Hourtick Review 2026: AI Tasks, Time Tracking, and Billing Controls

Hourtick is a candidate for small teams that want tasks, time records and AI-agent work in one place. Its important evaluation question is whether a manager can reconcile useful work with recorded time and cost. This is a research-based review; we have not tested the system and ratings are withheld.

The homepage presents task boards, team chat, time tracking, approval and reporting. It describes suggested missing time entries based on activity within Hourtick, with user approval before logging. These are vendor descriptions rather than verified synchronization or accounting results.

The agent page distinguishes an assistant acting as the connected user from a separately identified workspace agent. It describes MCP connections and agent time and cost reporting. Hourtick says it does not run or resell the models; AI-provider charges remain separate.

The pricing page lists unlimited people and features on Free, with 500 MB of monthly workspace usage. Pro is $29 monthly for 5 GB, with additional GB at $10 monthly each. Usage includes stored data and traffic sent and received, not simply uploaded-file size.

The homepage's at-a-glance panel gives a conflicting free-usage impression. Use the detailed pricing terms and confirm the allowance in your account. Do not equate an included agent connection with free model execution. Check provider invoices and any external service costs separately.

Our editorial view is that model spend, elapsed work time and client-approved billable time belong in separate columns. An agent taking longer does not automatically create more client value. A logged entry does not establish that a customer agreed to pay for it.

Write an internal rule for interrupted tasks, failed attempts and human correction. Decide who approves each entry and which evidence should accompany it. Agree on the treatment of AI work with clients before using those entries in an invoice.

Create a disposable project with one human and one agent task. Use invented project details. Track a manual baseline, complete a short agent-assisted job and reconcile the task result, recorded time and provider spend.

Test a corrected entry, a rejected entry and an approval. Export the report and compare totals with your manual record. Try a brief offline interruption with non-critical data and inspect the final state after reconnection. We have not performed these tests, so no synchronization reliability claim is made here.

For agent access, test the difference between a personal connection and a dedicated agent identity. Check that you understand who an action is attributed to, what it may change and how its credentials are revoked. Stop a disposable session and inspect what work remained in progress.

Review current retention, permissions and processing terms before importing client conversations or files. This review does not certify Hourtick's privacy or security posture. Keep the initial exercise non-sensitive and request the agreements your organization requires.

Migrate only a small project first. Reconcile exports, historical entries and approval states before moving the full archive. Keep the original records available until the new routine has passed your own acceptance checks.

Compare [ClickUp Brain](/tools/clickup-brain) when broader project coordination is the priority and [Notion AI](/tools/notion-ai) when documentation and knowledge work dominate. These are workflow comparisons rather than proof of feature equivalence.

Adopt Hourtick when reconciliation and approval are simpler in practice, the usage allowance is adequate and the total cost is clear. Our [Gemini agent pilot guide](/articles/gemini-work-agent-governed-rollout-2026) offers complementary questions for deciding how much authority an agent should receive.