How we score AI job agents
Through Route scores each AI job agent on six criteria, from 0 to 5, using only what the tool's own pricing, help and terms pages say, then weights them to a total out of 100. Sending applications counts most, at 25 points, then tailoring at 20. The home page picks go by use case, so a lower total can still win your route.
What did we check, and what did we not do?
On November 8, 2026 we read the public pricing page, help center articles, feature pages and terms of 13 tools. Every price, limit and feature on this site links to the page it came from, listed on Sources. Facts about Dreamwork (our publisher) come from its own product pages on dreamworkhq.com.
We did not sign up and test the tools, survey users, interview anyone, or measure interview or offer rates. We have no success rates to report, and we do not quote reviews. When a vendor's page did not load or did not state something, we left the fact out. Sonara's prices are an example: its pricing page did not load directly, so we list none.
What are the six criteria?
| Criterion | The question | Weight |
|---|---|---|
| Sends applications for you | Does it submit the application on its own, or do you still press the button? | 25 |
| Tailors each application | Does each application get its own resume, cover letter and answers? | 20 |
| You stay in control | Can you see and stop what goes out under your name? | 15 |
| Price and terms in the open | Can you see what it costs, and how billing and refunds work, before you sign up? | 15 |
| Reach | How many kinds of jobs and places does it say it covers, and does it name its sources? | 15 |
| Tracking | Can you see every application afterward, and the replies? | 10 |
Total = the sum of (score ÷ 5 × weight), rounded to a whole number. A tool that scores 5 everywhere gets 100.
Each criterion below lists the levels we award. A score not listed for a criterion is not used.
Sends applications for you, 25 points
Does it submit the application on its own, or do you still press the button?
- 5. Submits on its own and also offers a mode where you approve before it sends.
- 4. Submits on its own. No review-first mode is described.
- 3. Submits on its own, but only on one application system or under a small weekly cap.
- 2. Fills in the form. You submit.
- 0. Does not fill in or submit applications.
Tailors each application, 20 points
Does each application get its own resume, cover letter and answers?
- 5. Tailored resume, cover letter and answers to the employer's questions for each job, on the plan that applies.
- 4. Tailored resume and cover letter for each job.
- 3. Tailoring only on a higher tier or as a paid add-on.
- 2. AI writing tools, but you do the tailoring yourself.
- 1. A cover letter tool only.
- 0. No tailoring described.
You stay in control, 15 points
Can you see and stop what goes out under your name?
- 5. Nothing is sent without you, or an automatic mode comes with a review-first option and a match threshold or pause.
- 4. You approve each application, or preview materials, or set exclusions.
- 3. You choose which jobs or searches it runs on.
- 1. No controls are described.
Price and terms in the open, 15 points
Can you see what it costs, and how billing and refunds work, before you sign up?
- 5. Every plan's price is published, short billing periods exist, and a refund policy is stated.
- 4. Every plan's price is published with short billing periods. No refund policy found, or a trial needs a card.
- 3. Prices published, but only for long commitments such as a year.
- 2. Only a starting price such as "from $X".
- 1. No confirmed prices, but terms are published.
- 0. No prices published.
Reach, 15 points
How many kinds of jobs and places does it say it covers, and does it name its sources?
- 5. Says it works in every country and names its job sources.
- 4. Broad reach, and names its sources or application systems.
- 3. Broad but unnamed sources, or named sources with a narrow focus.
- 2. One country and one application system, or sources not named at all.
Tracking, 10 points
Can you see every application afterward, and the replies?
- 5. A tracker plus email, inbox or contact tools.
- 4. A tracker or dashboard.
- 3. You can view what was submitted.
- 1. No tracking described.
Why does a missing fact score low?
If a vendor does not publish a price or describe a review step, a job seeker cannot know it exists before paying. We score what you can check. Jobright and Massive score 0 on price for that reason, not because their prices are high. If a vendor publishes the missing fact, we re-read the page and change the score here and in the changelog.
How do we choose a route winner?
A route is a situation a job seeker is in, such as applying from outside the US or wanting to approve every application. For each route we pick the tool whose own pages show the capability that situation needs, then use the rubric to break ties. We wrote 12 routes, and Dreamwork takes two of them.
Does Dreamwork owning this site bias the scores?
Dreamwork publishes this site and makes one of the tools we score. We wrote the rubric too, and Dreamwork scores highest on it. The weights favor tools that send and tailor applications, because that is the job of an AI job agent. If you care more about price in the open or about reach outside tech, look at those columns on their own: AIApply, Simplify and Teal score higher than Dreamwork there.
Changelog
- . First edition. Read pricing, help and terms pages for 13 tools. Confirmed Sonara still operates under BOLD LLC's terms. Noted that Wonsulting folded AutoApplyAI into JobBoardAI, that Massive's main site now leads with AI training data, and that Jobright's pricing address returned a 404 error.
- . Renamed the publication to Through Route. Added Dreamwork's published refund and review terms; its price score stays at 4 because the trial asks for a card.
- . Moved the ownership disclosure from the top of each page to the footer, with a short "(our publisher)" note beside Dreamwork's name where we recommend it. Added pages for Huntr and LoopCV after re-reading their pricing and product pages. Raised LoopCV's tailoring score from 0 to 1: its auto-apply page says Premium drafts tailored answers to job-specific questions.