Skip to content
Zarif Automates

6-page PDF · about 30 minutes

AI Tool Picker

Choose a sensible first AI assistant for your task and budget. Compare practical buying criteria and choose one sensible first AI assistant without tool sprawl.

Sample page

Page 2 of 6 from the actual PDF, shown in full so you can try it before you sign up.

STEP 1 · START WITH THE JOB

Use this decision tree before a feature table

A product can be excellent and still be wrong for your task, information, organization, or workflow.

Follow the branches in order

  1. 1Name the job. Write one representative output and audience. For coding, creative production, or multi-step automation, use a task-specific comparison instead of this general-assistant shortlist.
  2. 2Apply the data gate. If the material is not approved for a candidate tool, stop and use the approved assistant or obtain policy guidance. Output quality never overrides this branch.
  3. 3Need current sources?. If claim-to-source checking is central, include Perplexity or a source-grounded research mode. If not, continue with the work-shape branch.
  4. 4Does the workspace decide?. If the job must stay inside Google Workspace or another approved suite, test its native assistant first. Otherwise continue.
  5. 5Match the work shape. Long supplied documents or careful synthesis → include Claude. Broad text, files, images, voice, or flexible everyday work → include ChatGPT, Claude, or Gemini.
  6. 6Apply the budget gate. If a free tier supports the representative test, start there. If paid capacity is required, set the maximum monthly seat cost before comparing plans.
  7. 7Keep two. Choose at most two policy-approved candidates and take the same input, prompt, time limit, and review rule into the bake-off.

A valid endpoint: use the approved tool

If your organization has approved one assistant for the data and task, start there. Do not move sensitive work into another product just to compare output quality.

The full 6-page PDF opens on this site right after signup. Get the full PDF.

What is inside

  • A job-first decision path for choosing what to test
  • A dated, plain-language assistant matrix and first-party verification links
  • A two-tool bake-off, policy checklist, and dated decision record

Safety notes

  • Verify current pricing, limits, data terms, and integrations on first-party pages.
  • Use the same non-sensitive task and scoring rule for both tools.
  • A policy or privacy deal-breaker cannot be averaged away by a higher feature score.