MVP update: what we built, what we learned, what's next

We launched the first version of Grovar to a small group of early users three weeks ago. Here's an honest breakdown of what happened.

What made it into v1

We originally planned to ship six agent types in the MVP. We shipped three.

The Growth Director, Strategy, and Campaign agents got cut from v1 — not because we don't believe in them, but because we needed to validate the core value proposition first: can we reliably run a campaign that performs better than what the business owner was doing manually?

What we did ship:

• Content generation connected to real product data. If you give Grovar your Shopify catalogue, it generates ads and posts that reference actual products, prices, and availability — not generic filler.
• Meta + Google campaign execution. Grovar can take a brief, generate 4–6 creative variants, set up A/B tests across audiences, and pause underperformers automatically.
• Customer chat agent. Handles FAQs, order status, and product recommendations. Escalates to a human when it doesn't know the answer.

What we learned

The bar for "autonomous" is higher than we thought. Early users wanted to understand why the system made the decisions it did. We underinvested in explainability in v1. This is now a first-class feature for v2.

Connecting to existing tools is the hardest part. Getting data out of Shopify, Meta, and Google in a consistent format took 60% of our engineering time. We expected 20%. This is actually our moat — every integration we build compounds.

Business owners don't think in channels. They think in outcomes. One user told us: "I don't care if it's Meta or Google, I just want more customers who buy high-margin products." We need to abstract the channel selection completely. That's the Growth Director Agent.

What's next

v2 (Q3 2026) — Growth Director Agent goes live. You set an outcome. The system decides the strategy. We're also adding TikTok and LinkedIn to the channel mix.

v3 (2027) — Full Growth Operating System. Autonomous planning, autonomous execution, autonomous optimisation across every growth function.

We're running a tight early access programme. If you want in, reach out — we're selective about who we onboard so we can actually learn from each customer.