Workflow test

Seamless.AI Single-Profile Test: A Saved Lead and Three Invalid Email Candidates

Workflow test completed · Free account · one self-profile · two routes

Sample / task scope: One tester-owned profile identity, two lookup routes.

Published 2026-10-23 · Report updated 2026-10-26 · Tested by Jimmy Dwyer

Jimmy Dwyer · Publisher and responsible editor
Jimmy runs StackVouch and is responsible for this article's conclusions, disclosures and corrections. About Jimmy · LinkedIn

We may earn a commission if you buy through links on this page. Affiliate disclosure.

No paid sponsorship. Seamless.AI did not review this article before publication.

Read the Seamless.AI buying review →

The result at a glance

Observed resultBuying implication
Name/company search offered prediction after no existing matchThe supplied name and company are not proof of a database identity match
One record saved and reopened in a named listStorage worked; a saved record does not establish usable contact data
First Raw CSV: 1 row, 183 columns, 4,825 bytesThree candidates were invalid/0%; Primary Email and all phone fields empty
URL enrichment: 1 unknown contact, no matching result rowIts 1-row, 156-column, 3,859-byte Raw file preserved only the input URL
Repeated First Name and Last Name headers in the first Raw CSVPowerShell 7.6.5 Import-Csv rejection reproduced October 25; Python 3.12.14 csv.reader retained positional fields
URL-job balance displayed 60 → 59 → 60Return mechanism unconfirmed; chatbot forwarded a policy inquiry to support

The findings apply only to the dated account workflow and stated conditions. See the full limits section before generalizing them.

On this page

Purpose and test conditions

Could Seamless find usable contact information for one known person and preserve the result through a list and CSV export? Jimmy Dwyer registered the account himself on October 22, 2026. The fixed target was Jimmy at StackVouch. This route supplied his name and company through the search filters; it did not submit his LinkedIn profile URL.

The header described the account as Free. Billing showed no current subscription or payment method, while its license rows used Universal (Free Trial) and Intent (Free Trial). Those labels do not establish permanent free-plan entitlements. We used the existing allowance, connected no new mailbox or CRM, bought nothing and sent no email or other outreach.

This is a one-person workflow check. It does not estimate Seamless’s database coverage, accuracy or deliverability. For product fit and alternatives, read our Seamless.AI review.

Search offered prediction after no existing match

We retained Company = StackVouch, added Full Name = Jimmy Dwyer and submitted that query. Seamless stated that it could not locate the contact in search and offered to predict or identify available contact information. The resulting row repeated the supplied name and company with a Find button.

The blurred fields before Find were placeholders, not verified results. We ran Find once on this one target. After Researching completed, My Contacts contained one Jimmy Dwyer record associated with StackVouch. The table offered Add Email and Add Phone rather than showing a usable address or number. A company website appeared, but an individual LinkedIn profile match was not established.

The search repeats Jimmy and StackVouch but explicitly says the contact could not be located before offering Find
Name/company search could not locate the contact and offered a prediction-based Find

One saved record reached a named list and CSV

We created a test list, added the saved record and reopened the list. Its entry showed one contact, and the filtered contact view retained Jimmy and StackVouch. This verifies the observed navigation and persistence through those views; it is not a duplicate-import or long-term retention test.

The named test list shows one saved contact owned by Jimmy Dwyer
Named test list contained one saved contact

Download Filtered offered Cleaned, Raw and Custom output types. We chose Raw CSV for the one-list filter so we could inspect the fields behind the table. The export completed and appeared in Downloads with one contact. We checked the downloaded file itself.

The CSV contained one data row across 183 columns. Name, company, website/domain and list tags were present. Primary Email, the individual LinkedIn profile field and all phone fields were empty.

Three candidate emails appeared in Email 1–3. Each carried Seamless’s own invalid label and a 0% Total AI value. We did not send to these candidates or independently verify their status. The Raw file therefore contains more detail than the table, but none of these candidates counts as a usable or verified email result.

The export also repeated the First Name and Last Name headers. On October 25, 2026, we reproduced the first Raw file’s duplicate-header rejection using PowerShell 7.6.5 on Windows: Import-Csv reported The member “First Name” is already present. Python 3.12.14’s csv.reader read the same one-row, 183-column file positionally and retained both occurrences. These are named reader behaviors, not a claim that every CSV importer rejects the export. Buyers using a fixed import schema should inspect Raw output before mapping it to another tool. This does not establish how every CSV reader handles the file.

The LinkedIn-URL route returned one unknown contact

On October 23 we tested the separate enrichment route using a single-row CSV containing only Jimmy’s own LinkedIn profile URL. The form accepted that identifier, displayed one credit per row and kept Add to My Contacts selected. We submitted the job once.

The task reached 100% and reported one unknown contact. Opening its result filter showed no matching contacts. The existing My Contacts count still referred to the previous day’s name/company record; it did not establish that the URL had matched.

The completed URL enrichment task shows one unknown contact and a final displayed balance of 60
Completed own-URL enrichment reported one unknown contact; displayed balance 60

Raw CSV from this task contained one row across 156 columns and was 3,859 bytes. Only the original input URL was populated; the other 155 fields were empty. The file preserved our input without returning an identified person or contact details. We did not submit a second enrichment request or change the sample to obtain a positive result.

Credits, onboarding rewards and price context

The account showed 55 available credits before the controlled Find and 60 afterward, including after export. Setup listed 50 earned for account creation and five earned for finding a first lead, alongside other reward tasks. These notices show onboarding incentives, but they do not fully reconcile the displayed balance or establish the gross cost of the Find. My Usage also reported an error loading lead stats, so its displayed zero counters were not reliable evidence of the operation count.

The separate URL job showed a balance of 60 before submission, 59 immediately afterward and 60 when we opened the completed result. That is the observed sequence. Without an itemized credit ledger, we cannot confirm a refund mechanism or general failed-match charging rule.

We requested clarification on October 23 about this job, the written unmatched-row rule and an itemized credit ledger. The in-app chatbot confirmed forwarding the question to support for an email reply. As of October 25, 2026, Jimmy confirmed he had received no human reply to the inquiry forwarded on October 23. The written unmatched-row rule and itemized accounting remain unconfirmed. This is two calendar days, not a measured business-day support-response result.

Seamless’s credit and license documentation, checked October 22, describes a 50-credit Free allocation without replenishment and feature-specific Universal Credit costs. Our account balance should not be substituted for that stated base allocation. A successful request, failed request, reward and refund would need separate accounting evidence to establish net pricing behavior.

The public pricing route redirected to a pricing inquiry form during the same day’s browser check. It did not provide numeric paid-plan prices. We did not submit the form or obtain a quote; paid pricing remains a custom quote rather than a verified starting dollar price.

What the findings change in our review

We can now support a narrow first-hand claim: Seamless processed one prediction-based name/company Find, saved its record and exported a one-contact Raw CSV. The result makes the distinction between a saved lead and usable contact data important. In this sample, the table had no usable contact details and the export exposed three candidates that Seamless itself labeled invalid. A separate URL-only enrichment returned one unknown contact and an input-only output file.

Buyers should check the contact-status fields in their own permitted sample and confirm how their target system handles Raw headers. This single case supports workflow and output observations; it does not support a general accuracy rating or a claim that other targets will return the same result.

Risks and limits

We did not test sending, deliverability, phone accuracy, CRM transfer, mailbox connection, API/MCP execution, team access, duplicate handling, paid-plan enforcement or contractual terms. No alternative person was researched to replace this result. Different inputs and output schemas distinguish the two tested routes; they are not two independent people or a representative sample.

Re-test and correction history

  • October 22, 2026: Original single-profile name/company test. One Find, one saved record, one named list and one Raw CSV export. Invalid candidate labels, duplicate headers, onboarding balance changes and a usage-report error documented. No outreach or purchase.
  • October 23, 2026: Second route: URL-only enrichment of the same person. One completed job reported one unknown contact; its Raw CSV preserved only the input URL. Displayed balance sequence 60 → 59 → 60 recorded without inferring a refund rule. No outreach, purchase or sample expansion.
  • October 23, 2026, support follow-up: Credit-accounting inquiry forwarded to the support team through the in-app chatbot. No human answer or written return policy confirmed; no additional product test performed.
  • October 25, 2026: Clarified the second route, reproduced CSV reader behavior with named versions, recorded Jimmy’s confirmation that no human support reply had arrived, and limited the buying verdict to product terms and workflow fit. No additional lookup or enrichment request.

CSV reader reproduction

On October 25, 2026, we reproduced the first Raw file’s duplicate-header rejection using PowerShell 7.6.5 on Windows: Import-Csv reported The member "First Name" is already present. Python 3.12.14’s csv.reader read the same one-row, 183-column file positionally and retained both occurrences. These are named reader behaviors, not a claim that every CSV importer rejects the export.

This minimal example reproduces the duplicate-name behavior; it is not the full Seamless export:

First Name,Last Name,First Name,Last Name
A,B,A,B
# PowerShell 7.6.5, Windows, October 25, 2026
Import-Csv -LiteralPath ./duplicate-name-headers.csv
# The member "First Name" is already present.

Original Raw file SHA256: dde72f536b13c6e963d92317dd9e2893deb39f3cb4dff47fc139792290086369. The original file remains private; no candidate address is supplied as a verified contact.