Aide

What it does

Done-for-you service

PartlyYou run the software, but Aide's own people do the implementation and, on Scale and above, build custom workflows for you, so it sits between tool and delivered service.Priority support + custom workflow buildsaide.app/pricing ↗

How much it does unattended

Runs autonomously

PartlyFull autonomy is opt-in per intent and gated on a confidence threshold: anything below it, or any failed condition, escalates to a human rather than acting.Automation is deployed per intent at the level you choose: none, human-in-the-loop, or fully agentic.aide.app/security ↗

Agent permissions

YesPermissions are enumerated per intent as a Scenario: the team fixes the conditions and the allowed actions, and anything outside escalates instead of firing.Every automation is a Scenario: conditions your team sets, actions your team allows, and a confidence gate it must clear.aide.app ↗

Not established: Multi-agent

Which models it runs on

Model choice

NoThe vendor picks the models and frames that as an advantage; no page offers a buyer any model selection, and the pricing FAQ ties cost control to running their own.Because we run our own models, we can afford to fix the root cause instead of farming the recurrence.aide.app ↗

Not established: Claude, GPT, Open models

Where you use it

In a browser

YesBesides living inside the helpdesk, Aide ships a customer-facing chatbot for your own website, included from the entry Growth plan upward.Automated chatbot on your websiteaide.app/pricing ↗

Whose machine it runs on

Self-hosted

PartlyNot customer-hosted software: the default is Aide's Azure tenancy, with private-cloud provisioning offered only where residency rules force it.Aide can be provisioned in public or private clouds for sovereign AI deployments where data residency requirements demand it.aide.app/security ↗

Not established: Open source

What it costs to run

How it meters

PartlyThe vendor's own pages disagree: the pricing page sells flat monthly platform tiers not metered by seats, while the homepage FAQ describes a platform fee plus a per-resolution rate and an implementation fee.Outcome pricing, done right: a platform fee plus a per-resolution rate that falls as you scale, plus a one-time implementation.aide.app ↗

Free tier

NoNo free tier, trial, or self-serve signup: the cheapest published plan is $3,000/mo and every plan button routes to a demo booking.aide.app/pricing ↗

API access

PartlyAPI access is gated: it is a listed capability of the $15,000/mo Operations tier and Enterprise, and is absent from the Growth and Scale columns of the same comparison table.Operations and Enterprise plans include API access and custom integration support.aide.app/pricing ↗

MCP server

PartlyAide speaks MCP outward as a client, connecting to your backends to gather context; nothing on the site says another agent can query Aide over MCP.Aide connects to your TMS, billing system, ERP, or custom back office via API or MCP, so answers reflect what your systems actually know.aide.app/integrations ↗

Not established: Bring your own key

Buying it for a team

A company can buy it

YesThis is an organisation-only purchase: annual contracts are the default, plans start at $3,000/mo, and pricing is scoped to the partnership rather than to an individual.Pricing is based on the level of technical needs and partnership you need, not seats. Annual contracts are our default — monthly billing is also…aide.app/pricing ↗

Seat model

YesThere is no seat model to negotiate: every published tier includes unlimited seats, so headcount changes never move the bill, and pricing moves on tier instead.Every plan includes unlimited seats.aide.app/pricing ↗

Pooled budget

YesWith unlimited seats on a flat account-level platform fee, the whole organisation draws on one shared entitlement; no per-user balance or allocation exists to divide.Every plan includes unlimited seats.aide.app/pricing ↗

Admin controls

YesConstraining the agent is the core of the product: it runs only team-written procedures, only within the account state configured, and escalates everything else.Aide executes only the procedures your team writes and approves, presents only the options configured for the account state, acts only when it is…aide.app/security ↗

Audit log

YesThe Action Trace logs every automated decision with the intent, data read, Scenario, confidence and outcome, per conversation and across the operation, and is exportable for review.The Action Trace records every automated action: which intent was recognized, which procedure ran, and what was done on the account.aide.app/security ↗

Not established: Single sign-on

What happens to your code

Opt out of training

YesNothing to opt out of: model providers operate under zero-data-retention agreements, so conversations are neither stored by them nor used for training, by default on every plan.Model providers operate under zero-data-retention agreements: the conversations they process are not stored or used to train models.aide.app/security ↗

Data residency

PartlyOne named region only, US East 2 on Azure, which is a problem for EU or UK residency; anything else requires a bespoke private-cloud provisioning conversation.Aide is hosted on Microsoft Azure (US East 2). For sovereign AI deployments with strict data residency requirements, Aide can be provisioned in…aide.app/security ↗

Getting out

PartlyThe audit record is exportable and retained from day one, but no page describes exporting your knowledge, Scenarios or conversation data at contract end, and there are no credits to expire.Filter by intent, Scenario, action type, or outcome, and export the record.aide.app/platform/action-trace ↗

Certifications

YesThree attestations claimed and dated by recurring audit: SOC 2 Type II verified by an independent auditor, plus GDPR and HIPAA, with the report available on request.Aide is SOC 2 Type II certified and GDPR and HIPAA compliant. Every conversation runs on enterprise-grade infrastructure, every action follows…aide.app/security ↗