The H-1B program DOL administers is attestation-based, not verified. When a company files a Labor Condition Application, it attests that hiring the foreign worker won’t harm similarly-employed U.S. workers — but DOL doesn’t check that claim for most filers. Real scrutiny (recruitment documentation, displacement questions) only applies to the narrow category of “H-1B-dependent” employers. Everyone else’s compliance is the filing itself. Compare that to the PERM green-card process, which does require documented American recruitment before a foreign worker can be sponsored permanently. The gap between those two programs is exactly the gap this report is about.

And the idea of the government requiring domestic labor with its own money isn’t novel or radical — it already exists elsewhere in federal spending. The Buy American Act prefers domestic materials on federal purchases. Davis-Bacon rules set wage floors for federal construction labor. The Jones Act requires domestic vessels and crews for maritime shipping between U.S. ports. The federal government routinely decides taxpayer money should preferentially build American jobs — just not, currently, for the IT-staffing and professional-services dollars this report is about.

Rather than pick one number, we built three, each narrower than the last. All three start from the same base: every DOL contractor in a staffing or IT-services spending category (NAICS 561320, 561311, 561330, 541511, 541512, 541519, 541990) since FY2016, cross-matched by company name against DOL’s own H-1B Labor Condition Application disclosure data, 2008–2026, then restricted to certified petitions at DC/MD/VA worksites filed 2020–2026 — the years and geography that plausibly overlap with these contracts.

The 91 vendors behind the 1,707 figure aren’t evenly distributed — a dozen companies account for two-thirds of it. These are the companies with the largest confirmed DC/MD/VA H-1B footprint among DOL’s own staffing and IT-services contractors.

This is a company-name cross-reference against federal spending and federal disclosure data, not a contract-by-contract audit. Be clear-eyed about the limits:

The Department of Labor exists, by its own founding statute, to “foster, promote, and develop the welfare of the wage earners, job seekers, and retirees of the United States” and “advance opportunities for profitable employment.” It has also spent $4.65 billion of taxpayer money since FY2016 on staffing and IT-services contracts, and by even the most conservative count we can defend, at least 1,244 certified H-1B positions sit inside the Washington-area workforce of the companies holding those contracts — with no requirement anywhere in the process that any of it go to an American worker first.

The federal government already accepts the principle that its own money should build American jobs: the Buy American Act does it for materials, Davis-Bacon does it for construction wages, the Jones Act does it for shipping. A poor West African nation applied the same principle to a foreign oil-services contractor in 1975 without a second thought. The question this report leaves open isn’t whether that kind of requirement is possible — it plainly is, elsewhere in the same federal government. It’s why the Department of Labor’s own contracts don’t have one.

The Office of the Attorney General — the entire subject of the original report — is Bansar’s 7th-largest customer. Health and Human Services Commission alone is 4.5× the size of the OAG relationship, spread across 214 separate staff-augmentation contracts since 2015.

Bansar’s Texas business roughly quadrupled between 2020 and 2021 and never receded — it has held at $20–26 million in new awards every year since, across an expanding roster of agencies. The jump lines up with the pandemic-era staffing surge at the state’s two largest human-services agencies, HHSC and DFPS, which together account for 40% of Bansar’s all-time total.

The state’s own procurement-method field labels 527 of Bansar’s 534 contracts (98.7%) “Competitive.” Taken at face value, that looks like a clean record. It isn’t measuring what it sounds like it’s measuring.

We obtained three consecutive Texas Alcoholic Beverage Commission purchase orders for a single Bansar contractor, Raja Konda, QA Manager on TABC’s IT modernization project: PO 24-40268 ($146,300, term 12/31/23–8/31/24), PO 25-40078 ($224,000, term 9/1/24–8/31/25), and PO 26-40053 ($224,000, term 9/1/25–8/31/26) — $594,300 total, three different contract numbers, zero-day gaps between each expiration and the next award. Each renewal was justified by a one-page “Best Value Statement” from TABC’s IT PMO manager arguing Konda specifically should be kept on: “he is the best person to perform these key and impactful duties.” TABC’s own internal procurement checklist — requiring three competing DIR vendor bids for purchases this size — has its verification boxes left unchecked on all three POs.

The “Competitive” label in the state’s contracts data refers to the one-time solicitation that established Bansar’s master DIR staffing contract (DIR-CPO-4534, renewed in 2024 as DIR-CPO-5452) — not to any individual agency’s decision to keep the same named person in the same seat, year after year. Every Bansar renewal we could check followed this shape, including one that names Konda directly in the statewide contracts registry itself: line item “ITSAC Bansar Contractor_ R. Konda,” recorded twice under TABC.

We ran the identical statewide pull for every vendor from the original report. In every case but one, the OAG relationship was a small fraction of the vendor’s real Texas government business — confirming this is a structural feature of how Texas buys IT staffing, not a Bansar-specific story.

This addendum draws on the same Texas Comptroller/DIR statewide contracts registry that fed the original report (confirmed: our pull reproduces the original report’s OAG figures for 10 of the 14 vendors within a percent or two). “Current Contract Value” in that registry is a not-to-exceed ceiling set at award, the same caveat that applies to LCA “worker positions” figures — it isn’t confirmed spend. We cross-checked it against the Comptroller’s actual payment-disbursement records: Bansar alone was paid $10.75 million in real cash across 22 agencies in just the first six months of 2026, so the ceiling figures in this piece are not sitting unused. In the course of this analysis we found and corrected one likely duplicate contract record (Red Salsa/OAG, detailed above) and flag Allied Consultants’ bare “ALLIED CONSULTANTS” vendor-name variant as verified — same underlying vendor ID as “Allied Consultants Inc” across all name-string variants, not a name collision.

The original report’s number was real, but it was a keyhole view. Bansar Technologies’ relationship with the Texas Attorney General’s office isn’t an outlier arrangement between one small Austin vendor and one agency — it’s the smallest visible slice of a $141 million, 23-agency, decade-plus staffing operation that has quadrupled since 2020 and shows no sign of slowing. And the mechanism that keeps it running — a single competitively-awarded master contract, followed by an unbroken chain of sole-source “best value” renewals for specific named individuals, each one issued a fresh contract number that erases the chain from public view — isn’t specific to Bansar either. Twelve of the other thirteen confirmed H-1B-sponsor vendors from the original report show the same pattern, just at different scales. Texas’s own procurement transparency data can tell you how much money moved and to whom. It cannot, on its own, tell you how many times the same person was renewed into the same job without ever being put back up for competition — that took the underlying agency paperwork to see.

DOL’s LCA disclosure files identify employers by a free-text name field with no employer ID, no deduplication, and no standardization pass. We normalized every employer name in the archive as conservatively as possible — uppercase, strip commas and periods, collapse whitespace, nothing fancier — and 24,703 of 188,152 “distinct” employer names collapsed into a name already used by another row. That’s a floor, not a ceiling: this check doesn’t touch abbreviations, “US” vs “U.S.A.”, or legal-suffix variants like “LLC” vs “L.L.C.”, all of which would fragment real companies further.

Every LCA asks the employer to attest whether it has previously been found a “willful violator” under the H-1B program — a designation that, once made, is supposed to be disclosed on every filing for five years. In the full archive, that field holds six different raw values for what should be a single yes/no answer (Yes, Y, No, N, N/A, and blank), and only 1,142 of 3,831,829 filings (0.03%) are marked positive at all. Tata Consultancy Services’ own filing history shows why that number is meaningless as a signal rather than reassuring:

DOL’s public H-1B debarred/disqualified employer list currently names four companies as confirmed willful violators. We checked each one’s own historical LCA filings for the self-attestation the form is supposed to require of them.

This is the same dataset every H-1B search tool on this site — and most of the H-1B journalism published anywhere — is built on. It has no employer ID, no name standardization, and at least one self-attestation field that’s essentially decorative. The practical consequence is that any “who are the top H-1B sponsors” ranking, including ones built from this exact archive, is undercounting real concentration by splitting single companies across multiple name entries, and any claim that leans on the willful-violator field as evidence — in either direction — is standing on a field nobody appears to be checking. None of this requires a conspiracy; it’s what happens when a federal disclosure system accepts free-text employer names from thousands of different law firms and staffing companies with no validation layer in between. The debarment list itself, by contrast, is small, curated, and appears reliable precisely because it’s DOL’s own enforcement output rather than employer self-report.