Add a featured image here
We wrote about business chatbots a few years ago, back when Unilever deploying an HR bot felt genuinely radical. We talked about ManyChat as an emerging platform and suggested that businesses of all sizes needed to get their heads around automation.
All of that was right. But the landscape has changed so dramatically since then that the original post is basically a history lesson now. So here's the update, because the question "should my business use AI chat tools?" has a very different answer in 2026 than it did when we first asked it.
Rule-based chatbots following scripted decision trees. Smart, but rigid. Required significant setup and ongoing maintenance to handle edge cases.
Generative AI that understands context, handles open-ended questions naturally, and improves over time. The gap between a human response and an AI response has narrowed significantly.
Unilever's Una: Still Going, Now Bigger
When we first wrote about Unilever's HR chatbot Una, she was a novel concept: an AI-powered assistant handling HR queries across 106 countries in 32 languages. The company made a bold commitment at the time, switching off HR phone numbers and email addresses when Una went live in each country.
It worked. Una is now referred to internally as "Unabot" and has expanded well beyond its original HR brief. By end of 2024, Unilever had trained 23,000 employees in AI usage, with Unabot serving as the first point of contact for employee queries spanning HR policies, IT support, and even campus logistics like shuttle bus timings. The chatbot that started as an experiment became foundational infrastructure.
The lesson from Unilever isn't just that chatbots work at scale. It's that the companies seeing the most value made a serious commitment: proper training, genuine integration into existing workflows, and the discipline to actually shift behaviour rather than just add another tool alongside the old process.
Did Claude and ChatGPT Replace Everything?
Fair question. When large language models went mainstream in 2022 and 2023, a lot of people assumed purpose-built chatbot platforms would get wiped out. Why would you build a scripted flow when you could just point a capable AI at your business and let it answer questions?
The reality is more nuanced. Claude, ChatGPT, and their equivalents are extraordinarily capable at open-ended conversation, content generation, and reasoning tasks. They're genuinely useful as business tools. But they have a fundamental limitation for many business use cases: they're general purpose, not action-oriented.
The businesses that are winning right now are combining both: a platform like ManyChat handling the engagement and action layer on social channels, with LLM-powered responses making those interactions feel genuinely human rather than scripted.
ManyChat Just Raised $140 Million
If you needed a signal that the purpose-built chatbot platform isn't going anywhere, ManyChat raised $140 million in a Series B round in April 2025, led by Summit Partners. The company now has around 1.5 million customers across 170 countries, sending billions of messages annually on behalf of clients including Nike, the New York Times, and Yahoo, as well as individual creators and small businesses.
ManyChat's CEO Mike Yan argues that what sets his platform apart from general AI chatbots is the focus on encouraging further action, not just answering questions. The difference is engagement versus information. A customer asking your bot a question wants an answer. A prospective buyer interacting with your Instagram needs to be guided toward a decision.
The platform has also integrated AI heavily into its own product. Current AI features include AI Replies that train on your website and business information, AI Comments that auto-reply based on your past reply style, AI Goals that follow up after interactions to hit specific targets, and an AI Flow Builder that generates entire automation sequences from a brief description.
What This Means for Small Business in 2026
The tools have never been more accessible or more capable. Here's the practical breakdown.
For customer-facing chat on your website
You can now train a chatbot on your entire website, product catalogue, and FAQ using tools like Chatbase or a custom Claude/GPT integration, and deploy it for a fraction of what it cost three years ago. A well-configured AI chat widget can handle the majority of pre-sale questions without human involvement.
For social media engagement
ManyChat remains the standard for Instagram and Facebook automation. The comment-to-DM flow, where someone comments on your post and automatically receives a direct message with your offer, is one of the highest-converting tactics available to small businesses right now, and it requires no coding. Starting price is $15 per month.
For internal operations
The Unilever model scales down. If you have repetitive internal questions about your processes, products, or team, training an LLM on your internal documentation and giving your team access is a meaningful time-saver. Tools like Notion AI, custom GPTs, and Claude Projects all support this use case.
For customer support
A tiered approach works best: AI handles the common questions automatically, escalates to a human for anything complex. Most good helpdesk platforms now have this built in natively.
The PlainBlack Take
The businesses that waited to see how AI chat would shake out now have clear, affordable, tested options. The conversation has moved from "should we?" to "which tool fits which problem?" If you haven't started somewhere, start with ManyChat for social and a simple AI chat widget for your website. Both are free to start and will show you results within days.
Our AI playbooks cover both in detail, with specific setup steps for your industry and business type.