Last updated: 4 May 2026
A DIY chatbot is a self-service tool (Intercom Fin, Drift, Tidio, Crisp, ChatGPT plugins) that a business owner signs up for and configures themselves on a monthly software subscription. A managed AI system is an AI tool deployed by an external operator over a two-week sprint and then maintained on a monthly retainer that includes conversation monitoring, prompt tuning, knowledge-base updates, integration maintenance, and performance reporting. DIY chatbots typically cost £50–£500/month in software fees plus 4–8 hours per week of the owner's time to keep them working. Managed AI systems typically cost £2,500–£6,000 to deploy plus £400–£1,500/month managed, with zero ongoing owner time. DIY wins on upfront cost; managed wins on total cost of ownership, conversion rate, and time recovered.
The 'AI chatbot' category looks like one market from the outside, but there are actually two products being sold under the same label, and they produce very different outcomes for small businesses.
A DIY chatbot is a software-as-a-service product. The business signs up, configures the bot through a dashboard, writes its own responses or imports its own knowledge base, connects its own integrations, and runs the monitoring itself. The vendor provides the infrastructure and pays for improvements to the underlying AI models. The customer provides everything else: content, training, monitoring, optimisation, integration work. Pricing is a monthly subscription, typically £50–£500 depending on the platform and conversation volume.
A managed AI system is a service. The provider deploys the AI over a short sprint, trained specifically on the client's business, integrated into the client's real tools, and then operates and maintains it on an ongoing retainer. The client provides access to information and workflows. The provider does everything else. Pricing is a fixed deployment fee plus a monthly retainer.
This distinction matters because the work of keeping an AI system good is substantial. DIY platforms make that work the customer's responsibility. Managed systems make it the provider's responsibility. For small businesses without in-house AI capability, the economics and outcomes diverge sharply.
DIY chatbot platforms market themselves on speed of setup. The ten-minute onboarding is real. What the marketing omits is the ongoing work required to keep the chatbot producing value after week two.
Training on the real business. Platforms like Intercom Fin and Drift AI let you 'point at your website' and claim to train on your content. In practice, the bot trained this way produces generic responses, hallucinates prices and policies, and fails to answer specific operational questions (availability, qualifications, specific service details). Real training requires curating inputs, not just pointing.
Response tuning as edge cases appear. Real users ask questions in ways the business didn't anticipate. Each unexpected question reveals a gap in the prompt. Somebody has to notice the gap, write a correction, and test it. On a DIY platform, that somebody is the business owner.
Integration maintenance. Calendar APIs change. CRM vendors deprecate endpoints. Email providers rotate authentication schemes. A DIY chatbot integrated with half a dozen tools will have at least one integration break every quarter. Diagnosis and fix takes hours.
Knowledge currency. Prices change. Services evolve. Team members join and leave. A bot still quoting 2025 pricing in 2026 is a liability, not an asset. Keeping content current requires a recurring update workflow.
Conversation review. Without reading what the bot is actually saying to real customers, there is no way to know if it is helping or quietly destroying trust. The business that deployed a DIY chatbot and then never reads transcripts has a statistical chance of an embarrassing failure that they never discover until the customer complains publicly.
For a typical service business, this ongoing work consumes 4–8 hours per week. In a team where nobody is hired to do it, it falls off. The chatbot slowly decays into a source of bad answers and is quietly disabled.
Both categories can technically do the same things. The difference is whether those things are maintained in production.
Natural-language conversations. Both DIY and managed AI deliver this well in 2026 — the underlying models are commodity. Tie.
Trained on the real business. DIY: possible but requires ongoing curation the owner has to do. Managed: done as part of the sprint and kept current on the retainer. Managed wins in practice.
Live calendar and CRM integration. DIY: supported by most platforms but configuration and maintenance are the owner's job. Managed: configured, tested, and maintained by the provider. Managed wins on reliability.
Qualification logic. DIY: most platforms support conditional flows; the owner writes and maintains them. Managed: the provider builds the qualification logic against the business's defined good-fit criteria. Similar capability, different burden.
Meeting booking inside the conversation. DIY: typically supported via calendar integration. Managed: same, but the provider handles the edge cases (timezone handling, buffer time, no-show follow-up) without the owner writing them.
Follow-up automation. DIY: many platforms support sequences; the owner writes them. Managed: the provider builds the sequences into deployment and tunes them based on results.
Compliance disclosure (EU AI Act Article 50). DIY: available on most platforms but requires the owner to configure and document. Managed: handled as part of deployment with documented compliance posture.
Monthly performance reporting. DIY: dashboards exist but the owner has to read them and interpret. Managed: the provider produces a written monthly report with recommendations.
The aggregate pattern: DIY platforms give you the tools; managed AI gives you the outcome.
Comparing sticker prices is misleading. The honest comparison is total cost of ownership over 12 months, including time and lost revenue.
DIY chatbot over 12 months. Software subscription £600–£6,000 (depending on platform and volume). Owner time 4–8 hours per week at realistic opportunity cost — even valued conservatively at £50/hour, this is £10,000–£20,000 per year of hidden labour. Most of this labour does not happen, which means the bot is under-maintained and the lead-capture lift is half or less of what is possible. Realistic 12-month cost including hidden labour and lost pipeline: £11,000–£26,000, producing a moderate lead-capture lift that degrades over time.
Managed AI system over 12 months. Sprint £2,500 + 12 × £400/month retainer = £7,300 all in. Owner time: near zero. Lead-capture lift: sustained across the year because the system is maintained. Realistic 12-month cost: £7,300, producing a sustained lead-capture lift.
The managed AI system is cheaper in total cost of ownership for any business where the owner's time is worth more than £20/hour, which is every service business. The apparent premium in the sticker price is an illusion created by moving the labour cost off the invoice and onto the calendar.
Where DIY is genuinely cheaper: businesses with very low enquiry volume where the lift from AI is not commercially meaningful anyway, and businesses with internal AI operations capacity where the maintenance labour already exists.
Managed AI is not always the right recommendation. There are three situations where a DIY chatbot is the correct choice.
The business has internal AI operations capacity. A growth-stage SaaS company with an in-house data team or AI engineer can run DIY platforms effectively because the maintenance labour already exists on payroll. For these businesses, a DIY platform at £200/month plus internal labour is cheaper than a managed retainer.
The use case is non-critical. An internal productivity chatbot for employee FAQ, a low-volume informational widget, or a prototype experiment does not justify the managed overhead. DIY is fine for low-stakes surfaces.
The business is validating the concept. A pre-revenue or early-stage business testing whether AI chat converts for their audience should use a DIY platform for the first 3–6 months to establish whether the category works for them at all. Once it is proven, switching to managed for production is the right move.
For every other profile — revenue-producing service businesses, businesses where response quality matters commercially, businesses without in-house AI capability — managed AI is the correct choice. The cost of a badly-maintained DIY chatbot in lost pipeline exceeds the cost difference.
For a small business deciding between DIY and managed AI, five questions resolve the choice.
Does the AI sit on a revenue-producing surface? If the chatbot's answers affect whether prospects book, buy, or sign, it is production. Production surfaces should be managed.
Does anyone in the business have the skills and time to maintain it weekly? If there is no named person with allocated hours, the DIY platform will decay. Pick managed.
Does the business change materially month to month? If services, pricing, team, or availability shift regularly, the knowledge base needs ongoing updates that DIY will not receive. Pick managed.
Does the business operate in a regulated sector or face compliance exposure? Healthcare, legal, financial services, and any AI touching hiring or scoring need documented compliance posture, which managed providers build in and DIY platforms leave to the customer. Pick managed.
Does the owner want to own an AI ops project, or want the lift without owning the project? This is the honest question. Some owners genuinely enjoy the tooling. Most do not and should not pretend otherwise. Pick whichever matches reality.
BRVO's Nerve is a managed AI lead capture deployment built for the 'production surface, no in-house AI ops' profile — which is the majority of UK service businesses. For businesses in the DIY-correct profile, a well-chosen platform and allocated maintenance time is a defensible alternative.