AI Agent Index
ByHeather MacAvelia·Independently reviewed·Published Jul 20, 2026·Updated Jul 20, 2026
Independently verified against live vendor data on Jul 20, 2026.

Oz by Anyreach is an AI teammate across Slack, Teams, email, voice, and web that returns finished work. Credit-metered from $33/seat/mo billed annually (USD); every send, payment, or delete waits for one-tap approval.

How we scored it

Autonomy

4/5

Integrations

5/5

Pricing clarity

4/5

Evidence

Be an early reviewer →

Setup

4/5

The facts

From

$33

subscription · annual

GitHub

Stars

G2

Rating

MCP

⚡ Server

Exposes server

Reviews

Be the first →

Oz is an AI teammate built by Anyreach that works across the tools a team already uses. You ask it for something in plain language in Slack, Microsoft Teams, Gmail, Outlook, on a call, or from the web dashboard and iOS app, and it returns finished work rather than a chat reply: research briefs, documents, slide decks, spreadsheets, pipeline updates, and operational tasks. Oz runs on a credit-metered model it describes as paying for outcomes, not tokens, with 14 built-in skills, standing orders, and scheduled (cron) automations. It is aimed at founders, operators, and small-to-mid teams that want one teammate to absorb recurring cross-functional work. Oz connects to more than 3,000 tools, with named deep integrations including Salesforce, HubSpot, Notion, Attio, Apollo, Stripe, GitHub, Linear, Google Drive, and Mercury, and it is API-first. It publishes its own MCP server, so Claude Desktop, Cursor, or any MCP client can connect into Oz's knowledge graph, and it exposes a documented REST API, webhooks, and an interactive API explorer. Reads run silently; every consequential action (a send, payment, merge, or delete) posts an Approve or Reject card before it executes, and teams can configure which action types auto-approve. This human-in-the-loop gate is a core design choice rather than a limitation, and it distinguishes Oz from coworkers such as Viktor (from $50/mo) that execute autonomously without per-step approval. Pricing is public and self-serve in USD. Solo is $33 per seat per month billed annually ($39 month-to-month) with 800 credits per seat and $0.12 overage; Team is $49 annually ($59 monthly) with 1,500 credits; Business is $74 annually ($89 monthly) with 3,000 credits, SSO, admin, audit, and custom skills; Enterprise is a custom committed pool gated behind Talk to sales, which is why pricing transparency is scored mostly-public rather than public. Annual billing is 2 months free. Credits are consumed by work done: a lookup is 1 credit, an account brief 12, a branded deck 40, and overage meters at the per-tier rate with no hard cutoff. What the entry price does not buy: the 800 Solo credits are a soft budget, not a cap, so heavy users pay overage on top of the seat fee, and shared team memory, standing orders, approval workflows, SSO, and pipeline access are gated to the higher tiers. Data residency, a signed DPA, and per-role fleets require Enterprise. Oz is not the right fit for every buyer. Teams that want fully hands-off execution with no approval step may prefer a coworker that acts without gates. Buyers who need a long independent track record should note that Oz is an early-access product from a 2024 company with no third-party review base yet on G2, Trustpilot, or Product Hunt. On security, Anyreach holds an independently audited SOC 2 Type 2 report and HIPAA certification and is AIUC-1 compliant; GDPR, ISO 27001, and PCI DSS are listed as in progress on its trust center, not yet certified. Customer content is not used to train foundation models, per the privacy policy. Current state Q3 2026: Oz is rolling out to early-access workspaces with a free trial, self-serve Solo, Team, and Business plans, native Slack, Teams, Gmail, Outlook, voice, web, macOS, Windows, and iOS entry points, and a first-party MCP server.

Pricing

subscription · $33/mo annual

View pricing ↗

Segment

b2b

Setup

easy

Verified

Jul 20, 2026

Transparency

Mostly Public

Contract

Monthly or Annual

Data training

Not Trained

Autonomy

Human Optional

Capabilities

autonomousworkflow-builderdata-analysisreportingcrm-syncscheduling

Pros & Limitations

Editorial assessment

Pros

  • Oz delivers finished work rather than chat replies: ask it in plain language from Slack, Teams, Gmail, Outlook, a call, or the web, and it returns research briefs, slide decks, spreadsheets, and pipeline updates across your connected tools in a single run. Pricing is outcome-metered rather than token-based, so a credit maps to a unit of work done and costs stay legible.
  • Human control is built into every consequential action: reads run silently, while every send, payment, merge, or delete posts a one-tap Approve or Reject card. Teams can set which lower-risk actions auto-approve, which makes Oz a strong fit for security-conscious teams that want autonomy without handing over unattended write access.
  • Unusually deep integration and security posture for an early-stage product. It offers 3,000+ integrations with named bidirectional connectors like Salesforce, HubSpot, Stripe, GitHub, and Notion, plus an API-first design and a first-party MCP server that external clients such as Claude or Cursor can query. It is backed by an independently audited SOC 2 Type II report, HIPAA, and AIUC-1, and customer content is never used to train foundation models.

Limitations

  • Credit-metered pricing means the $33 Solo seat is a floor, not a ceiling: the 800 included credits are a soft budget and heavy users pay per-credit overage on top of the seat fee.
  • No independent track record yet: Oz is an early-access product from a 2024 company with no third-party reviews on G2, Trustpilot, or Product Hunt, so buyers cannot yet triangulate quality.
  • Several controls are gated or pending: shared memory, SSO, and pipeline access require higher tiers, and GDPR, ISO 27001, and PCI DSS are still in progress on the trust center, not certified.

Technical Details

Deployment
cloudwebapimobiledesktop
Model architectureThird-party foundation models accessed through a commercial provider API under a no-training contract; specific model providers are not publicly named on the site or trust center.
Avg setup time< 1 hour (start free, install into Slack, Teams, Gmail, or Outlook, then ask Oz in plain language to run the first task)
Autonomous rateOz autonomously runs reads, research, drafting, and artifact generation across connected tools in a single pass, and executes standing orders and scheduled tasks on a set cadence. Consequential actions, meaning any send, payment, merge, or delete, post an Approve or Reject card and wait for a human unless auto-approve is configured for that action type.
Integrations
SlackMicrosoft TeamsGmailOutlookSalesforceHubSpotNotionAttioApolloStripeGitHubLinearGoogle DriveMercury
Security
SOC 2 Type IIHIPAAAIUC-1

Similar agents

Rating

On Our Radar

Independently reviewed. A full rating unlocks with community reviews or public signal.

Rate this agent
How we score this →

Recognition

MCP Server VerifiedListed 2026
Embed these on your site →

Industries

SaaSB2BStartupsEnterpriseSMB

Leave a review

Never displayed publicly.

Agent Stacks

See workflow stacks that feature Oz.

Compare

Related Content

Manage this listing

Badges, review collection, and listing updates for verified vendors.

Vendor dashboard →

Free · Every Two Weeks

AI Agent Price & Rating Tracker

Price changes, new agent launches, acquisitions, and rating updates across 344+ AI agents. Verified against live vendor data, not vendor marketing.

No spam. Unsubscribe anytime. We never share your email.