In September 2024, AAC announced a contract to keep supporting the National Library of Medicine. The work included networks, desktops, cloud administration, cybersecurity, and web development. The detail worth pausing over was the relationship: more than 30 years. A technology company can replace almost everything it works on over that span and still be asked to return. That is an interesting kind of staying power.
- AAC runs, secures, and modernizes technology for federal agencies.
- Its work joins the service desk to the cloud, the network, and the conference room.
- Its AI experiments begin with prototypes and operational checks.
Thirty years is a technical detail
Consider what that library renewal actually purchased. AAC’s announcement specified a ten-month base period and two twelve-month options. It covered both the machinery underneath the organization and the support people encounter at their desks. The customer was renewing an operating relationship across several kinds of technology, each with its own maintenance demands.
AAC’s president, Mark Birch, called it a 30+ year partnership
. The number is useful because it shifts attention from an installation to everything that follows. Someone must troubleshoot the network, administer the cloud environment, and help users through changes. A working system is a continuing obligation. AAC has made that obligation its business.
AAC’s description of its NLM relationship at the September 2024 renewal.
The work between the boxes
The company is a privately held, veteran-owned systems integrator serving civilian and defense agencies. Its customer list stretches from the Consumer Financial Protection Bureau and Federal Communications Commission to military organizations. These buyers need people who can connect technologies, support them, and operate within agency requirements. AAC supplies that work through contracts.
Its enterprise portfolio includes infrastructure engineering, cloud migration, endpoint management, service desks, and disaster recovery. Cybersecurity adds monitoring, vulnerability management, identity controls, and incident response. There is also audio-visual and media support, an apparently modest specialty until a meeting depends on it. The breadth makes sense when the unit of analysis is an agency’s working day.
A September 2026 EEOC award makes the proposition concrete. AAC announced systems administration, business analysis, and AV/video teleconferencing support for headquarters and nationwide field offices. Microsoft 365, Teams, endpoint platforms, and collaboration environments sit within the scope. The award carries a one-year base and three option years. A conference room and a managed laptop belong to the same service problem.
desks, devices
risk, response
applied AI
The meeting starts in five minutes. Every layer has a job.
Procurement is part of the product
There is another layer, less photogenic than a server room: the way government buys. AAC sells through acquisition vehicles and task orders. In July 2026, it announced a place on a multiple-award technology support contract for a federal legislative organization. The announcement said AAC would compete for work supporting thousands of personnel across the capital and hundreds of district offices.
That distinction matters. Selection for a vehicle provides access to future competitions; individual orders determine the work. Reading an award announcement as guaranteed revenue would skip an essential step. For a buyer, the practical question is which purchasing route covers the requirement and which contractor can deliver against it.
AAC also works through joint ventures. TA Federal and TA Federal 2 connect it with Tanvi IT Solutions, whose application development experience complements AAC’s federal operations background. AS Innovation joins AAC with SDS Solutions, bringing cloud-native development, automation, and high-performance computing expertise. The differentiation is a combination: technical coverage, institutional familiarity, and a procurement route.
The people who remember
Institutional familiarity has faces. CIO and CTO Ugur Koser joined AAC in 1984. Laura Schindler, now responsible for finance and human resources, began in 1986 as an accounts payable clerk. Birch returned in 2019 after leading another company. These biographies suggest a business where accumulated knowledge can survive successive technology cycles.
The company gives that knowledge a procedural framework. AAC lists ISO certifications for quality management, IT service management, and information security. It also reported a CMMC Level 2 self-assessment score of 110 out of 110 in December 2025. The word “self” belongs in that description. These credentials describe management and security practices; a buyer still has to examine the proposed team and scope.
Teach the machine the plumbing
AAC’s newer AI offering follows the same operational terrain. Its published capabilities include analytics, AI-assisted operations, private generative AI platforms, governance, and training. Its Emerging Technology Lab describes a secure private-cloud environment where experienced architects work with emerging engineers. The sequence starts with a problem, moves to a prototype, and checks the result against security and operational requirements.
A working system is a continuing obligation.
The practical lesson is visible in AAC’s AIOps data-pipeline paper. Operational signals arrive in different forms: logs, metrics, traces, and tickets. Before analytics can help, those inputs need collection, cleaning, and consistent interpretation. Two systems reporting time in different units can create confusion before any model makes a prediction.
Cost lives in that plumbing too. The paper warns about storage, data movement, and API calls, and recommends retention and tiering policies. Keeping everything readily accessible can become expensive. A useful experiment therefore needs boundaries around the information it collects and how long it keeps it.
- IdentifyChoose an operational problem.
- PrototypeTry the approach in a sandbox.
- ValidateCheck controls and deployment realities.
A reader can copy the sequence: select a specific problem, test a small implementation, and evaluate it against the environment where it must run. The inference has limits. Poor telemetry weakens the experiment; an unmet security requirement can prevent deployment. A convincing demonstration needs a workable path into daily operations.
Return to the library. AAC’s AI ambitions are easier to understand beside its maintenance contracts. The opportunity is to improve systems that people already depend on. The test arrives after the demonstration, when the network, the desktop, and the next meeting all need to work again.