Build vs buy: should your team code its own chatbot?
Building your own chatbot means owning WhatsApp API integration, template management, hosting, and ongoing maintenance — real engineering time beyond the initial build. Buying a platform trades that engineering effort for a subscription fee. Building tends to make sense only when your requirements are genuinely unusual and you already have the engineering capacity to maintain it long-term.
- The WhatsApp Business API itself is free to integrate with, but the integration work, template management and hosting are not free in engineering time.
- A platform's flow builder replaces weeks of engineering work with hours of no-code configuration — that time has a real cost either way.
- Building is easier to justify when you have in-house engineering capacity already, and your requirements do not fit any existing platform.
- A hybrid approach — a platform for the core bot, with custom code via webhooks or the REST API for special cases — often beats an all-or-nothing choice.
- Maintenance (Meta policy changes, template resubmissions, provider updates) is ongoing, not a one-time cost — budget for it either way.
The hidden costs of building your own
- WhatsApp API integration. Meta's Cloud API access itself has no separate licence fee, but connecting it, embedded signup, webhook handling, template submission and approval workflow, message-status tracking, is real development work, not a weekend project.
- Template management. Every Marketing, Utility or Authentication template needs to be created, submitted for Meta's approval, and resubmitted if rejected or if policy changes, an ongoing task, not a one-time setup.
- Voice, if you need it. Adding AI voice calling means integrating a telephony provider's API (Exotel, Plivo, Twilio or SIP), handling call state, DTMF/speech gather, and voicemail detection logic, a separate engineering effort from chat entirely.
- Hosting and uptime. Someone needs to keep the system running, patched, and monitored, an ongoing operational cost that does not stop after launch.
- Maintenance as Meta's platform evolves. Meta periodically changes template policies, rate limits and API versions, a custom build needs someone watching for these changes and updating accordingly.
What a platform trades that engineering time for
A no-code flow builder, pre-built voice blocks, template management with approval tracking, and a maintained integration layer replace the list above with a monthly subscription. The comparison is not 'free (build) vs paid (buy)', building has real, ongoing engineering cost, just paid in salaries and time rather than a subscription line.
A rough comparison framework
| Build in-house | Buy a platform | |
|---|---|---|
| Upfront cost | Engineering time (weeks to months, depending on scope) | Setup time (hours to days, using no-code tools) |
| Ongoing cost | Engineering maintenance time, hosting | Monthly subscription + usage costs |
| Flexibility for unusual requirements | High, build exactly what you need | Bounded by the platform's features, extendable via API/webhooks |
| Time to first working bot | Weeks to months typically | Days typically, for standard use cases |
Fill this in with your own team's actual capacity and hourly cost to make it a fair comparison rather than an abstract one.
When building genuinely makes sense
Building in-house is a reasonable choice when your requirements are genuinely unusual, deep integration with a proprietary internal system that no platform's connectors reach, or a product where the chatbot itself is your core product rather than a support tool, and you already have engineering capacity that would otherwise sit idle. If you would need to hire specifically for this project, the calculation shifts significantly toward buying, since that hiring cost rarely shows up in a simple build-vs-buy spreadsheet.
The hybrid path most businesses actually take
Rather than an all-or-nothing choice, most businesses use a platform for the core conversational flows (FAQs, bookings, order updates) and extend it with custom logic only where genuinely needed, via a JavaScript block for custom calculations inside a flow, an HTTP/API request block to call an internal system, or inbound/outbound webhooks to connect proprietary tools. This captures most of the speed benefit of buying while keeping room for the 10% of requirements that are genuinely unusual to your business.
Convo360's Flow Builder includes a JavaScript block, HTTP/API request block, Google Sheets and Webhook (HMAC-signed) blocks specifically so a no-code core does not become a dead end when a business has one unusual requirement. The REST API on the Business plan lets a development team pull or push data programmatically alongside the flow builder, rather than choosing build or buy as a one-time, irreversible decision.
Voice adds a separate layer of build cost
If AI voice calling is part of the requirement, the build-vs-buy gap widens further. A custom build needs a telephony provider account (Exotel, Plivo, Twilio or a SIP trunk), call-state handling, a gather step for DTMF or speech input, voicemail/answering-machine detection, and transfer logic to a human, each a separate integration on top of everything chat already needed. A platform that ships these as ready-made voice blocks (gather, DTMF menu, transfer, play audio, voicemail) turns weeks of telephony-specific engineering into flow configuration, which is usually the single biggest gap between the two paths once voice is in scope.
Questions to ask before committing either way
| Question | Why it matters |
|---|---|
| Do we have engineering capacity that is otherwise idle, or would we need to hire? | Hiring cost rarely appears in a simple build-vs-buy spreadsheet but is often decisive. |
| Is our requirement genuinely unusual, or does it just feel that way because we haven't checked a platform's connectors? | Many 'unusual' requirements turn out to be a webhook or API call away on an existing platform. |
| Who maintains this in two years if the original engineer leaves? | Custom builds carry key-person risk that a maintained platform does not. |
| Does the requirement involve voice as well as chat? | Voice roughly doubles the integration surface for a custom build. |
Neither choice is permanent. A business that starts on a platform can still hand specific, unusual workflows to an in-house developer later via webhooks or the API, without rebuilding the whole system. A business that starts by building can migrate to a platform later too, though it means re-creating flows and re-submitting WhatsApp templates, so it is worth trying a platform's free trial before committing engineering time, simply to see how much of the requirement it already covers out of the box.
Example conversation
See what you can build in Flow Builder
Start free in minutes — no code, no credit card. Bring your own AI keys and pay no per-message markup.
Frequently asked questions
Is the WhatsApp Business API free to use?
Meta's Cloud API access itself typically has no separate licence fee, though per-message charges apply per Meta's own pricing. The real cost of building in-house is the engineering time to integrate, maintain and extend it, not the API access itself.
How long does it take to build a WhatsApp chatbot from scratch?
It varies widely by scope and team experience, but integrating the API, building template management, and handling edge cases typically takes weeks to months of engineering time, compared to days on a no-code platform for standard use cases.
Can I switch from a custom build to a platform later?
Yes, though it involves migrating your WhatsApp Business Account, templates and contact data, and rebuilding your flow logic in the new platform's builder. Check how portable your existing setup is before starting a migration.
Does buying a platform mean giving up customisation?
Not entirely — most platforms offer extension points like webhooks, an API, and code blocks inside the flow builder, letting you customise specific parts without building the whole system from scratch.
What is the biggest ongoing cost of a custom-built chatbot?
Maintenance — keeping up with Meta's policy and API changes, resubmitting templates, and fixing issues as usage grows — is usually a bigger long-term cost than the initial build, and it is easy to underestimate upfront.
Related reading
Available across India — see WhatsApp & AI calling automation by state and city.