Every competitor claim rechecked against live primary sources · September 25, 2026

Any platform can chat or talk.
Only Delta Agents also remembers each contact across chat and phone, recovers quiet leads, and builds your tools with AI.

We compared Delta Agents against nine popular chatbot and voice-agent platforms on the things that actually close deals: memory, cross-channel context, lead recovery, and tools. Every fact about a competitor is footnoted to a source, primary wherever one exists. Competitors shipped a lot since our July edition, so we rechecked every claim and changed every mark that moved, including the ones that no longer favor us.

9 platforms compared 12 capabilities 89 sources Data as of 2026-09-25
Start for $7: full access See the evidence ↓
0/9

No competitor documents all three: memory, lead recovery and AI-built tools.

The only two with any form of lead recovery, GoHighLevel59 and CloseBot10, have no AI tool builder56,62. Every platform with an AI tool helper leaves recovery to campaigns, triggers or rules you set up yourself67,69,73,77,82,85,57. Delta Agents ships all three in one system, sharing one memory per contact.

0/9

No competitor re-engages quiet leads on its own with what they said on the phone.

The closest are GoHighLevel’s Auto Follow-Up59 and CloseBot’s cadences10, which follow up inside a live bot conversation, on delays you set. GoHighLevel’s standard bot does not recall calls3 and CloseBot has no voice13. Everyone else offers campaigns you launch from a list or rules you write67,69,73,77,82,85,57. Delta Agents’ recovery also picks up leads who went quiet in the 30 days before you switch it on, and writes each message with the contact’s memory and past calls.

4/9

Four competitors now keep per-contact memory. None of them carries a phone call into web chat, email and social DMs.

Up from two in July. Retell reads its memory back on calls only28; Fin’s covers chat and email but not calls86; Bland shares history across voice, SMS and iMessage on Enterprise plans34,78; Synthflow matches memory by phone number15. Delta Agents adds saved facts and past-call summaries to every chat and follow-up message, on every channel.

The comparison

Twelve capabilities. Ten platforms. Footnoted.

Capability marks reflect what each vendor documents publicly as of September 25, 2026, not marketing claims. Where a vendor claims something without documenting a mechanism, we say so.

✓Documented / shipped ◑Partial, see note ✗Not available or not documented
Capability Delta Agentschat + voice platform GHL AI EmployeeCRM-bundled AI CloseBotGHL-ecosystem chat Synthflowno-code voice SaaS Vapideveloper voice infra Retell AIvoice agent platform Bland AIenterprise voice infra ElevenLabs Agentsagents platform Chatbasesupport chatbot Finsupport AI agent
Chat and voice in one system ✓One contact record and one memory across web chat, SMS, email, social DMs and phone calls ◑Conversation AI and Voice AI are separate products with separate builders2,60 ✗Chat only, no voice13 ◑Voice-first; chat is a separate agent type66 ◑Chat API and inbound SMS; chat sessions expire after 24 hours20 ◑Voice, chat & SMS agents; converting creates a separate chat agent27 ◑One agent for voice, SMS and iMessage; messaging is Enterprise-only36,78 ✓One agent: web, phone, WhatsApp40 ✓Chat + inbound voice (launched April 2026)44 ✓Same agent: chat, email, voice48
Chat remembers your phone calls ✓Summaries of the contact’s recent calls added to every chat & follow-up message ◑Only inside one workflow action, not the standard bot3 ✗No voice to remember13 ◑Voice↔SMS/WhatsApp only, matched by phone #15 ✗No built-in cross-call memory24 ✗Memory is read back on calls only; chats never receive it28 ◑Shared history across voice, SMS & iMessage (Enterprise); web chat not documented34,78 ✗Not documented; DIY via Mem041 ✗Not documented45 ✗Fin Memory covers chat and email, not calls86
Per-contact long-term memory ✓Agent saves & corrects facts per contact, recalled on every turn ◑Workflow-scoped rolling summary3,7 ✗Summaries saved to CRM fields only11 ✓Memory matched by phone number (summary injection)15 ✗None; only data your server supplies at call time24 ✓Contact memory keyed to phone number (June 2026)28 ✓Native Memory: facts, summary, open items33,34 ✗Memory tool still not GA41 ◑Attribute storage, not AI-curated45 ✓Fin Memory (beta, Aug 2026): facts & summaries86
Autonomous lead recovery ✓Opt-in: finds leads who went quiet and re-engages them by SMS or email with full context: memory, events, past calls ◑Auto Follow-Up chases contacts who stop replying to the bot; calls need a workflow59,4 ◑Preset follow-up cadences10 ✗Campaigns & workflows you build (Enterprise)67 ✗Campaigns from a CSV you upload69 ✗Batch calls you schedule73 ✗Event triggers; nothing detects inactivity77 ✗Batch calling only82 ✗WhatsApp campaigns you launch85 ✗Rule-based nudges for unresponsive support chats57
Agent schedules its own follow-ups ✓Mid-conversation, the agent schedules, moves and cancels its own follow-ups ◑Honors a requested time; later steps use preset delays59 ◑Up to 4 fixed cadences per bot10 ✗Callback loops in workflows, Enterprise only67 ✗Build it yourself21 ◑Post-call workflow can open a CRM task or callback72 ✗Rule-based triggers only77 ✗Not documented ✗Not documented ◑Rule-driven auto follow-up57
AI builds your custom tools ✓Describe it in chat: AI writes & reviews the tool, then saves it switched off for you to test in a sandbox ✗Manual webhook config56 ✗Manual four-step wizard62 ◑Aurora sets up & tests HTTP actions; Enterprise only63 ◑Composer (alpha) configures tools; you provide the endpoint68 ◑Conductor configures functions; you supply the endpoint70 ✓Norm builds a tool & test-runs it on your real endpoint75 ◑Architect (alpha) edits tool settings in plain English80 ◑Backstage AI sets up webhook actions; plan-capped count83,46 ◑AI drafts Procedures; AI-built HubSpot connectors49
Human handoff + async alerts ✓Pause or take over any conversation from the inbox; the agent can alert your team on Slack, email or SMS ✓Handover action + call transfer5 ✓Pause + milestone notifications11 ◑Warm transfer; post-call email summaries to your team65 ◑Transfers; alerts on quality thresholds89 ◑Warm/cold transfer; metric alerts by email or webhook31,74 ✓Transfers; Slack alerts via Triggers; warm transfer Enterprise-only35,77 ✓Transfers; agent can flag a conversation for review39,81 ✓Helpdesk escalation + lead email alerts52 ✓Native inbox escalation48
Native GoHighLevel integration ✓Deep: contacts, custom fields, tags, conversations, calendars, pipelines, plus one-click import & nightly pipeline sync ✓Is GoHighLevel ✓Built for GHL8 ✓Marketplace app + sub-account import16 ✓Prebuilt tools for contacts & calendar23 ✓Native: contact sync + CRM & calendar tools30 ✗Not in its integration list79 ✗Not documented ✗Not documented ✗Not documented
Knowledge base with site crawling ✓Files, whole folders, URLs, recursive crawl, vector search ✓Crawler, up to 4,000 URLs6 ✓Website scrape, daily refresh; unlimited uploads11,61 ✓Files, URLs, site crawl, Zendesk import64 ◑Files only, <300 KB each; no crawl22 ✓Crawl + 24h auto-sync29 ✓Files + website via sitemap76 ✓Site crawl with auto-sync (July 2026)38 ◑1 to 40 MB per agent by plan43 ✓Articles, PDFs, URLs + helpdesk sync48
Run many clients (multi-tenant) ✓Agency dashboard; duplicate & share agents across locations; row level isolation87 ✓Sub-accounts; rebilling AI needs $497/mo plan2 ✓White-label portal, $397/mo8 ◑Sub-accounts; in-product rebilling removed Sep 15, 202617 ✗Orgs at $20/mo each extra; wrappers from $29/mo19,25 ◑One workspace per client; no white-label71 ◑Enterprise terms, negotiated32 ✗No first-party white-label42 ◑Client workspaces & agent-level access84 ✗Not documented
Your own LLM keys & models ✓Your OpenAI, Anthropic, Google or Groq key; main model + backups per agent ◑Model menu on GHL’s meter; no BYO keys1,3 ◑Menu of 4 providers; no BYO keys8,9 ◑Synthflow LLM only unless the account team enables BYOK18 ✓BYO keys, provider bills you19 ◑Model menu; custom-LLM endpoint for devs26 ✗Their own model stack32 ✓Custom LLM endpoint37 ◑Model picker; credits ×1 to 6 by model58 ✗Proprietary Fin models only50
Pricing model Included with Operator Base Elite: $199/mo ($7 first 7 days, full access, cancel anytime)88 · AI, voice & Twilio on your own accounts at provider cost, no Delta markup Pay per use, or $50 or $97/mo per location; realtime voice $0.10 to $0.20/min; phone extra1 Business from $97/mo by reply volume; agency $397/mo + $0.012/message8 Sales-led; enterprise contracts from $30k/yr; no public per-minute rate14 $0.05/min hosting + providers at cost; its calculator: $82 to $129 per 1,000 min19 Stacked per-minute; advertised $0.07 to $0.31/min; chat from $0.002/msg26 $0.14/min, or $0.12/min + $299/mo; transfer minutes & telephony extra32 $0 to $990/mo tiers; $0.08/min overage; LLM billed on top37 $40 to $500/mo ($32 to $400 billed yearly); credits ×1 to 6 by model43,58 $0.99 per outcome, assumed ones included; $9.99 per sales qualification47

✓ / ◑ / ✗ reflect capabilities documented in each vendor’s public docs and pricing pages as of September 25, 2026 (numbered footnotes link to the sources ledger). “Not documented” means the vendor’s public documentation does not describe the capability. Delta Agents marks cover features live for customers on the same date; features still in pilot are left out. SMS, email and social DMs, lead recovery, self-scheduled follow-ups and alerts by email or text run through a connected GoHighLevel account. Vendors ship fast, so verify anything load-bearing before you buy.

Start for $7 and run it on your own leads
7-day full access trial · $199/mo after · cancel anytime
Why it wins

Five things no other platform puts in one system.

Context engine

Your chat agent knows what was said on the phone.

Every Delta chat and follow-up message, on web chat, SMS, email, social DMs or a recovery send, is briefed with the contact’s saved memory facts, activity events, and summaries of their recent phone calls. A customer who called on Tuesday can text on Friday and the agent picks up mid-thought.

Everyone else: four competitors now keep per-contact memory, and each stops short. Retell reads its memory back on phone calls only; its chats write to it but never receive it28. Fin’s Memory (beta) covers chat and email, not calls86. Bland shares history across voice, SMS and iMessage, with messaging on Enterprise plans34,36. Synthflow keys memory to a phone number15. Vapi has no cross-call memory at all; third-party vendors sell the missing piece24. GoHighLevel documents call-transcript recall only inside a single workflow action, not its standard bot3.
Lead recovery

Revenue recovery that runs itself.

Switch it on and Delta Agents looks for leads who went quiet, including ones who were already quiet for up to 30 days, and can skip anyone who recently booked or bought. Then the same agent, with the full relationship in context, writes each message and sends it by SMS or email through your GoHighLevel account, texting only in allowed local hours.

Everyone else: GoHighLevel’s Auto Follow-Up59 and CloseBot’s cadences10 only follow up inside a live bot conversation, and neither knows what was said on a call3,13. The rest offer batch campaigns you launch from a list69,73,82 or rules you write yourself57,77.
Tool builder

Tools written by AI, switched on by you.

Say “look up order status from our API” and the builder reads the API docs, writes the tool and reviews it, then saves it switched off. Test it in an isolated sandbox and turn it on when you are happy. New capability in minutes, no developer.

Everyone else: AI tool helpers are now common, but most fill in a request to an endpoint you already run: Vapi’s Composer (alpha)68 and Retell’s Conductor70 ask you for the endpoint, Chatbase’s Backstage sets up webhook actions83, and Synthflow’s Aurora is Enterprise-only63. Fin’s AI builds HubSpot connectors49. Bland’s Norm comes closest: it builds a tool and test-runs it against your real endpoint75. GoHighLevel and CloseBot are still manual forms56,62.
Agency-native

Built multi-tenant from the first line of code.

Per-location agents, an agency-wide dashboard, one-click duplication and sharing of proven agents across client locations, one invite for a teammate across several locations, and every tenant’s data isolated in the database by row level security87.

Everyone else: GoHighLevel and CloseBot are agency-native too2,8, and Delta Agents plugs into GoHighLevel rather than replacing it. The voice platforms are not: Vapi charges $20 a month per extra organization and leaves serving clients to third-party wrappers (Vapify from $29/mo)19,25, Retell tells agencies to open one workspace per client71, Synthflow removed in-product client rebilling on September 15, 202617, and ElevenLabs offers no first-party white-label42.
Honest economics

Your keys. Provider cost. No meter games.

Delta Agents runs on your own keys: you pay OpenAI, Anthropic, Google or Groq their actual price. Voice runs on your own Retell account, or on your own OpenAI key with your own Twilio number, and texting goes through your own GoHighLevel account, each billed by that provider at its own rates. Delta adds no credit multipliers, no per-resolution fees and no markup. A few background jobs, such as memory extraction and search indexing, run on our keys at no charge to you.

Everyone else: Chatbase charges 1 to 6 credits per response depending on model58. Fin bills $0.99 per outcome, including when a customer simply leaves after an answer47, a design its own community calls flawed51. GoHighLevel runs AI on its own meter, with realtime voice at $0.10 to $0.20 a minute plus phone charges1. Vapi adds $0.05 a minute of hosting on top of your providers19.
Shipped since July

What’s new in Delta Agents.

Everything below is live for customers today. Features still in pilot are not on this page.

A second voice engine: OpenAI realtime.

Voice

Build a voice agent on OpenAI’s realtime voice model, running on your own OpenAI key, alongside the Retell engine. Hear a sample of each of ten voices before you pick one, and talk to the agent in your browser before you connect a number.

A greeting in the agent’s own voice.

Voice

On the OpenAI engine, the opening line is pre-recorded in the agent’s voice and checked word for word, so calls open with exactly the words you wrote. It can greet a known caller by first name.

Outbound calls that handle voicemail.

Voice

OpenAI engine agents can place outbound calls from your CRM automations, detect voicemail and transfer a caller to a number you set. Unanswered calls can be retried automatically, and calls get a written summary, sentiment and success flag, produced on your own key.

Your own Twilio account.

Telephony

Connect your own Twilio account and number for the calls your OpenAI engine agents make and take. Twilio bills you directly at its own rates, and Delta adds nothing on top.

Your GoHighLevel data on day one.

CRM

Import your GoHighLevel contacts, custom fields, tags, pipelines and opportunities in one click. A nightly sync then keeps imported pipelines current, and tags too for agents running lead recovery.

It sees the photos customers send.

Chat

On a vision-capable model, agents look at the photos customers text in and reply about what they show.

Alerts that tell you the fix.

Operations

Every alert explains what broke, why it matters and how to fix it, and critical ones are emailed to the account owner. You also get an alert when your own AI provider account runs out of credit.

Workflows and forms, drafted by AI.

Automation

Build automations with triggers, waits, branches and SMS or email steps, draft them with AI and dry-run them on a real contact. Hosted forms and surveys can start a workflow when someone submits.

Also new: take over any conversation from the inbox (the AI steps back while you handle it), connect your own MCP servers to an agent, add a whole folder to the knowledge base at once, export a location’s data or erase a single contact’s, and optional safety screening of inbound messages on each agent. How we protect data, and what we do not claim, is on our Security page.

Know the categories

What you’re actually comparing.

“AI agent platform” covers four very different kinds of product. Delta Agents is the only one in this comparison purpose-built as a done-for-you chat + voice agentic system with a context engine, for businesses and the agencies that serve them.

CRM-bundled AI

GoHighLevel AI Employee

Convenient if you live in one CRM, but chat and voice are separate products with separate builders2,60, memory is workflow-scoped3, there are no BYO keys, and AI runs on GoHighLevel’s meter: pay per use, or $50 to $97 per location per month1. Its Auto Follow-Up only chases contacts who go quiet mid-conversation59.

GHL-ecosystem chatbots

CloseBot (ZappyChat merged in12)

Capable text-only appointment setters. CloseBot’s own blog concedes the voice side to other vendors13, so the moment your leads want to talk, you’re integrating and paying for a second platform with a second brain.

No-code voice SaaS

Synthflow · ElevenLabs Agents

Strong phone agents. Synthflow’s chat is a separate agent type66 and its memory keys on phone numbers15; ElevenLabs now spans web, phone and WhatsApp40 but its memory tool is still not generally available41. Agency economics are tough: Synthflow is sales-led from $30,000 a year14 and removed in-product client rebilling17, and ElevenLabs has no first-party white-label42.

Developer infrastructure

Vapi · Retell AI · Bland AI

Excellent parts, increasingly productized. Vapi still leaves memory24 and follow-ups to you and needs a wrapper to serve clients25. Retell and Bland now ship contact memory28,34, and Retell a native GoHighLevel integration30, but their memory is built around the phone number, recovery is still campaigns you schedule73, and serving clients means a workspace per client71 or an enterprise contract. They reward teams with engineers.

Support-desk chatbots

Chatbase · Fin (+ Voiceflow, Botpress)

Built first to deflect support tickets; lead handling is a newer add-on (Fin charges $9.99 per sales qualification47). Fin, now a Salesforce company50, bills $0.99 per outcome, including assumed ones47,51. Builder platforms like Voiceflow and Botpress hand you a canvas and leave the memory and the build to you53,54,55.

The complete agentic system

Delta Agents

One system that chats and calls, keeps one memory per contact across channels, re-engages quiet leads on its own, writes custom tools with AI, and runs every client location from one dashboard, on your own keys.

Honesty, both ways

When you should not pick Delta Agents.

A comparison you can trust has to cut both ways. Three cases where a competitor is the right call:

You employ engineers and want raw infrastructure.

Pick Vapi or Retell

They’re excellent building blocks19,26. You’ll own every line of code and every provider bill, and your team can build much of what’s on this page themselves.

You need support-ticket deflection at scale, nothing else.

Pick Fin

Fin is the category benchmark for resolving support tickets on an existing helpdesk48. If closing leads isn’t the job, we’re not the tool.

You want the cheapest add-on inside one CRM.

Pick GHL’s AI Employee

Pay per use, or $50 to $97 a month per location1, and it lives where you already work. If cross-channel memory, recovery across your list, and AI-built tools don’t matter to your business, it’s right there.

Still here?

Then this was built for you

You close leads across chat and phone, you can’t afford an agent with amnesia, and you don’t have engineers to spare. That’s exactly the operator Delta Agents was built for.

One system. Chat + Voice. One memory.

Delta Agents keeps one memory per contact across chat and phone, recovers the leads you’d have lost, and writes custom tools with AI. Included with Operator Base Elite.

Start for $7: full access
$199/mo after your 7-day trial · your own API keys · cancel anytime
Receipts

Sources ledger.

Every numbered footnote above resolves here. All sources were fetched and read on September 25, 2026 (first edition: July 3, 2026).

GoHighLevel

  1. [1] AI Product Pricing (three plans: pay per use, $50/mo Growth, $97/mo Unlimited per location; realtime voice $0.10 to $0.20/min; phone charges separate; model menu) · help.gohighlevel.com/…/155000006652-ai-product-pricing
  2. [2] AI Employee: Access, Rebilling and Reselling ($497/month agency plan to rebill AI usage; Conversation AI and Voice AI covered as separate products) · help.gohighlevel.com/…/155000003906-ai-employee-overview
  3. [3] Workflow Action: AI Agent (conversation memory, cross-channel history, call-transcript retrieval) · help.gohighlevel.com/…/155000007600-workflow-action-ai-agent
  4. [4] Voice AI Outbound Calling (calls placed through a workflow) · help.gohighlevel.com/…/155000006598-voice-ai-outbound-calling
  5. [5] Conversation AI Human Handover Action · help.gohighlevel.com/…/155000005615
  6. [6] Knowledge Base Web Crawler (up to 4,000 URLs per knowledge base) · help.gohighlevel.com/…/155000008417
  7. [7] Idea board: “AI to Remember Recent Conversations” (user-disputed) · ideas.gohighlevel.com/conversation-ai/p/ai-to-remember-recent-conversations
  8. [56] Voice AI Custom Actions (manual webhook configuration) · help.gohighlevel.com/…/155000005461-voice-ai-custom-actions
  9. [59] Conversation AI Auto Follow-Up (re-engages contacts who drop off in a bot conversation; up to 5 steps per scenario on delays you set) · help.gohighlevel.com/…/155000005500
  10. [60] Create, Test, and Deploy a Voice AI Agent (its own builder) · help.gohighlevel.com/…/155000004107

CloseBot / ZappyChat

  1. [8] Plans & pricing (Business Core from $97/mo; Agency Core $397/mo + $0.012/message rebillable, white-label client portal; “we do not allow bring your own key”; native GoHighLevel and HubSpot) · closebot.com/plans
  2. [9] AI providers (OpenAI, Anthropic, Gemini and Grok, supplied by CloseBot) · docs.closebot.com/…/11403917
  3. [10] Follow-ups (up to 4 preset cadences) · docs.closebot.com/…/11529498-follow-ups
  4. [11] “17 Features You Don’t Have to Wait For” (webhooks, notifications, summaries-to-fields, daily website re-check) · closebot.com/blog/17-features…
  5. [12] CloseBot acquires ZappyChat (sunset Dec 31, 2025) · closebot.com/blog/closebot-acquires-zappychat
  6. [13] “CloseBot for Conversation AI and others for Voice AI” (no voice product) · closebot.com/blog/why-smart-marketing-agencies…
  7. [61] Unlimited uploads on every plan (Sep 24, 2026) · closebot.com/blog/ai-appointment-setter-closebot-analytics
  8. [62] Custom tools built through a four-step wizard (Sep 8, 2026) · closebot.com/blog/highlevel-alternatives…

Synthflow

  1. [14] Pricing (sales-led; enterprise contracts from $30,000/yr) · synthflow.ai/pricing
  2. [15] Memory (cross-channel memory matched by phone number; voice, SMS and WhatsApp) · docs.synthflow.ai/memory
  3. [16] GoHighLevel integration · docs.synthflow.ai/gohighlevel
  4. [17] Reselling migration guide (in-product reselling through Stripe removed Sep 15, 2026) · docs.synthflow.ai/reselling-migration-guide
  5. [18] Changelog, Aug 25, 2026 (third-party models no longer available on standard plans; Synthflow LLM only) and OpenAI key setup (shown only when BYOK is enabled for your account) · docs.synthflow.ai/changelog/2026/8/25, docs.synthflow.ai/openai
  6. [63] Aurora (scaffolds, creates, attaches and tests actions; Enterprise plan only) · docs.synthflow.ai/aurora
  7. [64] Documentation index, Knowledge Base entry (PDFs, documents, single URLs, website crawl, Zendesk import; the article page itself was in a redirect loop on Sep 25, 2026) · docs.synthflow.ai/llms.txt
  8. [65] Post-call email (call summary to up to 50 addresses) · docs.synthflow.ai/post-call-email
  9. [66] Create an agent (voice and chat are separate agent types) · docs.synthflow.ai/create-an-agent
  10. [67] Workflows (Enterprise plan only) · docs.synthflow.ai/workflows

Vapi

  1. [19] Pricing ($0.05/min hosting, providers at cost; calculator $82 to $129 for 1,000 min; extra organizations $20/org/mo; extra concurrency $10/line/mo) and provider keys (the provider bills you directly) · vapi.ai/pricing, docs.vapi.ai/customization/provider-keys
  2. [20] Chat quickstart and SMS chat (session-based context; sessions expire after 24 hours) · docs.vapi.ai/chat/quickstart, docs.vapi.ai/chat/sms-chat
  3. [21] Custom function tools (you host the endpoint) · docs.vapi.ai/tools/custom-tools
  4. [22] Knowledge base (file uploads, <300 KB recommended) · docs.vapi.ai/knowledge-base
  5. [23] GoHighLevel tools (contacts and calendar) · docs.vapi.ai/tools/go-high-level
  6. [24] “Vapi voice agents have no built-in cross-call memory” (third party) and Vapi’s personalization guide (data your server supplies at call time) · hindsight.vectorize.io (Jun 3, 2026), docs.vapi.ai/assistants/personalization
  7. [25] Vapify white-label wrapper pricing (Starter $29/mo, one sub-account) · vapify.agency/pricing
  8. [68] Composer (alpha; “Your part: Provide your API endpoint URL”) · docs.vapi.ai/composer
  9. [69] What’s new (dashboard Campaigns: upload contacts, pick an assistant) · docs.vapi.ai/whats-new
  10. [89] Monitoring quickstart (alerts on quality thresholds by email, Slack or webhook) · docs.vapi.ai/observability/monitoring-quickstart

Retell AI

  1. [26] Pricing (stacked per-minute, $0.07 to $0.31/min advertised; chat from $0.002/msg; 20 concurrent calls free) · retellai.com/pricing
  2. [27] Chat Agents launch and chat agent docs (conversion creates a separate chat agent) · retellai.com/blog/meet-retell-chat-agents…, docs.retellai.com/build/create-chat-agent
  3. [28] Contact memory (matched by phone number; “Chats write to contact fields but don’t receive them as variables”) · docs.retellai.com/integrations/build-contact-memory
  4. [29] Knowledge base (crawl + 24h auto-sync) · docs.retellai.com/build/knowledge-base
  5. [30] GoHighLevel integration (contact sync; contact, opportunity, tag and calendar tools) · docs.retellai.com/integrations/gohighlevel, retellai.com/integrations/go-high-level
  6. [31] Transfer call (warm/cold) · docs.retellai.com/…/transfer-call
  7. [70] Conductor (“For a custom function, provide the endpoint details”) · docs.retellai.com/conductor/build-and-refine
  8. [71] Workspaces (“create one workspace per client”) · docs.retellai.com/accounts/workspace
  9. [72] Agent workflows (post-call follow-up task or callback) · docs.retellai.com/agent/agent-workflow
  10. [73] Batch calls · docs.retellai.com/deploy/make-batch-call
  11. [74] Alerting (email or webhook when a metric rule triggers) · docs.retellai.com/features/alerting-overview

Bland AI

  1. [32] Pricing (Start $0.14/min, no platform fee; Build $0.12/min + $299/month; Enterprise custom; transfer minutes extra; telephony separate; Bland runs its own model stack) · bland.ai/pricing
  2. [33] Memory + Slack + HubSpot launch (Mar 7, 2025) · bland.ai/blog/memory-slack-hubspot-integrations
  3. [34] Memories docs (facts, summary, open items) · docs.bland.ai/tutorials/memories
  4. [35] Warm transfer (Enterprise-only) · docs.bland.ai/tutorials/warm-transfer
  5. [36] SMS (Enterprise-only; $0.02/message) · bland.ai/product/sms, docs.bland.ai/tutorials/messaging/sms
  6. [75] Norm (builds a tool, test-runs it against the real endpoint, attaches it) · docs.bland.ai/integrations/mcp/norm
  7. [76] Knowledge (website discovery from the sitemap) · docs.bland.ai/agents/knowledge
  8. [77] Triggers (event, conditions, action; Slack posts; “Nothing here is gated by plan”) · docs.bland.ai/agents/triggers
  9. [78] Messaging overview (SMS, iMessage and voice keep one shared history) · docs.bland.ai/tutorials/messaging/overview
  10. [79] Integrations (Salesforce, HubSpot, Zapier, Make) · bland.ai/integrations

ElevenLabs Agents

  1. [37] Agents pricing ($0 to $990/mo tiers; $0.08/min for extra minutes; LLM billed on top) and custom LLM docs · elevenlabs.io/pricing/agents, elevenlabs.io/docs/…/customization/llm
  2. [38] Knowledge base and the Jul 27, 2026 changelog (crawl jobs with sitemap URLs and automatic synchronization) · elevenlabs.io/docs/…/knowledge-base, elevenlabs.io/docs/changelog/2026/7/27
  3. [39] transfer_to_number system tool · elevenlabs.io/docs/…/transfer-to-number
  4. [40] “Introducing ElevenLabs Agents” (phone, web and apps) and the WhatsApp channel docs · elevenlabs.io/blog/introducing-elevenlabs-agents, elevenlabs.io/docs/eleven-agents/whatsapp
  5. [41] No native cross-call memory; Mem0 workaround (memory tool schema only, in the Jun 15, 2026 API changelog) · memu.pro, docs.mem0.ai, elevenlabs.io/docs/changelog/2026/6/15
  6. [42] No first-party white-label (third-party wrapper guide, updated Sep 15, 2026) · voiceaiwrapper.com/insights/white-label-elevenlabs-ai-guide-2026
  7. [80] Architect (alpha; “you describe what you want and Architect makes the edit”) · elevenlabs.io/docs/eleven-agents/operate/architect
  8. [81] Flag issue for review system tool · elevenlabs.io/docs/…/flag-issue-for-review
  9. [82] Batch calling · elevenlabs.io/docs/eleven-agents/phone-numbers/batch-calls

Chatbase

  1. [43] Pricing (Hobby $40, Standard $150, Pro $500 per month; 20% off yearly; 1 to 40 MB training content per agent) · chatbase.co/pricing
  2. [44] Chatbase Voice launch (Apr 8, 2026; same agent answers calls) · chatbase.co/changelog/chatbase-voice-is-now-available
  3. [45] Contacts (attribute storage + identity verification) · chatbase.co/docs/…/contacts-overview
  4. [46] Custom actions (manual config; plan-capped count per agent) · chatbase.co/docs/…/custom-action
  5. [52] Instant email notifications for new leads · chatbase.co/changelog/instant-email-notifications-for-new-leads
  6. [58] Credit multipliers by model (1 to 6 credits per response) · chatbase.co/docs/faq
  7. [83] Backstage (AI assistant that creates actions such as Slack notifications and webhooks) · chatbase.co/docs/…/backstage
  8. [84] Agent-level permissions (Sep 2026; for agencies managing several agents) · chatbase.co/changelog/agent-level-permissions-here
  9. [85] Outbound campaigns (WhatsApp is the only campaign channel) · chatbase.co/docs/…/outbound-campaigns

Fin (formerly Intercom)

  1. [47] Pricing ($0.99/outcome; $9.99/sales qualification; Fin Voice custom) and outcomes (a customer who leaves after Fin’s answer counts as an assumed resolution) · fin.ai/pricing, intercom.com/help/…/8205718-fin-ai-agent-outcomes
  2. [48] Channels & Voice (same agent, unified knowledge across channels) · fin.ai/channels, fin.ai/voice
  3. [49] Fin Procedures and the HubSpot connector (“AI inspects your HubSpot schema, runs a test request, and assembles the connector”) · intercom.com/help/…/12599391, intercom.com/help/…/16299042
  4. [50] Salesforce completes acquisition of Fin (Sep 10, 2026; proprietary Fin model suite) · salesforce.com/news/press-releases/2026/09/10
  5. [51] Community: “Fin’s flawed assumed-resolved pricing design” · community.intercom.com/…/8929
  6. [57] Auto follow-up for unresponsive customers (rule-driven) · intercom.com/help/…/11793341
  7. [86] Fin Memory (beta since Aug 13, 2026; Messenger and email) · intercom.com/help/…/16001516-fin-memory, intercom.com/changes/…/154444

Others

  1. [53] Voiceflow pricing & billing (credits; prices shown inside the product) · voiceflow.com/pricing, voiceflow.com/docs/…/billing
  2. [54] Botpress pricing (Plus and Team plans; no way to turn off auto-recharge for conversation packs) · botpress.com/pricing
  3. [55] Botpress pricing update (May 14, 2026: per-conversation, $150/mo entry, mandatory auto-recharge) · botpress.com/blog/pricing-update-may-2026

Delta Agents & Operator Base

  1. [87] Operator Base Security (tenant isolation with row level security; API keys encrypted with AES-256-GCM; what we do not claim) · try.operatorbase.app/security
  2. [88] Checkout (“Charged $7 today. Day 8 onward: $199/mo.”) · try.operatorbase.app/buy