Only 160 federal AI jobs posted, yet spending hits $7.2B

The Bet: Earlier Signal, Shared Substrate

Pryzm has secured FedRAMP High authorization and a $12.2 million seed round led by Andreessen Horowitz, PR Newswire reported, to turn its federal procurement CRM into an AI operating system that detects opportunities before solicitations go public and maps the relationships that decide who wins. Founded by alumni of Palantir and Lockheed Martin, the company converts public data — contracts, congressional hearings, budget documents, news reports — into a personalized dashboard that shows unallocated funding amounts, the government officials responsible for allocating it, and the historical patterns behind previously awarded contracts. CEO Nick LaRovere describes the output as a "highly targeted heat map or signal into where opportunities may be and who you need to work with." The goal: let contractors engage when influence is still possible.

DoD Impact Level 5 clearance, secured alongside FedRAMP High in November 2025, lets Pryzm host Federal Contract Information and Controlled Unclassified Information on the same substrate where its AI models run. That compliance milestone unlocks end-to-end procurement workflows that move from opportunity detection through proposal development without leaving the authorized environment. The Defense Innovation Unit selected Pryzm in March 2025 for its Enterprise Workflow and Reporting Platform effort, using the company's funding analysis and project management workflows to replace manual budget tracking across technology portfolios. Early deployments with Forterra, HII, Vannevar Labs, and REGENT Craft are the proving ground for the broader rollout.

With the seed capital, Pryzm plans to expand engineering and go-to-market teams across Washington, D.C., Boston, and New York while deepening those deployments. The Arlington, Virginia office opened in 2025 serves as the federal-facing hub for that work.

What distinguishes this push from the incumbent CRM layer is the bidirectional design. Contractors get early signal; agencies get a digital thread of every defense dollar that streamlines decision-making. Both sides operate on the same authorized substrate. If the model holds, the procurement timeline compresses not because paperwork moves faster, but because the right conversations start months earlier.

How Contractors Work Today: And Where the Gap Lives

The federal contracting workflow has long been defined by a brutal asymmetry: the government posts thousands of opportunities, but the real decisions — who gets funded, which requirements get shaped, which program managers hold the pen — happen months before a solicitation ever goes public. "By the time something is publicly posted it's already too late," LaRovere told TechCrunch in August 2024. Startups and mid-tier contractors spend hours each week sifting through SAM.gov, agency forecast pages, congressional hearing transcripts, and press releases, trying to reverse-engineer where unallocated money sits and who controls it. The manual effort is not trivial; it is a full-time job for capture teams at companies that can afford them, and a fatal distraction for those that cannot.

Pryzm's platform attacks this directly. Its AI ingests contracts, congressional hearings, budget justification books, and news reports, then fuses that public corpus with each customer's proprietary data (past proposals, teaming agreements, contact maps) to produce a personalized dashboard. The dashboard surfaces the amount of currently unallocated money, those officials, and details on previously won contracts so contractors can see how competitors secured similar work. LaRovere reiterated that the platform surfaces opportunities and key contacts. That heat map replaces weeks of analyst time with a query that returns in seconds.

The practical shift shows up in two dimensions: earlier entry and sharper targeting. Contractors using Pryzm can identify opportunities at the "pre-RFP" stage — when requirements are still being written and program managers are looking for industry input — rather than reacting to a finished solicitation. That early window is where the "soft power side of contracting" lives: shaping statements of work, influencing evaluation criteria, building relationships with the decision-makers who will eventually score the bids. Pryzm's relationship-mapping capability shows not just who holds a title, but who has historically championed what technologies, which program offices have execution authority versus advisory roles, and where a contractor's existing network overlaps with the buying chain. For a startup that has never sold to the DoD, that map is the difference between a cold call and a warm introduction.

FedRAMP High authorization means these contractors can now load such data directly into the platform, enabling what co-founder Justin Deckert called "true end-to-end procurement automation for the first time." The authorization eliminates the air-gap workaround where sensitive capture data lived in separate, unsearchable systems. Now a capture lead can trace a requirement from a congressional mark through a program element number to a specific program manager's inbox without leaving the platform.

The win-rate implication is structural. In a market where incumbents win because they know the unwritten rules, such as which contracting officer prefers oral presentations and which technical evaluator penalizes certain architectures, Pryzm codifies that tribal knowledge. It does not guarantee a win, but it removes the information penalty that has historically favored primes with decades of institutional memory. For the defense industrial base, that penalty has been a tax on innovation: every hour spent hunting for the right program manager is an hour not spent engineering the solution. Pryzm's bet is that removing the tax changes who shows up to compete.

Rivals Move: Palantir Scales, Deltek Expands, Accenture Grows Federal Presence

Palantir's Artificial Intelligence Platform, launched in April 2023, represents the most direct competitive signal in the federal AI race. Built to securely integrate large language models with classified and sensitive data, AIP arrived well before Pryzm's recent funding, but its adoption curve has steepened precisely as the market Pryzm targets has heated up. On the latest earnings call, Chief Financial Officer Dave Glazer said total contract value of bookings in the quarter soared 185 percent to $1.1 billion, with U.S. commercial revenue jumping 93 percent year over year. "AIP continues to drive existing customer expansion and new customer conversions in the U.S.," Glazer said. Government revenue still leads at 53.7 percent of the 2024 total, but the commercial flywheel is spinning faster.

The Pentagon has backed that momentum with contract authority. In May 2025 the Department of Defense lifted the ceiling on Palantir's Maven Smart Systems contract for AI capabilities to $1.3 billion. NATO acquired the Maven Smart System in March 2025 and reached full technical operational capability three months later. Admiral Pierre Vandier, NATO's supreme allied commander transformation, said there is currently no competitor for Palantir and that European alternatives would need to prove themselves in months or years, not a decade. Palantir is also under contract to deliver 10 TITAN mobile ground stations, AI-equipped vehicles that fuse space-sensor data for field intelligence, to the U.S. Army.

A November 2024 partnership with Anthropic and Amazon Web Services brought Claude models into Palantir's government environments, letting unspecified intelligence and defense agencies run the models inside Palantir's IL5-authorized SaaS. The same announcement noted a broader trend: Meta offering Llama to defense partners and OpenAI pursuing closer Defense Department ties. Palantir reported one unnamed American insurer used 78 AI agents powered by the platform and Claude to shrink an underwriting process from two weeks to three hours, a commercial proof point the company cites to government buyers.

Deltek, the Herndon, Virginia-based ERP and project-lifecycle vendor founded in 1983, continues to expand its GovWin intelligence platform. The company's private-equity ownership (Roper Technologies acquired it in 2016) means its AI investments are measured against retention and upsell in a captive base.

Accenture Federal Services has grown its federal AI presence. The contrast is instructive. Palantir sells a horizontal decision layer (Gotham for analysts, Foundry for enterprises, AIP for model orchestration) that sits atop existing data. Deltek embeds AI inside the ERP and capture-management workflows where contractors already run their businesses. Pryzm sits between them: a purpose-built CRM for national-security business development that promises generative opportunity detection and relationship mapping without the implementation weight of an ERP or the platform breadth of Palantir.

Palantir's valuation — 226 times forward earnings, over 80 times forward revenue — reflects market conviction that its government moat and commercial breakout are durable. Both Palantir and Deltek are moving faster than they were two years ago. The question for contractors is whether the next wave of procurement intelligence comes from a platform they already own, a CRM they might adopt, or a decision layer they plug into both.

Pryzm's customers include Forterra, which builds autonomous vehicles and has raised $243 million in total funding, according to PitchBook, TechCrunch's data shows, coding startup SysGit and defense tech venture firm Decisive Point.

Federal Demand: From Experiment to Line Item

Federal AI spending has moved from experiment to line item at a pace that would have been unimaginable three years ago. Brookings tracked the obligated value of federal AI contracts at $7.2 billion in the 2026 budget, a near tenfold jump from 2024, while the potential award ceiling hit $91.8 billion, up nineteen times. The Department of Defense alone accounted for 1,319 contracts worth a potential $90.7 billion, representing 98.9 percent of the entire federal AI market. Twenty-eight agencies now hold AI contracts, up from 17 in 2022. Large agencies documented more than 3,600 use cases in 2025, a 69 percent increase over 2024 and five times the 2023 total. On average, each large agency reported 211 use cases.

The policy machinery behind those numbers is running at full throttle. The Trump administration's July 2025 AI Action Plan set three pillars: accelerate innovation, build American AI infrastructure, and lead in international AI diplomacy. OMB memoranda preserved the chief AI officer roles, procurement reforms, and use-case inventories established under the prior administration. In February 2025, the Department of Government Efficiency launched with a mandate to "use modern technology" against "waste, fraud, and abuse." Deloitte's 2025 aerospace and defense outlook captured the resulting mindset: "Speed to field" has become the unifying metric across portfolios.

Agencies are not short of ideas. They are short of capacity. Nearly 60 percent of documented use cases remain in pilot or pre-deployment stages. Contractors build the majority: 63 percent of use cases with disclosed development models were contractor-led or hybrid, and that share rises to 72 percent at full deployment. The federal workforce cannot absorb the demand. Since January 2025, only 160 full-time federal postings explicitly required AI expertise. Hiring timelines stretch long, security clearances longer, and salaries trail the private sector. Early-2025 workforce reductions cut into the very talent pools agencies had just recruited.

Procurement rules are bending to close the gap. Commercial solutions openings and Other Transaction Authority vehicles are expanding to attract nontraditional suppliers and compress timelines. Contracting officers are directed to favor commercially available tech when feasible. The Federal Acquisition Regulation, the Defense Federal Acquisition Regulation Supplement, and the DoD's Software Acquisition Pathway are being rewritten for outcome-focused, software-driven procurements. Deloitte notes these reforms aim to enable a new generation of smaller, software-centric firms to scale rapidly. IDC forecasts U.S. aerospace and defense spending on AI and generative AI will reach $5.8 billion by 2029, 3.5 times 2025 levels.

Pryzm sits at the intersection of that pressure and that reform. Its Contextual Relationship Management platform was built for the national security ecosystem, mapping relationships, tracking opportunities, and now layering generative AI to automate opportunity detection and relationship intelligence. For contractors, that means managing the flood of new use cases, the shift to multiyear contracts, and the preference for commercial solutions without drowning in SAM.gov noise. For agencies, it means a vendor base that can surface relevant capabilities faster, respond to OTAs and commercial solutions openings with structured data, and reduce the "black box" opacity that complicates audits. The platform's value proposition is not abstract efficiency; it is survival in a procurement environment where the rules, the volumes, and the technology are all changing at once.

The next bottleneck is not funding. It is trustworthy data and auditable outputs at scale.

Where the Model Meets Reality: Compliance, Accuracy, Integration

Pryzm's FedRAMP High authorization and DoD Impact Level 5 baseline clearance represent a genuine breakthrough; they let the platform ingest Federal Contract Information and Controlled Unclassified Information without violating federal security standards. But the authorization is a threshold, not a finish line. Every new data source Pryzm connects, every agency workspace it enters, and every prime contractor it onboards adds surface area for compliance risk. The company's own announcement acknowledges that "growing its federal compliance posture with new certifications and authorizations" remains a forward-looking work stream, not a completed checklist.

Model accuracy presents a quieter but harder problem. Pryzm fuses public data (contracts, congressional hearings, news reports) with each customer's proprietary capture history and internal documents. That hybrid corpus is messy by design: government solicitations use inconsistent taxonomies, program names change mid-cycle, and budget line items shift between continuing resolutions. The platform's value hinges on resolving those ambiguities correctly enough that a business-development lead trusts the output without manual verification. If the model hallucinates a program officer's authority or misreads a color-of-money constraint, the contractor wastes pursuit budget, exactly the failure mode Pryzm exists to prevent.

Integration hurdles compound the accuracy challenge. The landscape is a thicket of "disparate government websites," "fragmented systems," and "countless information silos." Pryzm's pitch is that it replaces the manual sifting across SAM.gov, beta.SAM, USAspending, agency-specific portals, and service-specific acquisition systems with a single digital thread. But each of those sources has its own API quirks, update cadences, and access controls. The DIU eWARP selection validates the technical approach for program, budget, and contract management workflows, yet scaling from a pilot environment to enterprise-wide deployment across multiple agencies, each with its own Authorizing Official and continuous-monitoring requirements, is a different engineering discipline than building the core model.

The competitive clock is ticking. Deltek has been expanding its AI-driven ERP and PSA positioning. Accenture Federal Services has expanded its federal AI footprint. Palantir — whose alumni founded Pryzm — continues to deepen its own federal foothold. Each rival brings deeper integration into existing contractor workflows and longer compliance track records. That funding fuels the next phase of product velocity, but the company has signaled it must expand its teams across those hubs simultaneously, a hiring sprint in a talent market where Palantir alone lists offensive security engineers at $145,000–$200,000 and platform engineers at $135,000–$200,000.

The federal acquisition reform movement adds external pressure. The government's push to prioritize commercial-off-the-shelf solutions and accelerate vendor selection means Pryzm must demonstrate measurable cycle-time reduction, not just data aggregation, to stay on the right side of the new "speed, flexibility, and accountability" mandate. LaRovere's stated goal, that "when it comes time to allocate money, Silicon Valley's up-and-comers are top of mind," only materializes if the platform's opportunity detection and relationship mapping survive contact with real capture teams under real proposal deadlines. The next funding round will be priced on whether that digital thread holds when the money moves.

Working in frontier tech? Zero G Talent tracks the openings: see every open Palantir Technologies role, browse frontier tech jobs, the companies hiring, and the people building the field.