Halo AI agent pricing in 2026: what it really costs

What Halo AI agent pricing includes

The current Halo pricing page describes an all-inclusive platform licence with AI capabilities, including triaging and suggested replies. It offers named and concurrent licences at the same per-agent price. End-users accessing the portal are not licensed agents.

Onboarding packages are required and priced according to scope. That is a reason to request a line item, not to apply a universal $4,000 estimate. The enterprise offer is a separate option; do not treat its terms as the default contract.

A named licence belongs to a specific agent; concurrent licensing counts simultaneous agent use. Review the relevant working pattern before deciding which count represents your team.

Get a quote that can be checked

This retained Halo product example shows assistance inside a ticket, not a pricing quote. Confirm the features and configuration covered by your Halo agreement.

The public calculator's extracted values did not show a consistent selected configuration during this review. This article therefore does not reuse an older £55 price or claim a 250-agent floor. Check the calculator with your own inputs and obtain the selected terms in writing.

I would ask for the following:

This is a buying checklist, not a list of extra fees Halo necessarily charges. Have the supplier distinguish included work from separately priced work.

For a unit price quoted per month and paid annually, the base licence calculation is licence count × monthly unit price × 12. Add only the other charges that apply to your agreement. Keep currencies separate until you choose and document an exchange-rate assumption.

Separate the first year from later years

An onboarding project should not be silently repeated in every annual estimate. Equally, recurring model usage or internal administration should not disappear after launch. Put one-time costs and recurring costs on separate lines.

If you already run Halo, compare the incremental cost of the native AI workflow with the incremental cost of an alternative. If you are replacing the service desk, include the broader platform decision. Those are different buying situations.

Compare per-agent and task-based costs fairly

The Halo Virtual Agent example illustrates a conversation. An offer to log an incident does not establish completion or a measured cost saving.

A licence buys access to the platform, not just one AI answer. A task-based agent charges for work it performs, but may still require the helpdesk you already pay for. Neither model is automatically cheaper.

Use the same scope for both sides: the same request types, knowledge, necessary actions, review process, and integrations. Compare the cost of a useful outcome, including human review and corrections, rather than assuming every AI-handled request saves the same amount of time.

For eesel's standard pay-as-you-go plan, the pricing guide lists regular support/chat tasks at $0.40 each, no platform fee, and no monthly minimum. Tasks are billed regardless of outcome. Enterprise adds a platform fee, so specify the plan when comparing.

As a hypothetical usage-only example, 500 regular tasks at $0.40 would cost $200. That is not a quote for replacing Halo: it excludes any existing helpdesk licence, integration work, other task types, and the team's review time. It also does not predict how many tasks your workflow will generate.

Review an alternative AI budget through eesel CLI

The eesel CLI works with the same eesel agent and workspace as the dashboard. A person can use it from a terminal, a script can read its JSON output, or Claude Code, Cursor, or Codex can help collect the evidence. For a pricing review, start with inspection rather than generating additional work.

Identify the workspace and current workload

With Node.js 18.17 or newer:

npx @eesel/cli login
npx @eesel/cli whoami
npx @eesel/cli agents
npx @eesel/cli billing
npx @eesel/cli automations --agent TEST_AGENT_ID
npx @eesel/cli activity --agent TEST_AGENT_ID
npx @eesel/cli status --agent TEST_AGENT_ID

Replace TEST_AGENT_ID with the selected agent. Billing describes the workspace; do not attribute the whole amount to one agent merely because the activity command was scoped. Check the dates, included agents, and task types before comparing totals.

These inspection commands are different from chat: the CLI guide identifies chat as billed work. A command that reads billing does not estimate future usage, update a cap, or validate a Halo connection.

Ask a coding agent to produce a bounded cost review

This eesel screenshot uses a Zendesk workspace, not Halo. It illustrates activity inspection, not an invoice breakdown or the results of the commands in this article.

A useful first request is:

Read the selected eesel workspace's billing state and the selected agent's automations and activity. Identify the reporting period and what each total includes. Separate observed usage from assumptions. Do not run chats, change billing, create automations, or connect systems. Report anything you cannot attribute or verify.

The coding agent can collect JSON results and prepare a comparison. Have it flag missing information instead of filling gaps with a guessed task count. For example, an activity list may not cover the full billing period or establish which invoice line belongs to a particular workflow.

Define a small pilot before forecasting savings

Choose one service-desk request type and an approved procedure. Check the selected agent's instructions and permissions first:

npx @eesel/cli instructions --agent TEST_AGENT_ID
npx @eesel/cli integrations --agent TEST_AGENT_ID

Existing automations may already be active. Keep account-changing actions and outgoing messages off for a fictional policy test. If the approved procedure is available, an example is:

npx @eesel/cli new --name "service-desk-budget-pilot" --agent TEST_AGENT_ID
npx @eesel/cli chat "Fictional evaluation: an employee cannot connect to the VPN. Use the approved procedure to propose the next step, identify missing information, and say when a human should take over. Do not contact anyone or change accounts." --agent TEST_AGENT_ID

This is a billed chat test, not a free historical simulation. A new conversation does not disable the agent's tools. Check the answer against the source and record the review effort; a correct-looking answer is not proof of a resolved employee request.

Before estimating a monthly volume, define which real events would start work, which requests are excluded, and how repeat or escalated cases are handled. Do not enable a broad automation just to obtain a sample bill.

Review spending controls and integration costs

Open Workspace settings > Billing & Usage to inspect the monthly cap. The pricing guide says a card is needed before the limit becomes editable. When the limit is reached, new work stops while work already in progress finishes. Decide how your human team handles requests during a pause.

Treat Halo connectivity as a separate requirement. The eesel documentation index checked for this guide does not list a built-in Halo connector. General Network Access may be relevant to a custom API design, but authentication, permissions, ticket events, and write behavior must be verified for your deployment.

Include the effort to build and maintain that connection if it is required. A successful billing inspection does not establish that eesel can read or update your Halo tickets, and a domain allowlist is not a read-only permission guarantee.

Make the price comparison match the decision

Ask Halo for a quote with the selected licence and onboarding terms. For eesel, use the CLI to inspect actual work and define the assumptions behind a pilot budget. Keep platform replacement, AI task usage, and integration work distinct.

Try eesel with one approved service-desk scenario before projecting savings across the queue.