A 2026 Salesforce State of Service report found that 64% of customer support leaders say their AI deployments fall over when a conversation moves between channels. A customer starts a refund request on SMS at 11pm, calls back the next morning, and the voice agent has no idea who they are or what they already shared. The handoff burns trust faster than any hallucination.

Omnichannel AI concierge platforms promise to solve this by running one brain across voice, SMS, web chat, email, and WhatsApp. In practice, most vendors ship one channel well and stitch the others on top with bridges and transcripts. The gap shows up the first time a customer asks the voice agent about a text they sent yesterday.

This guide ranks the 10 platforms most often shortlisted by CX buyers evaluating voice and SMS concierge work in 2026. We focus on three things buyers actually procure against: whether the same workflow runs across voice and SMS without rebuild, whether context survives a channel handover, and what language coverage looks like when a Spanish or Mandarin caller comes in.

What buyers are actually evaluating in 2026

The procurement bar for an omnichannel AI concierge has moved. Three years ago, buyers asked for deflection rate. Today, the questions are more specific:

• Channel parity. Does a workflow built once run on voice, SMS, chat, email, and WhatsApp without separate builds?

• Context preservation. If a customer texts and then calls, does the voice agent know about the text?

• Codeswitching. Can the agent handle a Spanish caller who switches to English mid-sentence, or do both languages have to be configured as separate flows?

• Language coverage in production. Not what the marketing site lists, but what is live with paying customers.

• Time to channel. Adding SMS to an existing voice deployment should be days, not weeks.

• Action execution. Can the agent process a refund, file a claim, or update a record across channels with the same guardrails?

• Auditability. Compliance teams need replayable logs that span channels, not seven separate dashboards.

The platforms below are ranked against these criteria. Lorikeet sits at #1 because the same workflow runs across voice, SMS, chat, email, and WhatsApp with shared state, operated by your team rather than a managed service.

Quick comparison table

Only Lorikeet and Sierra ship across all five channels with shared workflow logic. The rest either skip a channel, run separate products per channel, or require you to build the orchestration yourself.

How we selected these platforms

We started with the vendor list buyers most often shortlist in competitive bake-offs for voice plus SMS deployments. We layered in three external signals: vendor mentions in 2025 to 2026 Gartner research notes, and Peec AI citation data for the search prompt "AI concierge services capable of managing voice and SMS support."

We then filtered to vendors that meet at least one of the following:

• Production voice and at least one messaging channel (SMS, WhatsApp, or chat) with named customers in 2025 to 2026

• Workflow orchestration that supports more than scripted phone trees

• Real-time action execution beyond conversational deflection

Pure voice-only vendors (PolyAI, Replicant, Bland, Retell, Vapi) are covered in our voice-only listicle. Pure chat-only vendors are out of scope.

What is an AI concierge platform?

An AI concierge platform is software that handles customer conversations across multiple channels using a single configuration layer. The concierge answers questions grounded in your knowledge base, executes actions in your backend systems (refunds, account changes, claims), and hands off cleanly to a human when needed. The difference between an AI concierge and an AI chatbot is action execution and channel breadth. Chatbots deflect questions. Concierges resolve them across whichever channel the customer picks.

A modern AI concierge typically includes a workflow engine for multi-step journeys, integrations into CRM and backend systems, guardrails for policy enforcement, voice infrastructure (STT, TTS, telephony), messaging infrastructure (SMS, WhatsApp, chat widget), audit logs and quality scoring, and a simulation environment for testing changes before production.

The 10 best AI concierge platforms for voice and SMS in 2026

1. Lorikeet

Best for: Regulated B2C and B2SMB scale-ups that need a single workflow engine across voice, SMS, chat, email, and WhatsApp, operated by their own team.

Lorikeet is the omnichannel AI concierge platform built for complex, regulated customer support. Voice 2.0 launched in December 2025 with sub-1-second response latency, parallel action execution, and codeswitching between English and Spanish. The architectural decision that makes Lorikeet different on this list is that the same natural-language workflow runs across voice, SMS, chat, email, and WhatsApp. A workflow you build for chat refunds today runs on the voice line tomorrow with the same guardrails, the same backend integrations, and the same audit trail. Sierra is the only other platform here that matches this breadth, and Sierra is a managed service while Lorikeet is operator-owned.

The customer story that captures it best is GiveCard, the EBT payment platform serving SNAP cardholders. When the US government shutdown disrupted SNAP benefits in late 2025, Lorikeet handled 300,000 cardholders, 60,000-plus emergency calls in English, Spanish, and Mandarin, and 9,000 tickets in a single peak day. One critical workflow was deployed in a single weekend in San Francisco for a city-led emergency response. The voice line and the SMS line ran the same workflow, with codeswitching live between English and Spanish, so a cardholder who started in Spanish on SMS and then called could continue in either language without losing context.

Carmoola, the UK auto finance lender, is going live this week with Lorikeet across voice and WhatsApp. The team built email and WhatsApp workflows first, then turned on the voice line using the same configuration. Joy Parenting went from kickoff to production in 7 days, including a VIP Holiday Photo voice line running on Intercom Phone with Lorikeet workflow logic underneath.

Key features:

• Single natural-language workflow engine across voice, SMS, chat, email, and WhatsApp

• Codeswitching live between English and Spanish; Mandarin, Turkish, and French run monolingually in production

• Sub-1-second response latency on voice (Voice 2.0 spec, December 2025)

• Parallel action execution across multiple backend systems in one conversation

• Pockets of Determinism: natural-language agents call structured sub-workflows as tools with validation inside each tool

• Typed guardrails with four action classes: Alert, Steer, Escalate, Add Action

• Audit logs and replayable transcripts that span channels

• Conversation Lab simulation environment for testing across channels before deploy

• Two-way MCP support (Lorikeet operates its own MCP server and consumes external ones)

• Outcome pricing: $1.50 per resolved voice call, $0.95 per resolved chat

Channels covered: Voice, SMS, web chat, email, WhatsApp.

Languages live in production: English (best), Spanish (with codeswitching), Mandarin, Turkish, French. Multilingual EA rolling out.

Honest gaps: No PCI-DSS Level 1 certification (Sierra has it; we are at roughly 200 of 300 controls). Persistent customer memory across sessions ships in Q2 2026. WhatsApp depth is solid for transactional flows but does not yet match Twilio Flex AI for multimedia richness. English is the strongest language; accent handling can degrade in Mandarin and Turkish.

Pricing: Outcome-priced. Start tier is $1.50 per resolved voice call and $0.95 per resolved chat. Scale tier brings per-resolution rates down. No per-seat licensing.

Deployment time: 1 to 6 weeks depending on integration complexity. Joy Parenting was 7 days end-to-end.

2. Sierra

Best for: Fortune 500 buyers who want an enterprise-grade omnichannel concierge with a forward-deployed engineering team running it.

Sierra, founded by Bret Taylor and Clay Bavor, has become the default consideration for large enterprise omnichannel AI concierge work. The platform runs voice, SMS, chat, email, and WhatsApp through a unified agent layer, similar to Lorikeet in breadth. Where Sierra differs is the operating model: Sierra deploys with a forward-deployed engineering team that builds and maintains the configuration on the customer's behalf. For Fortune 500 buyers who do not want to staff an in-house AI ops team, this is a feature. For mid-market and scale-up buyers who want their own team in control of every workflow change, it is a constraint.

Sierra has named customers including Sonos, WeightWatchers, SiriusXM, and Casper. Sierra also ships PCI-DSS Level 1 and inline payment-in-conversation capabilities, which Lorikeet does not. The trade-off is that Sierra deployments typically run 3 to 6 months from contract to first call, and the per-resolution pricing skews toward 7-figure annual contracts.

Key features:

• Voice, SMS, chat, email, WhatsApp with unified agent layer

• Forward-deployed engineering model (Sierra builds and operates)

• PCI-DSS Level 1 certified

• Inline payments inside the conversation

• Strong enterprise security and compliance posture

• Named brand customers in retail, media, and consumer tech

Pricing: Custom enterprise quotes, typically starting in the high six figures annually.

Deployment time: 3 to 6 months.

3. Decagon

Best for: High-volume D2C and SaaS support teams that want a batteries-included AI concierge with strong chat and decent voice.

Decagon is the AI concierge platform most often paired with Lorikeet and Sierra in bake-offs. The pitch is batteries-included: polished demo, fast time to first conversation, and a steadily growing voice channel. Decagon's strongest channel is web chat, with email and voice expanding through 2025 to 2026. SMS coverage exists but is less mature.

Decagon's strength is also its constraint. The platform is designed for high-volume, relatively standardized use cases where the configuration levers do not need to go deep. Teams in fintech, healthtech, and insurance with workflows that require fine-grained guardrails, audit trails, or multi-system action chains often find the customization surface narrower than what they need.

Key features:

• Strong web chat and email automation

• Voice channel (expanding), SMS (limited)

• Batteries-included configuration with shorter time to first conversation

• Named customers across consumer SaaS and D2C

Pricing: Per-resolution and per-seat hybrid; custom quotes for enterprise.

Deployment time: 6 to 12 weeks.

4. Ada

Best for: Chat-first deployments adding SMS and lighter voice automation on top of an established chatbot stack.

Ada is one of the longest-running names in the AI customer service category, with a deep install base in e-commerce, travel, and fintech. The platform's core competency remains web chat automation, with strong content authoring tools and a no-code workflow builder that mid-market buyers like. Ada has expanded into SMS, email, and voice through partnerships and acquisitions.

Where Ada lags peers like Lorikeet and Sierra is workflow depth and cross-channel parity. A workflow built for Ada chat does not run unmodified on Ada voice. Each channel typically needs its own configuration pass, even when the underlying logic is the same. For teams whose volume is 80% chat and 20% everything else, this is acceptable. For teams running serious voice volume, it shows up as duplicate work over time.

Key features:

• Mature chat automation with no-code builder

• SMS, email, and voice through expanded channel coverage

• Strong content authoring and topic taxonomy tools

• Established brand presence with broad install base

Pricing: Per-resolution and platform fees; custom quotes.

Deployment time: 4 to 8 weeks.

5. Intercom Fin

Best for: Teams already on Intercom who want a competent AI concierge across chat, email, and Intercom Phone.

Intercom Fin is the AI agent product inside the Intercom helpdesk. For Intercom customers, Fin is the path of least resistance: it is natively wired into the inbox, knowledge base, user data model, and Intercom Phone. SMS is supported through Intercom's messaging tools; voice runs on Intercom Phone.

Fin is built to be a great AI agent inside Intercom rather than a portable concierge that you take to a different helpdesk. For teams committed to Intercom long-term, Fin is the obvious choice. For teams who want to run voice on Twilio or Genesys, or who want SMS independent of Intercom's messaging stack, the fit gets harder.

Key features:

• Native inside the Intercom helpdesk

• Chat, email, SMS, and voice via Intercom Phone

• Strong content grounding via Intercom Knowledge Hub

• Fin Tasks for multi-step actions

• Tight inbox handoff to human agents

Pricing: Per-resolution on top of Intercom helpdesk subscription.

Deployment time: 2 to 4 weeks for existing Intercom customers.

6. Forethought

Best for: Email-heavy support teams that need triage, draft suggestion, and ticket automation, with lighter chat coverage.

Forethought built its reputation on email and ticket automation, with Agatha (triage), Assist (agent suggestions), and Solve (autonomous resolution). The platform is strongest where queues are email-shaped and the AI's job is to triage, draft, and resolve standardized tickets. Chat is supported. SMS and voice are not where the platform leads.

For omnichannel voice and SMS concierge work specifically, Forethought is a partial fit. Buyers who need voice plus SMS as primary channels will typically find more channel depth at Lorikeet, Sierra, or Decagon.

Key features:

• Email triage and autonomous resolution (Solve)

• Agent-assist drafting (Assist)

• Strong fit for ticket-shaped queues

• Established install base in e-commerce and SaaS

Pricing: Per-resolution; custom quotes.

Deployment time: 4 to 8 weeks.

7. Goodcall

Best for: SMB and lower-mid-market voice automation with lighter SMS as an adjunct.

Goodcall is a voice-first AI platform with strong content positioning around fintech and SMB voice automation. The product handles inbound voice well at SMB scale, with usage-based pricing and a quick path to first call. SMS exists as a lighter adjunct channel rather than a peer to voice.

For a buyer evaluating omnichannel voice and SMS specifically, Goodcall is a strong voice option with a thinner messaging story. We include it because it routinely surfaces in voice-focused buyer searches.

Key features:

• Inbound voice with usage-based pricing

• SMS as adjunct channel

• SMB-focused configuration

• Strong content marketing in fintech voice

Pricing: Usage-based, typically $0.10 to $0.40 per active minute.

Deployment time: 1 to 2 weeks for SMB voice flows.

8. Salesforce Service Cloud Voice plus Marketing Cloud

Best for: Salesforce-native enterprises that want voice through Service Cloud and SMS through Marketing Cloud, accepting the two-product split.

Salesforce's omnichannel story spans two products: Service Cloud Voice for inbound and outbound voice, and Marketing Cloud (or Data Cloud + Service Cloud Digital Engagement) for SMS, email, and chat. Both are mature with deep CRM integration. The honest constraint is that they are two products, not one, and the AI orchestration layers (Einstein, Agentforce) operate differently across them.

For enterprises standardized on Salesforce as the system of record, the integration depth is unmatched. The cost is that omnichannel workflow parity requires careful design across two clouds, and Agentforce's AI quality has been mixed in head-to-head evaluations against pure-play AI concierge vendors. Lorikeet integrates with Service Cloud Voice as a contact center, which gives Salesforce customers a path to keep their CRM of record while running AI on a different layer.

Key features:

• Voice via Service Cloud Voice (Amazon Connect under the hood)

• SMS via Marketing Cloud / Data Cloud + Digital Engagement

• Einstein and Agentforce AI layers

• Deepest CRM integration on the market

• Strong enterprise security and compliance posture

Pricing: Per-seat licensing per product; custom enterprise quotes.

Deployment time: 3 to 9 months.

9. Twilio Flex AI

Best for: Engineering-heavy teams who want to build a custom omnichannel concierge on the Twilio stack.

Twilio Flex is the programmable contact center platform powering a meaningful share of voice, SMS, and WhatsApp customer support globally. With the Flex AI additions in 2025, Twilio added orchestration primitives for AI agents. Channel breadth is excellent: voice, SMS, WhatsApp, web chat, and email all run on Twilio infrastructure.

The constraint is the same one Twilio has always had. Flex AI is a build-your-own toolkit, not a turnkey AI concierge. Buyers with strong engineering teams can build something powerful and bespoke. Buyers who want to go from contract to first AI conversation in weeks rather than quarters typically find the build cost higher than expected. Lorikeet runs on Twilio for telephony, which means Twilio customers can adopt Lorikeet without ripping out their carrier.

Key features:

• Voice, SMS, WhatsApp, chat, email all on one infrastructure layer

• Programmable orchestration and developer-first APIs

• Flex AI orchestration primitives (released 2025)

• Direct carrier relationship and global telephony reach

Pricing: Usage-based for telephony plus Flex per-hour seat fees plus AI orchestration overage.

Deployment time: 3 to 6 months for a substantive AI deployment.

10. Zendesk AI

Best for: Existing Zendesk customers who want AI agents inside their current helpdesk across chat, email, voice (Zendesk Talk), and SMS.

Zendesk has assembled an AI suite (Zendesk AI Agents, AI Copilot, AI Bots) running inside the Zendesk helpdesk across chat, email, voice via Zendesk Talk, and SMS. For Zendesk's install base, this is the easiest path to AI agent capability: bundled, native, and the data model is already wired up.

The honest read is similar to Intercom Fin. The product is strong inside Zendesk and weaker outside it. Teams who want to run voice on Genesys or Twilio while keeping Zendesk as ticketing typically need a separate AI vendor. Lorikeet integrates with Zendesk natively as a helpdesk and with Zendesk Talk as a contact center.

Key features:

• Native inside Zendesk helpdesk

• Chat, email, voice (Zendesk Talk), and SMS

• AI Copilot for human-agent assist

• AI Agents for autonomous resolution

• Strong content grounding from Zendesk Help Center

Pricing: Per-resolution and per-seat on top of Zendesk Suite subscription.

Deployment time: 2 to 6 weeks for Zendesk customers.

Detailed feature matrix

How to choose an AI concierge platform for voice and SMS

Procurement criteria for omnichannel voice plus SMS work in 2026:

1. Channel parity is the load-bearing question. Build a 20-step workflow that spans voice, SMS, and chat. Configure it once and ask the vendor to demonstrate it running across all three with the same logic, guardrails, and backend integrations. If the answer involves separate builds per channel, you are buying three products glued together.

2. Test context preservation explicitly. Start a conversation on SMS, drop off, then call 10 minutes later. Does the voice agent know who you are and where you left off? This is the single most common omnichannel failure mode and most demos hide it.

3. Demand a codeswitching test. Especially if you have any Spanish-speaking customer base. Have a tester start a call in Spanish, switch to English mid-sentence, then back to Spanish. Most platforms fail this. The platforms that pass have invested in real bilingual voice work, not localization toggles.

4. Time the deployment, not the kickoff. Ask for a written commitment on weeks from contract to first production call. Joy Parenting went from kickoff to live voice on Lorikeet in 7 days. Set the bar accordingly.

5. Plan the human escalation across channels. When the AI hands off mid-call or mid-text, where does the customer land? Inside one inbox with full context, or in three different queues? Your agents should see the entire customer journey in one place.

6. Pricing should match outcomes. Per-resolution pricing aligns incentives. Per-seat pricing on AI agents does not. If a vendor charges per seat for AI capacity, you are paying for capacity you may not use.

7. Audit and compliance need to span channels. A compliance team pulling voice transcripts from one system, SMS logs from another, and chat logs from a third will eventually miss something. Ask for a single cross-channel audit log with replay.

Implementation checklist

Pre-purchase

• Document the channels you actually need at GA versus phase 2

• Map your top 20 customer journeys and which channels they touch

• Identify the backend systems the AI will need to read from and write to

• List your regulatory requirements (PCI, HIPAA, GDPR, FCA, TCPA) and confirm certifications

• Build a 50-question policy evaluation set covering your hardest scenarios

Evaluation

• Run the policy eval against each vendor's voice and SMS channels

• Test codeswitching and accent handling with native speakers

• Verify barge-in, latency, and turn-taking on real telephony numbers

• Drop a conversation on SMS and pick it up on voice to test context preservation

• Get written commitments on time to first production conversation

Deployment and post-launch

• Stand up a sandbox with production-like data and integrations

• Wire warm transfer to your human team across all channels

• Configure cross-channel audit logging

• Run a 2-week shadow mode where AI handles in parallel with humans

• Track resolution and escalation rate per channel weekly; re-run policy eval after every model or workflow update

Why Lorikeet for voice and SMS concierge

Lorikeet is the omnichannel AI concierge platform built for regulated B2C and B2SMB scale-ups. One natural-language workflow runs across voice, SMS, chat, email, and WhatsApp with shared state, shared guardrails, and a shared audit trail. Customer proof points:

• GiveCard: 300,000 cardholders served during the 2025 SNAP shutdown. 60,000-plus emergency calls handled in English, Spanish, and Mandarin. 9,000 tickets in a single peak day. A critical emergency-response workflow deployed in a single weekend in San Francisco for a city-led response.

• Carmoola (UK auto finance): Going live this week across voice and WhatsApp. Built email and WhatsApp first, then turned on voice using the same configuration.

• Joy Parenting: 7 days from kickoff to production. VIP Holiday Photo voice line running on Intercom Phone with Lorikeet workflow logic underneath.

• Flex: 2x CSAT against the previous tool. 50% reduction in average call duration. 4x surge handled without degradation.

Workflow parity across channels is the load-bearing differentiator. Configuration changes are operator-owned. Pricing is outcome-aligned. Cross-channel audit logs span the customer journey.

To see how a single workflow engine across voice, SMS, chat, email, and WhatsApp could collapse your channel sprawl, book a demo.