Richgro has been helping
Australians grow things
since 1916. Now their
website does too.
For 110 years, Richgro's expertise lived in the hands of its agronomists and the knowledge of its staff. Customers who called got answers. Customers who visited the website — Australia's most visited garden brand, with over 3 million gardeners in its customer base — mostly didn't. They left with unanswered questions and unpurchased products.
We were brought in to change that. Not with a chatbot. With a purpose-built AI platform that thinks, recommends, and sells like Richgro's best person on the floor.
Richgro's product catalogue is vast — fertilisers, potting mixes, soil conditioners, pest control, weed management — each with specific application logic depending on plant type, season, and soil condition. No generic chatbot could navigate that. And Richgro's customer base isn't monolingual: across Australia's multicultural cities, a significant share of home gardeners don't speak English as a first language. Support staff were fielding the same questions at scale. Marketing had no visibility into what visitors actually wanted. The gap between "visits the website" and "buys the right product" was leaking revenue every single day.
An AI partner, not a SaaS vendor. Richgro needed custom intelligence — trained on their exact catalogue, their brand voice, their agronomic expertise — that could serve as a knowledgeable garden advisor at scale. One that could answer "what do I use on my frangipani?" in Mandarin, recommend the right product, and route that customer toward a purchase. And one that could feed every interaction back as structured data, so the marketing team could finally understand what their 3 million customers actually needed.
Four capabilities.
One platform. Built for Richgro.
We don't deploy off-the-shelf tools. Every capability below was designed from the ground up around Richgro's products, customers, and business model.
- Understands gardening intent — not just keywords. "My roses are going yellow" triggers the right diagnosis and the right product recommendation, not a search result.
- Serves customers in English, Mandarin, Vietnamese, Arabic, and four additional languages — switching naturally mid-conversation without any user prompt.
- Actively guides purchase decisions: from problem identification through to product selection, stockist lookup, and checkout — 24 hours a day.
- Branded as Richgro's own. Customers interact with Sprout, not with any third-party AI tool.
- Every inbound query is automatically classified by intent — product discovery, pest diagnosis, purchase issue, shipping — and routed accordingly. No human sorting required.
- Follow-up workflows triggered automatically from conversation signals: a customer asking about a product three times gets a targeted email. A stalled cart gets a Sprout prompt.
- Seamlessly integrated into Richgro's existing operations stack — no rip-and-replace.
- The support team now fields genuinely complex issues. Routine queries are handled entirely by the platform.
- The dashboard surfaces top product intents, seasonal demand spikes, and catalogue gaps in real-time — information that previously required surveys, focus groups, or guesswork.
- Sessions are automatically categorised by intent: product discovery, buy intent, pricing, problem diagnosis, how-to — giving the marketing team a live feed of customer needs.
- Weekly AI-generated summaries replace manual reporting cycles entirely.
- Audience segments built from real behavioural signals power more relevant, higher-converting campaigns.
- A customer photographs yellowing leaves, root rot, or an unfamiliar pest. Sprout identifies the condition, recommends the correct Richgro product, and explains how to use it.
- Dramatically reduces friction for non-expert gardeners who don't know the terminology to search for a solution.
- Visual catalogue QA runs automatically — product images are checked for quality and consistency at scale, flagging issues before they reach the storefront.
- Creates a support experience that no competitor — and no generic AI tool — can replicate without Richgro's proprietary product data.
This is Sprout.
Serving real customers, today.
Our team didn't just solve one problem. We looked at the whole operation and built AI into parts of the business that helped customers 24 * 7. The ROI showed up faster than we expected — and it's compounding."
What changed, and by how much.
| 300+ |
Daily active AI-assisted sessions
From zero to 300+ unique users per day engaging with Sprout — within weeks of launch, without any paid promotion to the feature.
|
|
| 4× |
Sales conversion uplift from chat-initiated sessions
Customers who engage with Sprout before purchase convert at four times the rate of those who don't. The recommendation engine works.
|
|
| 91% |
Queries fully resolved by AI — no human required
Nine in ten customer enquiries are diagnosed, answered, and resolved by Sprout without any staff involvement. The remaining 9% are complex edge cases escalated appropriately.
|
|
| −68% |
Reduction in inbound support tickets
Workflow automation intercepted the high-volume, low-complexity tier of support entirely — freeing the team for work that actually requires a human.
|
|
| 100+ |
Languages served — automatically
Sprout detects and responds in the customer's language with no setup, no manual routing, and no degradation in product recommendation quality across any language.
|
Your customers are leaving
with unanswered questions.
We can fix that.
We build full-spectrum AI for businesses with complex products and real customers. We don't sell templates or retrain generic tools. We build bespoke intelligence — trained on your business, integrated into your operations, live in weeks.