Perspective AI is a conversational research and intake platform built around four purpose-built agent types: Interviewer, Concierge, Evaluator, and Advocate. Instead of static forms or surveys, the platform runs adaptive text or voice conversations that ask real follow-up questions and adjust based on what a participant says.
The company was founded in 2024 by Guy Nirpaz, previously an EVP of Engineering at GigaSpaces Technologies, and is headquartered in Milpitas, California. Its stated goal is to replace rigid forms and one-shot surveys with structured, schema-aligned data captured through natural conversation.
Each agent has a distinct job. The Interviewer runs qualitative research interviews at scale, the Concierge replaces intake forms and lead-qualification flows, the Evaluator turns rating-scale surveys into a conversation that explains the score, and the Advocate listens to a person’s situation before delivering tailored, persuasive messaging from a playbook.
Perspective ships a documented MCP server and SDK, so teams can create agents, check project status, and analyze conversations directly from Claude Code, Claude Desktop, Cursor, or VS Code rather than only through the web dashboard.
Pricing
Perspective AI offers a free plan that includes 100 credits toward AI-powered conversations, with no card required to start. The Pro subscription starts at $99 per month for 1,000 credits, aimed at teams running continuous customer research or ongoing feedback programs, with fresh credits issued monthly and cancellation available anytime. Credits are consumed per activity: a full conversation costs 10 credits, while a research project or an analysis session costs 1 credit each. Enterprise plans and a full-service Research as a Service option, where Perspective’s team designs, runs, and analyzes the study, are both available by contacting sales rather than through published tiers.
* Disclaimer: Please note that pricing information may not be up to date. For the most accurate and current pricing details, refer to the official website.
Key Features
-
✓
Four dedicated agent types: Interviewer, Concierge, Evaluator, and Advocate
-
✓
Dynamic follow-up questions probe vague or incomplete answers in real time
-
✓
Structured data extraction turns dialogue into schema-aligned, exportable fields
-
✓
Automated synthesis surfaces themes, sentiment, and trust scores from transcripts
-
✓
MCP server connects agent creation and analysis to Claude Code and Cursor
-
✓
Embeddable widgets deploy conversations on websites, funnels, and landing pages
Use Cases
Scaled Qualitative Research
Researchers deploy the Interviewer agent to run hundreds of open-ended conversations at once, without manual scheduling. Perspective probes deeper on vague answers and returns synthesized themes across the full participant pool.
Conversational Lead Qualification
Sales and marketing teams use the Concierge to greet website traffic, infer budget and timeline from natural conversation, and route qualified prospects to a meeting link. Completion rates rise compared with static forms.
Applicant and Candidate Screening
Recruiting teams use the Evaluator agent to score responses against defined rubrics asynchronously. Instead of a static test, candidates answer scenario questions and get follow-up probes on incomplete answers.
Adaptive Client Intake
Legal, insurance, and financial services teams replace multi-field intake forms with a Concierge conversation that captures case or policy details, validates them against business rules, and hands off structured records.
Personalized Outreach and Objection Handling
Customer teams deploy the Advocate agent to listen to an individual’s situation before responding with tailored messaging drawn from an approved playbook, addressing specific objections rather than sending generic outreach.
Strengths & Weaknesses
Strengths
Four purpose-built agent types cover distinct jobs instead of one generic chatbot.
Published pricing starts free and scales through a transparent, itemized credit system.
A documented MCP server lets teams build and analyze agents from Claude Code or Cursor.
Structured data extraction routes conversation output directly into CRM and workflow tools.
A large template library and multiple embed formats speed up deployment across use cases.
Weaknesses
Pricing above the $99 per month Pro tier requires an enterprise sales conversation.
Public docs name Slack, HubSpot, and Gmail as automation targets but do not list a full native integration catalog.
Results depend on participants engaging in multi-turn conversation rather than quick clicks.
Complex or highly technical evaluations may still need human review to catch misinterpretation.
Who Is This For?
UX and market researchers who need qualitative interviews at scale without manual scheduling or transcription.
Recruiting and HR teams that want to standardize first-round candidate screening against a rubric.
Marketing, sales, and RevOps teams that want to qualify inbound leads and personalize outreach conversations.
Developers and technical teams who want to create and analyze agents from Claude Code, Cursor, or VS Code.
Frequently Asked Questions
What is the difference between the four Perspective AI agents?
The Interviewer runs open-ended qualitative interviews, the Concierge replaces forms with structured intake conversations, the Evaluator scores responses against a rubric, and the Advocate delivers tailored persuasive messaging after listening first.
How much does Perspective AI cost?
The free plan includes 100 credits. The Pro subscription starts at $99 per month for 1,000 credits, and enterprise pricing is available by contacting sales.
How do Perspective AI credits work?
Each full conversation costs 10 credits, while a research project or analysis session costs 1 credit. Credits refresh monthly on paid plans and can be topped up as needed.
Can I try Perspective AI without paying?
Yes. Sign-up includes a free plan with 100 credits toward AI-powered conversations, and no credit card is required to start.
Does Perspective AI integrate with my existing tools?
Perspective’s integration tooling references Slack, HubSpot, and Gmail as automation channels, alongside webhooks and a documented API and MCP server for custom connections. A full catalog of native integrations is not published.
Can I connect Perspective AI to Claude or Cursor?
Yes. Perspective runs a native MCP server at getperspective.ai/mcp that connects to Claude.ai, Claude Desktop, Claude Code, Cursor, and VS Code, letting you create and analyze agents from those tools directly.
How steep is the learning curve for building an agent?
Agents are configured through a prompt-based design flow covering goals, rubrics, tone, and guardrails, plus over 180 templates by role and industry, which shortens setup for common use cases.
Can I export the structured data Perspective AI collects?
Yes. Conversations produce schema-aligned structured fields that can be reviewed in the dashboard and sent downstream to CRM or workflow tools through automations.
Is Perspective AI suitable for highly technical or regulated evaluations?
It can handle structured intake for regulated workflows like insurance or legal intake, but complex technical domains may still benefit from human review to catch edge cases the agent misreads.
Does Perspective AI publish security or compliance certifications?
Perspective maintains a public trust page at trust.getperspective.ai, but specific SOC 2, ISO, or HIPAA certification status is not confirmed on the pages reviewed here.
Perspective AI offers a documented REST API and SDK, plus a native MCP server (getperspective.ai/mcp) for Claude.ai, Claude Desktop, Claude Code, Cursor, and VS Code. Its automation tooling names Slack, HubSpot, and Gmail as supported channels, and webhooks are available for custom delivery. A full public list of pre-built native app integrations beyond these examples is not published.