Question 1

What technology powers Alpha Level's chatbots?

Accepted Answer

Every Alpha Level chatbot is a custom build on Anthropic's Claude model family — Sonnet 4.6 as the default reasoning engine, Haiku 4.5 as a cheaper fallback for narrower intents, and Voyage embeddings to power the vector knowledge base. We pick the model per intent rather than locking the whole conversation into one tier: a billing-lookup intent that needs deterministic tool use runs on Haiku, while a discovery question that needs multi-step reasoning routes to Sonnet. The knowledge base is versioned in a vector store so updates ship without retraining, and every conversation is logged with the model, latency, and resolution outcome so we can tune costs and accuracy together. No fine-tuning on customer data, no model lock-in.

Question 2

How much of customer support can an Alpha Level chatbot actually resolve?

Accepted Answer

Up to eighty percent of conversations resolved without a human, with clean escalation when the bot is out of scope. Resolution rate depends on three inputs: how clean your knowledge base is at kickoff, how repetitive your top-twenty query categories are, and whether the bot has read-write access to the systems where action happens (Shopify, Stripe, Zendesk, CRM). Bots get smarter monthly because failed queries are auto-flagged, reviewed by a senior trainer, fed back into the knowledge base, and the escalation rate drops over time. We publish the monthly resolution-rate chart in the client dashboard so the improvement curve is visible. The SupportHub deployment hit seventy-eight percent inside the first ninety days.

Question 3

When does a chatbot NOT make sense for my business?

Accepted Answer

Four scenarios where Alpha Level will tell you upfront the math doesn't work. First, under fifty support conversations per month — not enough volume to justify the build cost against simply hiring an extra few hours of support time. Second, extremely diverse and unpredictable queries that can't be templated, where the bot would escalate so often it adds friction rather than removing it. Third, a catalogue or documentation that changes daily without versioning, because the knowledge base would drift faster than we can re-train it. Fourth, regulated industries with explicit compliance rules that forbid any AI in customer-facing touchpoints (some healthcare, legal and financial services use cases). A thirty-minute discovery call is enough to tell you which bucket you fall into.

Question 4

When does an Alpha Level chatbot make business sense?

Accepted Answer

Strong fit when four conditions hold together. First, you handle five hundred or more tickets per month with a small team and queue times that are starting to hurt CSAT scores. Second, roughly eighty percent of those tickets fall into twenty or fewer repeating categories — order status, product specs, returns, account access, booking changes — which is the long-tail pattern most service businesses sit on. Third, you sell online and want a twenty-four-seven sales assistant that can answer pre-purchase questions when human staff are offline, recovering revenue from the abandoned-cart window. Fourth, your internal team wastes time digging through documentation, where an internal-facing bot pays for itself in week one by cutting context-switching out of senior-engineer hours.

Question 5

What integrations does an Alpha Level chatbot support?

Accepted Answer

We connect to whatever the bot needs to actually take action, not just retrieve information. Commerce stack: Shopify for catalogue and order status, Stripe for billing-cycle lookups and refund flows, Zendesk for ticket creation and merge. CRM stack: HubSpot or Pipedrive for lead capture, lifecycle-stage updates, and pipeline progression. Booking stack: Cal.com or Calendly with multi-team availability awareness, time-zone normalisation, and automated rescheduling. Internal knowledge: Notion, Confluence, Google Drive, with permissions respected so the bot never leaks documents the user shouldn't see. Team-facing surfaces: Slack and Microsoft Teams with thread-aware context. Each integration is a separate scope item with its own auth and test pass — we don't ship a bot that has stale read-only credentials.

Question 6

How long does it take to deploy an Alpha Level chatbot?

Accepted Answer

Roughly thirteen to eighteen calendar days across four sequential phases. Discovery (days one to three) — we map your top-twenty query categories, audit the knowledge sources, and lock the success metric. Knowledge-base ingestion (days four to eight) — your documentation, FAQs, product catalogue and support transcripts go into a versioned vector store, with chunking and metadata tuned for retrieval accuracy. Integration (days nine to fifteen) — read and write access to live systems (Shopify, Stripe, Zendesk, CRM, calendar) including auth, error handling, and edge-case test coverage. Internal beta plus prompt tuning (days sixteen to eighteen) — your team runs the bot first, surfaces gaps, and we tune before public launch. The SupportHub deployment shipped at eighteen days end-to-end.

Question 7

Is an Alpha Level chatbot just a wrapper around a chatbot SaaS like Intercom or Drift?

Accepted Answer

No. Each Alpha Level deployment is a unique build on your data, with your tone of voice, your integrations, and your conversation flows — not a generic chatbot widget that gets rebranded with your logo and outgrown in six months. You own the architecture (prompts, retrieval logic, integration code), you own the knowledge base (vector store, ingestion pipeline, versioning), and you own the data the bot generates (every conversation, escalation, resolution). No lock-in to a proprietary SaaS, no per-seat pricing that scales punitively as your support volume grows, no opaque "AI features" controlled by a vendor's product roadmap. If you ever want to take the bot in-house, every artefact is portable and the documentation explains exactly how it works.

Layer	Integrations	What the bot can do
Commerce	Shopify, WooCommerce, Stripe, Paysera, Zendesk	Order status, refunds, subscriptions, ticket merge
CRM	HubSpot, Pipedrive, Salesforce	Lead capture, lifecycle update, round-robin assignment
Booking	Cal.com, Calendly, Google Calendar	Multi-team availability, time-zone normalisation, rescheduling
Knowledge	Notion, Confluence, Google Drive, SharePoint	Permission-aware retrieval, never leaks unseen docs
Team	Slack, Microsoft Teams	Thread-aware context, internal Q&A, escalation routing

An employee that never sleeps.

What does an Alpha Level chatbot actually do?

Customer support

Lead qualification

Booking & scheduling

E-commerce concierge

Internal knowledge bot

Live analytics

How is the chatbot trained on my business?

Discovery

Knowledge base

Integrations

Train & ship

What Alpha Level chatbots integrate with.

Which AI model powers Alpha Level’s chatbot?

Does it make sense for you?

It probably won’t pay off if

It’s a slam-dunk if

SupportHub: 80% auto-resolved, €50K saved.

Ready to automate your support?