AI chatbot development services for customer service, support, and ecommerce. We ship RAG-grounded chatbots on Claude Sonnet 4.6, Haiku 4.5, and GPT-4o-mini — to your web widget, WhatsApp, voice, or Slack/Teams. Eval-gated, guardrailed, token-optimized. First chatbot live in 30 days, behind a feature flag.
Every customer service chatbot, ecommerce chatbot, and internal Slack chatbot below has been shipped from this exact playbook. Each one comes with an eval suite, audit logging, confidence gating, and a per-turn cost target — not a polished demo.
The crown-jewel use case. RAG-grounded chatbot over your help center + ticket history that handles password resets, order status, refund eligibility, and policy lookups. Confidence gate at 0.7; anything below escalates to a human with the AI's draft attached. We've shipped these to 30–45% tier-1 deflection on Zendesk and Intercom queues.
Conversational AI grounded in your Shopify / WooCommerce catalog. Product recommendations from natural-language criteria, order status ("where's my package?"), return initiation, size and fit Q+A. Function calls into your OMS + 3PL APIs so the bot can act, not just describe.
Chatbots inside Slack or Microsoft Teams that retrieve from Notion, Confluence, Drive, or your code repo. Onboarding Q+A, policy lookups, on-call runbooks. Built on Sonnet 4.6 + tool use for the 6+ step internal workflows where a one-shot answer isn't enough.
Where the user actually is. WhatsApp Business chatbots (Meta Cloud API) for international support and lead capture. Voice chatbots on Twilio / Vapi over the OpenAI Realtime API or Deepgram + Sonnet 4.6 for sub-second voice. Latency profiles differ — we'll tell you which channel makes sense before we build.
When the answer lives in a 10,000-document Drive folder, a contract library, or a research archive. Retrieval-augmented chatbot with pgvector or Pinecone, top-k 5 retrieval re-ranked by bge-reranker-v2, eval-tested on your real question set before launch. We measure groundedness, not just BLEU.
Website chatbot that runs a structured intake (BANT, MEDDIC, or your custom rubric), drops qualified leads into HubSpot / Salesforce with a transcript, and books a meeting via Cal.com or Chili Piper. Less marketing-deck, more pipeline — built for revenue teams that actually want call shows, not vanity completion rates.
Six stages every production chatbot turn moves through — from user message to logged outcome. Skip any one and you get the demo competitors ship instead of a chatbot that deflects tier-1 traffic. Each stage carries its own latency budget, model pick, and failure mode. Hover any stage for the operator detail.
Latencies and token counts are typical production traces from shipped chatbots. Your eval set sets the real budgets.
A real customer-service replay — left side, a chat surface streaming. Right side, the same six stages from the anatomy lighting up as each turn fires. This is what 'shipped' looks like, not what the demo deck shows.
Illustrative per-decision economics — typical engagement-band figures, not literal client numbers.
Most chatbot vendors will quietly say yes to any channel. We won't. Pick a channel to see the deployment surface, latency profile, the actual stack, and — the part competitors hide — where it fails. Channel mix is decided in the audit, not in the sales call: web widget, WhatsApp, voice, Slack + Teams, Discord, Telegram, Instagram + Messenger, and SMS + iMessage.
Embedded floating widget on your marketing site, in-product help center, or post-login dashboard. We ship a Preact-based widget that loads under 35KB gzip and streams responses token-by-token. Same widget surface across desktop and mobile.
<800ms first token · streamed · p50 1.4s end-to-end
Web widget bounces hit ~70% of sessions in B2C. If your customers aren't already on your website (e.g. they're in WhatsApp or your mobile app), the web widget is the wrong channel. Pick the channel your buyers already live in.
The chatbot reply is half the story. When a turn fires a tool, an automation graph kicks in — classify, lookup, decide, act. Here's a real WhatsApp refund flow playing out node-by-node, with a branching decision that escalates to a human when policy says it should.
Built on n8n, LangGraph, or custom — depending on your stack. Cost chips are illustrative per-decision economics.
The naming has drifted — every vendor calls everything an “AI agent” now. The honest distinction: chatbots are scoped and short-turn; agents are multi-step and long-horizon. Most teams asking for an agent need a chatbot first. Per-dimension honest comparison below.
Generalizations from shipped client work. Specifics vary per workload; we benchmark on your eval before recommending.
A production chatbot is not one model — it's a routed pipeline. Cheap classify in front, grounded generate in the middle, cheap log + eval at the back. Here's the default chatbot stack we ship; we'll re-pick per workflow if your eval data demands it.
Five tactics stacked, in order of impact for chatbots. Most chatbot pilots see effective per-turn cost drop to 6–10% of the naive baseline at the same eval-suite quality. This optimization pass is included in every chatbot pilot, post-cutover.
Four stages, milestone-billed, with a walk-away point at the retrieval baseline. Most chatbot failures happen because the team skipped the eval set or skipped retrieval tuning — both are in week 1 and week 2 here, not bolted on at the end.
We harvest 50–200 real questions from your ticket archive (or run a structured user interview if you're greenfield) and build the eval set the chatbot will be measured against. Scope locked: channels, knowledge sources, tool surface, escalation rules.
Ingest your docs into pgvector or Pinecone, run chunking experiments (semantic vs fixed-size, header-aware vs not), tune top-k and re-ranker, and score retrieval against the eval set. Most chatbot quality issues are retrieval issues, fixed here.
Wire the full anatomy: classifier → retrieval → tool use → generate → guardrail → log. Behind a feature flag, in your repo (or ours, your call). PII scrub, refusal rules, confidence gate, audit-log every turn. Channel-specific UI shipped in parallel.
Shadow mode against your existing channel for 2 weeks. Score on the eval set, score on real traffic, score on cost. Roll out at 10% → 50% → 100% if numbers hold. Run the token-optimization pass — most chatbots see 60–85% cost reduction at the same quality.
The same chatbot turn — classify → retrieve → tool → generate → guardrail → log — across Sonnet 4.6, Haiku 4.5, and GPT-4o-mini. Pick a model on the left; the model= line swaps and the per-turn cost stat updates. This is how we choose: run your eval, then look at the bill.
from anthropic import Anthropic
client = Anthropic()
def chat_turn(user_msg: str, history: list[dict]) -> dict:
# 1. Intent classify with Haiku 4.5 (~$0.0001 / turn)
intent = classify_intent(user_msg)
# 2. RAG retrieve from pgvector (top-k 5, re-ranked)
docs = retrieve(query=user_msg, k=5, rerank=True)
# 3. Tool-aware generate — switch the reply model:
response = client.messages.create(
model="claude-sonnet-4.6",
max_tokens=600,
system=SYSTEM_PROMPT + format_docs(docs),
tools=[zendesk_create_ticket, order_status_lookup],
messages=history + [{"role": "user", "content": user_msg}],
)
# 4. Guardrail: confidence + PII + policy
verdict = guardrail.check(response, intent=intent)
if verdict.action == "escalate":
return handoff_to_human(response, verdict)
# 5. Log to Langfuse for nightly eval
langfuse.log(trace_id, response, verdict, tokens=response.usage)
return response
Same pricing as our other engagements. Most clients begin with the audit to scope channels + scope retrieval, run a 4–6 week pilot on the highest-ROI channel, then move to monthly for the next 2–3.
Find the chatbot workflow worth shipping before you commit a budget.
One chatbot shipped end-to-end on your highest-ROI channel — with eval data, not a demo.
Embedded squad shipping the next chatbot channel + tuning the live one.
Three anonymized chatbot capability patterns drawn from real engagements. Named references shared under NDA once we know what you're building.
Inbound Zendesk queue averaging 6-hour first-response time; tier-1 reps spending 60% of time on password resets, billing questions, and feature-availability lookups.
Web-widget chatbot with RAG over the help center + ticket history. Haiku 4.5 classifier, Sonnet 4.6 generate, function calls into Zendesk for ticket creation. Confidence gate at 0.7; sub-threshold escalates with a drafted reply attached for the agent. Voice-channel sibling: see the published openai realtime voice agent case study for the same deflection pattern on inbound voice at $0.10/call.
International D2C brand getting product Q+A and 'where's my order' inquiries via WhatsApp from 14 countries — manual reply queue 18 hours long at peak.
WhatsApp Cloud API chatbot grounded in Shopify catalog + 3PL tracking data. Multilingual via Sonnet 4.6 native multilingual; function calls into Shopify Admin API + Aftership for live order status. Refund initiation gated to human review.
On-call rotation drowning in repeated questions about runbook locations, alert ownership, and dashboard URLs — same 12 questions answered nightly by senior engineers.
Slack chatbot with RAG over Notion runbooks + the team's PagerDuty service catalog. Tool calls into PagerDuty for on-call lookup and Grafana for dashboard linking. Escalates to senior on-call if confidence drops or alert is sev-1.
Three engagement tiers. A 1–2 week chatbot audit is $3,000 — discovery, channel recommendation, RAG architecture, model pick, eval-set design, and a 90-day roadmap. A pilot is $10,000–$25,000 fixed price, 4–6 weeks — one chatbot shipped end-to-end on your chosen channel with eval, monitoring, and a token-optimization pass. A continuous chatbot team is from $5,000 per month — embedded engineer + PM + ops analyst, shipping new channels and tuning the live one. Run-cost (model calls + vector DB + monitoring) typically lands at $200–$2,000 per chatbot per month depending on volume and channel mix.
A chatbot is scoped, single-turn or short-turn, and grounded — user asks, system retrieves, maybe makes one tool call, replies. Latency budget is sub-2s. A chatbot answers customer service questions or qualifies a lead. An <a href="/services/ai-agent-development/">AI agent</a> is multi-step and long-horizon — plans, calls multiple tools, observes results, re-plans, eventually completes a task. Latency budget is 10s–10min. An agent files a refund across three systems, researches a prospect, or runs a deployment. Most teams asking for an "AI agent" actually need a chatbot first; we'll tell you which during the audit. Cost per interaction differs by ~50×.
Both are production-ready for customer service chatbots. <a href="/services/claude-development/">Claude Sonnet 4.6</a> wins on long-context RAG, multilingual support without separate language models, and tool-use stability when the chatbot has 6+ functions to choose from — these are the dimensions that matter most for support. <a href="/services/openai-development/">GPT-4o-mini</a> wins as the cheap classifier in front (intent + routing) and as the voice-channel reply model via the OpenAI Realtime API. Our default chatbot stack is Haiku 4.5 or GPT-4o-mini for intent classify, Sonnet 4.6 for the grounded reply — we're model-agnostic and we'll show you the eval-set numbers before recommending.
Most pilots ship in 4–6 weeks after a 1–2 week audit. Realistic distribution: simple chatbots (single channel, single-language, narrow scope like password reset + billing FAQ) in 3–4 weeks. Mid-complexity (RAG over a 1,000-doc knowledge base, 3–5 tool calls, web + WhatsApp) in 4–6 weeks. Complex (regulated industry with PII handling, voice channel, multilingual across 5+ languages, 10+ tools) in 8–10 weeks. The audit phase tells us which bucket you're in before any pilot contract. We don't quote a 30-day chatbot for work that takes 90 days.
A RAG (retrieval-augmented generation) chatbot grounds its replies in your actual data instead of relying on the model's general knowledge. The flow: user asks → system retrieves the top 3–5 most relevant chunks from your knowledge base (pgvector / Pinecone) → those chunks plus the user message go to the reply model (Sonnet 4.6) → the model composes an answer cited to those chunks. You almost certainly need one. The only chatbots that don't are pure-personality bots ("chat with a brand mascot") or chatbots over data the model was trained on (general programming Q+A). Every customer service, support, ecommerce, and internal-knowledge chatbot is a RAG chatbot. Most chatbot quality issues are retrieval issues, not generation issues — which is why we tune retrieval before tuning prompts.
Yes — multi-channel deployment is standard. WhatsApp via Meta's Cloud API (business verification + template approval, typically 1–3 business days). Voice via Twilio Voice or Vapi over the OpenAI Realtime API (sub-second first-token latency) or a Deepgram + Sonnet 4.6 pipeline. Slack via the Bolt SDK with event subscriptions + slash commands. Microsoft Teams via the Bot Framework SDK with admin scope approval. Same RAG corpus and tool surface across channels; the UI differs (streaming for web, message-edit-streaming for Slack, audio streams for voice). We'll recommend which channels matter during the audit — most teams over-deploy and end up with three channels they don't measure.
Honest answer — there isn't a single best, but the question to ask any AI chatbot development company is: do you ship eval suites, channel-specific honesty notes, and token-cost projections, or do you ship demos? Listicle sites rank chatbot vendors by review count and case-study polish, neither of which predicts whether your chatbot will deflect tier-1 traffic in production. We score ourselves on operator detail — we use Claude Code daily, we run model-agnostic across Claude + OpenAI, and we publish a $3K audit-to-roadmap engagement before any chatbot build kicks off. If your shortlist includes vendors that can't show you their eval methodology in 30 minutes, that's the disqualifying signal. <a href="/services/ai-consulting/">AI consulting</a> + audit is a $3K way to scope what's worth building before you sign a six-figure chatbot agency contract.
Four layers, stacked. (1) RAG grounding — the reply model sees retrieved chunks from your real data, and the system prompt instructs it to answer only from those chunks or say "I don't know." (2) Confidence gating — every reply gets a self-rated confidence score; sub-threshold replies escalate to a human with the AI's draft attached, never auto-send. (3) Guardrails layer — separate from generation, a policy-check pass runs PII scrubbing, refusal rules ("never quote a price", "never confirm an account number"), and competitor-mention blocking. Fail-closed by default. (4) Nightly eval — Langfuse / Helicone logs every turn; we run an eval suite against held-out questions nightly and alert on regression. The combination, not any single piece, is what makes a chatbot production-safe. We include this stack in every pilot.
Book a free AI chatbot audit. We'll review your existing chatbot or support queue, recommend channels (web · WhatsApp · voice · Slack/Teams), pick models per stage (Sonnet 4.6 / Haiku 4.5 / GPT-4o-mini), project token cost vs your current spend, and give you a 90-day chatbot roadmap. No deck, no obligation to build.
Building a chatbot often connects to a sibling AI service. These pages go deeper on the adjacent decisions.
Anthropic Claude integration — the default reply model for our chatbots.
Read more 02GPT-4o-mini as classifier · Realtime API for sub-second voice.
Read more 03When you need multi-step planning, not single-turn replies.
Read more 04Plug your chatbot into Salesforce, Zendesk, HubSpot, NetSuite.
Read more 05Strategy and roadmap before the chatbot build.
Read more 06When the workflow is bigger than a single chatbot turn.
Read more 07HIPAA-grade patient-intake chatbots — BAA, PHI scrub, audit log.
Read more 08Shop-floor chatbots — diagnostic Q&A on MES + historian with citation logging, no PLC writeback.
Read more 09Privilege-aware chatbots — matter intake + KM Q&A on rings 2/4 with citation logging.
Read more 10Travel chatbots across pre-trip nudges, in-trip exception handling, and post-trip review handlers.
Read more 11FERPA-aware chatbots — Socratic tutors, advisor Q&A, and L&D micro-coaches with integrity-zone gating.
Read more 12HR-services chatbots — policy Q&A, benefits intake, ER triage with hard-escalation rules and ADA accessible alternatives.
Read more 13Insurance AI audit + roadmap — claim-lifecycle state machine, underwriting capacity sankey, and fraud-network mapping before any core-system integration.
Read more 14Fintech AI audit + roadmap — risk-score gauges, payment-rails routing, KYC tier-ladder, model-risk-management before any production inference.
Read more