New A/B testing and analytics for your agents

AI agents that answer every conversation, and show their work

Put AI agents on Gmail, Outlook, GoHighLevel, SmartLead, Instantly and your website chat. Design them as visual job flows, let them reply, book, tag and hand off, and see every thought and tool call behind every message.

Free plan forever · no card needed · bring your own AI keys

Messages arrive by SMS, email and web chat; ReplySetter routes each through its flow, calls the right tool, replies on the same channel and logs every stepINBOUNDMayavia SMSHi! Can I book a cleaningthis week?Danavia GmailDo you offer paymentplans for implants?Visitorvia Web chatI'd rather talk to areal personCHANNELSGmail · Outlook · SMS · WhatsAppInstagram · SmartLead · Instantly · WebFront-desk agentjob flow · AvaStartAnswerBookHand-offscenariocheck_availabilitysearch_knowledgehandoff_to_humanREPLY · SAME CHANNELOf course! Thursday 3pm orFriday 10am?sent via SMSYes: 0% interest over 6 or 12months, approved same day.sent via GmailNo problem! Sam from our teamwill reply in a few minutes.sent via Web chatRun traceStepBookCharacterAvaToolcheck_availabili…Thu 15:00 · Fri 10:00 freeThinking412 tokensLatency2.1 sCost$0.006bookingStepAnswerCharacterAvaToolsearch_knowledgeFinancing.pdf · 0.93Thinking288 tokensLatency1.7 sCost$0.004groundedStepHand-offCharacterAvaToolhandoff_to_humanAssigned to Sam · notifiedThinking164 tokensLatency0.9 sCost$0.002scenario
Gmail Outlook GoHighLevel SMS WhatsApp Instagram Facebook SmartLead Instantly Website chat Anthropic OpenAI OpenRouter Pinecone Supabase
10+channels, from email to Instagram
350+AI models, with instant failover
130+MCP tools for automation
100%of replies traced and explainable
How it works

From connected inbox to confident autopilot

Four steps, and you stay in control at every one of them.

Connect your inboxes

Gmail, Outlook, GoHighLevel, SmartLead, Instantly or a website chat. OAuth, a token or an API key, and history syncs in.

Design the job

Pick a character, then lay out steps, objectives, exits and scenarios on a visual canvas. One step is already a working agent.

Test it safely

Play the contact in sandbox conversations, read every thought and tool call, and let Diagnose propose fixes.

Go live, your way

Start in suggest mode and approve everything, then move to draft or full auto as the results earn your trust.

Visual job flows

Agents built like a playbook, not a prompt

Lay out the job step by step. Each step collects its objectives, picks its exits and hands over to the next, while scenarios catch the off-script moments anywhere in the conversation.

  • Steps that stay until the job is done
  • AI conditions and switches that decide instantly
  • Scenarios catch the moments your script didn't plan for
A conversation moves through a job flow: objectives are collected, a condition routes it, and a scenario lane watches every messageMAIN FLOWStartQualify leadAgent step · AvaOBJECTIVESBudget$5kTimelinethis monthAddressLeedsQualified?AI conditiontruefalseBook callAgent stepHand offStopCUSTOM SCENARIOS · checked on every messagePricing questionsensitivity 6Wants a humansensitivity 5Opt-outpriority 10Wait, how much does this cost?Pricing question scores 8/10: jump, then return
CRM actions & booking

From conversation to booked appointment

Agents check calendars and book, then tag the contact, save every answer to the right field, move the deal and start the workflow, all inside the conversation.

  • From “does Thursday work?” to booked, in one reply
  • Every answer lands in the right field
  • Tags, deals, notes, tasks and workflows
The agent checks calendar availability, books Thursday at 3:30, then tags the contact, moves the deal, adds a note and starts a workflowSMS · Jordan LeeCan I come in Thursdayafternoon?You're booked forThursday at 3:30pm.See you then!check_availability(Thu)book_appointment(Thu 3:30pm)MonTueWedThuFri1 pm2 pm3 pm4 pm5 pm3:30JLJordan LeeGoHighLevel contactSERVICETeeth whiteningTAGSbookedwhiteningOPPORTUNITYNew leadAppointment bookedNOTESBooked Thu 3:30 via SMS.Wedding in June.WORKFLOWSAppointment reminders
Test & Debug

Every thought and tool call, on the record

Click any reply to see which step answered and why, the model's reasoning, every tool call and the exact prompt. Rate a reply and the AI proposes the fix.

  • Test conversations that can't embarrass you
  • Why, Thinking, Tools and Prompts
  • Rate it, diagnose it, apply it, re-run it
Clicking a reply opens the debugger, which cycles through Why, Thinking, Tools and Prompts for that turnTest · Happy path (SMS)Hi! Is there parking?Yes! Free parking right behindthe studio. Want me to bookyour first class?Yes please!Type as the contact…Debugger · turn 2WhyThinkingToolsPromptsStepQualify leadCharacterAva · claude-sonnet-5Exit takennone (objective pending)Learnedwants: parking infoSCENARIO SCORESPricing question1/10Wants a human0/10Opt-out0/10STATE{ step: "qualify", objectives: { budget: null, goal: "beginner classes" } }The contact is asking about parking. That's afactual question about the studio, so I shouldcheck the knowledge base rather than guess.Found: free parking behind the building.Budget objective is still open, but it's tooearly to ask. Answer, then offer to book thefirst class to keep momentum.812 reasoning tokenssearch_knowledge240 ms{ query: "parking at studio" }→ 2 passages · 0.91, 0.74save_objective12 ms{ goal: "beginner classes" }→ savedsend_reply8 ms{ text: "Yes! Free parking…" }→ queued (test: simulated)systemYou are Ava, a friendly concierge…## Step: Qualify leadObjectives: budget (required), goal…Tools: search_knowledge, …user<message>Hi! Is there parking?</message>assistantYes! Free parking right behind…in 3,412 · out 92 · 1.4 s · $0.0118
A/B testing

Prove what converts on real conversations

Split new conversations between variants of a step, character or offer, and let proper statistics tell you when a winner is real.

  • Sticky variants on real conversations
  • Win on bookings, tags, replies, lead score or sentiment
  • A winner only when it's really a winner
New conversations are split between variant A and variant B; B converts more often and is proven the winner at 97% confidenceNEW CONVERSATIONS50 / 50A/B testA · ControlOpens with a questionbookedbookedB · ChallengerOpens with a complimentbookedbookedGOAL · BOOKED18.0%A · 412 conv26.9%B · 405 convLIFT+49%CONFIDENCE97% · needs 95%B proven
Pricing

Start free, scale when it pays off

Every plan uses your own AI keys, so there's never a markup on usage.

Free

$0forever

1 agent · 1 connected inbox · Bring your own AI keys

Start for free

Agency

$297/ month

Unlimited agents · 25 connected inboxes · Many GHL sub-accounts per agent

Start with Agency

Compare all plans

FAQ

Questions, answered

What is ReplySetter?

ReplySetter puts AI agents on your inboxes. Each agent follows a visual job flow, speaks as a character you design, uses tools like calendars and your CRM, and replies on the same channel the message came from, with every decision recorded.

Which channels does it work with?

Gmail, Outlook, GoHighLevel (SMS, email, WhatsApp, Facebook, Instagram and live chat), SmartLead, Instantly, and an embeddable website chat widget.

Which AI models can I use?

Anthropic, OpenAI and 350+ models through OpenRouter. Each character can have up to five backup models from any provider that take over instantly if the main one fails.

Will the AI send messages without my approval?

Only if you let it. Each agent has an autonomy level: suggest (everything waits for approval), draft (sends wait, the rest runs) or auto.

Can I see why an agent replied the way it did?

Yes. Every run stores the step and character that answered, scenario scores, the model's reasoning, every tool call and the raw prompts, shown in a debugger.

Can I control it from Claude Code or Cursor?

Yes. A built-in MCP server exposes 130+ tools, so you can build, test and operate agents from any MCP client with a personal API token.

Get started

Put an agent on your inbox today

Connect an inbox, describe the job, test it in a sandbox and go live with exactly the autonomy you're comfortable with.