At Alhena, we ship what matters most to our customers. Here's a rundown of the most significant releases of the past month:
A new Returns Management Agent
For most ecommerce brands, returns are a big unmanaged cost that often eat 15 to 30% of revenue. And every returns conversation is also a retention moment: handle it well and the customer comes back, handle it badly and they don't.
So we built a dedicated Returns Management Agent — a purpose-built AI agent with its own prompts, logic, and handoff rules, trained specifically for the nuance of returns conversations. It's not a generic chatbot learning returns on the side. It's an agent built for the job.
And because we know ops teams don't have time for config work, it auto-registers across all your bot profiles the moment it's enabled. No setup. No ticket to engineering. It just starts working.
Enhanced BigCommerce Integration
This month, BigCommerce stores receive deeper integration with Alhena— REST API connectivity, product catalog sync, real-time stock checks, order management, the product finder, and full dashboard integration. Everything ships together, not drip-fed release by release.
If you're on BigCommerce and you've been waiting for AI that treats your stack seriously: the wait is over. See our full BigCommerce integration guide for a complete walkthrough.
Precision control for Shopify merchants
The brands we work with are increasingly sophisticated about what AI should and shouldn't be allowed to do on their behalf. "Let the bot do everything" is not a serious answer for a serious CX team.
So we've broken out every individual Shopify action into its own toggle on the dashboard. Cancel an order, look up a gift card, check return eligibility, create a return, create an exchange, pull up a discount, list active promotions. Each one is now controlled independently. You decide exactly what your AI can and can't touch.
We've layered a safety guardrail on top of it: if the underlying Shopify permission hasn't been granted, the toggle stays disabled, and we tell you exactly what's missing and how to re-authorize. No mysterious failures, no silent permission gaps.
One quieter improvement that some of our most sophisticated customers will love: nested metaobject references now resolve up to three levels deep. Translation — if you store FAQ groups, ingredient lists, or product specs as linked objects in Shopify (as most mature brands do), that content now flows through to your AI agent. Previously, it was silently dropped. That's fixed.
More voice options to suit your brand
Voice AI has crossed a threshold. The newest voice models no longer sound like machines reading a script — they sound like people. And as voice becomes a serious channel for shopping and support, the specific voice you choose becomes part of your brand.
We've added ElevenLabs as a TTS provider, bringing eight new voices alongside our existing Cartesia lineup: Eryn, Joseph, Justin, Katherine, Liz, Nathanial-C, John Fernandes, and Katie X. You now have real depth of choice to match a voice to your brand personality, not just "settle for what's available."
A new home for AI search optimization
This one matters more than it might sound. Your customers are no longer starting on Google. They're asking ChatGPT, Gemini, Perplexity, and Claude "what should I buy?" and those answer engines are making recommendations without you in the room.
The new AI Visibility tab in the dashboard is where you take that conversation back. In one place, you get a full picture of how your catalog shows up across AI answer engines: content coverage, gap analysis, search volume, prompt opportunities, and crucially a detail view with the specific steps to get recommended more often. Pagination, filtering, and sorting let you work through opportunities systematically instead of drowning in them.
And because different brands have different audiences, you can now choose which AI engines are used for your visibility analysis - OpenAI, Gemini, Perplexity, or Claude. Prompt limits are now tiered transparently by plan (25 for Essentials, 75 for Growth, 150 for Scale), with a clear prompt if you hit your ceiling. No surprises, no hidden throttling.
Smarter defaults, more capable models
Three changes that sound small but compound:
- Suggested questions are on by default. The agent that generates smart question suggestions now runs on every bot profile out of the box — not just new ones. Existing customers get the lift immediately, without having to know the setting existed.
- Chat Completions is now the default API. We've moved from OpenAI's Responses API to Chat Completions as our default mode, for broader model compatibility and more consistent reliability across the board.
- Gemini 3 thinking levels are exposed. You can now dial Gemini 3's reasoning depth to low, medium, or high, depending on how much "thinking" the task actually needs. Faster and cheaper when speed matters; deeper when quality matters.
Under the hood: a cleaner, more capable platform
The unglamorous work that keeps a product great:
- Rich text editor migration. Every text box in Alhena (the ones your team types into every day) was built on an older library called Draft.js, which is no longer actively maintained by its creators. We've migrated everything to Lexical, its modern, actively supported successor, with full feature parity: markdown, headings, code blocks, links, lists, raw mode. Same experience on the surface, faster and more stable underneath, and future-proof.
- The Pro plan now has three tiers — Essentials, Growth, and Scale — so instead of a one-size-fits-all package, customers pay for the size they actually are. Feature limits at each tier map to where our customers genuinely are in their journey, whether that's just getting started or running at real scale.
- A tools dashboard cleanup. Deprecated tools (lead generation, conversation markdown, OSS manager, related question prompt) have been retired; the sidebar is refreshed; and developer API settings now live in the manage admins section, where that kind of technical configuration logically belongs.
New to Alhena? Sign up to get started, or book a demo to see it live.
