.4 and Claude 4 release new versions, monthly performance reporting with resolution rate and escalation metrics, and priority support response within 4 business hours for production issues.
Most UK businesses can have a custom AI chatbot live within 4-8 weeks - a timeline that is realistic, deliverable and confirmed in writing after our free discovery week. The biggest variable is not development speed but client readiness: projects where clients arrive organised, stay engaged during UAT and make decisions promptly consistently finish 5-10 days ahead of schedule. Projects with complex multi-channel requirements - WhatsApp Business API, CRM read-write access, ICO-grade GDPR obligations - sit at 8-16 weeks, a figure driven by integration complexity and third-party approval processes, not development capacity. Over 60 chatbot projects, the single most reliable predictor of on-time delivery has been a well-run discovery week with all documents ready before the first call.
Planning a chatbot project? Get a scoping call with Softomate - we will confirm your realistic timeline, identify any GDPR obligations upfront and give you a fixed-price quote. Most scoping calls take 45 minutes.
Written by the Softomate Solutions AI Development Team, Barking, East London. We have delivered custom AI chatbots for UK businesses in property, financial services, healthcare and professional services.AI chatbot development costs in the UK range from £3,000 for a simple FAQ chatbot to £25,000+ for a fully integrated conversational AI with CRM and booking system integration. Monthly running costs are typically £100-£400. Softomate Solutions builds AI chatbots from £3,500 with a 3-4 week delivery timeline and full UK GDPR configuration included.
For customer-facing use, a custom AI chatbot trained on your specific business knowledge, pricing and services significantly outperforms a generic ChatGPT integration. A custom chatbot knows your products, your pricing, your service area and your compliance requirements. It also integrates with your CRM, booking system and WhatsApp - capabilities ChatGPT plugins cannot replicate without custom development.
More from the blog
Let us help
Talk to our London-based team about how we can build the AI software, automation, or bespoke development tailored to your needs.
Softomate Solutions Limited
Deen Dayal Yadav
Online
