meta
Vatsal Shah

AI Chatbots & Voice Agents

Production voice and chat agents for Canadian businesses. Built on real conversations, integrated with your stack, monitored after launch.

Talk to Vatsal

A voice or chat agent that actually handles real calls — not a pilot in a sandbox. We build, ship, and monitor production-grade agents that run inside your existing telephony or chat surface. Most engagements ship a working pilot by week four and reach production by week six.

Fit for

  • Canadian businesses with 50+ employees and at least one high-volume customer-facing workflow (calls, chat, intake)
  • Teams that already have transcripts, scripts, or playbooks for the work the agent will absorb
  • Operators willing to test against real conversations, not just synthetic prompts

Not a fit for

  • Companies looking for a generic chatbot widget on a marketing site — SaaS tools do that for less
  • Pre-product startups that haven't talked to customers yet
  • Teams expecting an autonomous agent without any human-in-the-loop oversight

What you walk away with

  • A working voice or chat agent deployed to your production telephony or chat surface
  • Integration with your CRM, helpdesk, or ticketing system (HubSpot, Salesforce, Zendesk, Front, etc.)
  • An eval harness covering 50+ real conversation scenarios pulled from your historical data
  • Monitoring and alerts for production quality — latency, intent accuracy, escalation rate, error patterns
  • Runbook for your team — what to monitor, when to retrain, how to handle the edge cases the agent declines
  • Thirty-day post-launch optimization window built into the engagement

How the engagement runs

  1. 1

    Discovery · Week 1

    We work through sample conversations, the current workflow, and the metrics that define success for your team.

  2. 2

    Architecture · Week 2

    Model selection, integration plan, latency budget, escalation policy. You see the system shape before we write the first agent.

  3. 3

    Pilot build · Weeks 3–4

    Working agent in a sandboxed environment, tested against real conversations from your data. Weekly demo, weekly course-correct.

  4. 4

    Production deployment · Weeks 5–6

    Phased rollout into your live telephony or chat surface with monitoring already wired. Start with a single team or queue, expand from there.

  5. 5

    Optimization · Weeks 7–10

    Eval against real production data, model tuning, handoff of the runbook so your team owns the agent end to end.

By industry

For Canadian Financial Services

Voice agents for member service teams at credit unions, MGAs, and insurance brokerages. Handle account inquiries, claim status checks, and routine policy questions — escalating to a human when the conversation goes off-script. Integrates with your AMS, CRM, and core banking systems.

For Canadian Healthcare

Voice agents for multi-site clinics, dental groups, and physiotherapy networks. Handle appointment booking, prior-auth status, prescription renewals, and patient intake. Tuned for your provincial health framework, integrates with your EMR or practice management system.

For Canadian Customer Operations

Voice and chat agents for support and operations teams across SaaS, e-commerce, and B2B services. Tier-1 deflection, ticket triage, and answer-with-citations from your help center and internal documentation.

Selected work

Recent voice and chat work.

“Vatsal is an excellent full stack developer and highly skilled project manager. He identified our business needs quickly and established a very strong framework. His incredible speed should be noted, this is a developer who doesn't waste time and hit every target date we threw at him.”
Josiah Liesemer
Josiah Liesemer
IT Specialist and Developer, Zucora Home
Read more in our engineering log →

Frequently asked

Can the agent take calls in production from day one?
No. Week 1–2 is discovery and architecture; weeks 3–4 is a sandboxed pilot tested against your real conversations; weeks 5–6 is a phased production rollout into a single queue or team first. Most teams want this gradual ramp rather than a cold cutover, and the eval work we do in weeks 3–4 is what makes production safe.
What does it cost?
Pricing is custom per engagement and depends on call volume, integration complexity, language requirements, and the scope of the production rollout. We share pricing on the first discovery call once we understand what you actually need.
Do you build voice agents in Canadian English versus US English?
Yes. We tune voice models for Canadian intonation, region-specific terms (RRSP, OHIP, Service Canada, etc.), and bilingual English/French when the deployment requires it for Quebec or federal-scope work.
Can you integrate with our existing telephony or chat platform?
Yes. We deploy on Twilio, Telnyx, RingCentral, Zoom Phone, Dialpad, Aircall, and most major Canadian telephony providers. For chat, we integrate with Intercom, Zendesk, Front, HubSpot Service Hub, and direct in-app surfaces. We don't require you to switch platforms.
What happens when the agent gets something wrong in production?
Every conversation is instrumented. Errors trigger alerts; misclassifications feed back into the eval set; your team gets a runbook for handling escalations and a hard handoff to humans when the agent is uncertain. We treat production errors as data, not failures — and the engagement includes time to act on them.
Do we own the code and the conversation data?
Yes. Everything we build ships into your repository from commit one. Conversation data stays in your infrastructure unless you explicitly opt into a model-tuning loop with us. No vendor lock-in, no hidden tooling.
Where is the agent deployed?
Default is your existing cloud — AWS, Azure, or GCP. For Canadian data residency requirements, we deploy on AWS ca-central-1, Azure Canada Central, or Bedrock Canada. Inference can run through Anthropic, OpenAI, Google, or open-weights models depending on cost, latency, and residency constraints.
Who does the work?
Two to three engineers from our Toronto-based team, led by Vatsal. The people who scope the engagement are the people who write the code. The full team is named on our team page — you can see and talk to them before we start.