Rob Dolin

🛑 Government shutdowns don’t just halt politics—they halt progress. For nonprofits, schools, and local governments already operating on razor-thin margins, every day without federal funding deepens the gap between needs and resources. It’s not just about missed paychecks or paused contracts. ➡️ An over-reliance on one funding stream can tank a program. ➡️ Weak reporting systems can lead to contracts canceled “for convenience.” ➡️ Interrupted investments stall critical programs—housing, food access, climate resilience, youth development. And the worst part? Communities carry the cost. The shutdown reminds us: systems matter. If your funding strategy lives across spreadsheets, inboxes, and 10 different portals—you’re not set up to weather disruptions, let alone scale. That’s why I’m energized by AI and agentic workflows—not as buzzwords, but as tools to: ✔️ Find and win diverse, aligned funding ✔️ Track and report spend in real-time ✔️ Centralize data and compliance across departments ✔️ Free up staff to focus on mission, not maintenance At Eddi, we help organizations turn unstable grants into sustainable growth—with tech that’s built for the real-world mess of public funding. 📣 How has the shutdown impacted your team or your community? Do you have specific funding needs that need immediate support?

1

Access to care shouldn’t depend on geography. AI can help rural hospitals stay financially strong and serve their communities better. The claims denial navigator—available now in the GitHub Models catalog—is a free, AI-powered tool developed by Microsoft Partners and rural health leaders from our Microsoft’s Rural Health AI Innovation Lab (RHAIL) Read more: https://msft.it/6046sFc0Q #healthcareAI

1

One of the most common questions I hear is this. If government agencies are all part of the same system, why is it still so hard for them to share data? On the surface, it sounds simple. Connect the systems. Enable access. Move information from one place to another. But in reality, it is rarely that straightforward. From experience, the challenge is not just about technology. It is about structure, trust, and long-established ways of working. Many institutions were built to operate independently. Over time, each developed its own processes, priorities, and systems. So when the need arises to connect them, it is not just systems that must align. It is people, expectations, and sometimes even mandates. There is also the question of ownership. Who controls the data? Who is responsible if something goes wrong? Who decides how it is used? These are not technical questions. They are institutional ones. And then there is trust. For any system to work across multiple agencies, there has to be confidence that the data being shared is accurate, secure, and will not be misused. Without that trust, even the best-designed system will struggle. What this has taught me is simple. Interoperability is not just about connecting systems. It is about connecting institutions. And until we pay equal attention to both, progress will always feel slower than it should. #DigitalGovernment #eGovernance #GovTech #PublicSector #DigitalTransformation #DataGovernance #Interoperability #Leadership #Nigeria #Innovation

9

One pattern I’m seeing consistently across federal agencies right now: a strong preference for RAG-based architectures over fully custom-trained models. It makes sense. Retrieval-Augmented Generation (RAG) approaches let teams ground GenAI in authoritative policy, guidance, and program data while keeping governance, audibility, and updates manageable in heavily regulated environments. Over the past few years, I’ve designed and implemented Bedrock driven assistants in the federal space with knowledge bases, document pipelines, and guardrails that agencies actually trust enough to use in production. With recent advances in Amazon Bedrock around agent governance, evaluations, and knowledge base patterns, it’s getting easier to scale these solutions for citizen services and internal teams supporting them. All for a surprisingly low cost and ramp up time. If you’re navigating GenAI decisions in the federal space (RAG vs. fine-tuning, governance models, or how to operationalize this beyond a demo), I’m always happy to compare notes and lessons learned.

12

2 Comments

U.S. energy demand is surging — don’t get left behind. New gas projects and large industrial loads are reshaping power markets at breakneck speed. In this high-stakes landscape, timely and accurate data is essential. Halcyon’s Data Products, powered by a regularly updated authoritative data catalog of 4.5+ million documents from all 50 public utility commissions, ensure you always have the latest energy market intelligence. Download this guide to see how AI can help power sector stakeholders find new growth opportunities first.

6

GovDash just raised a Series B. I'm an engineer, CTO, and one of the co-founders of GovDash. I spend most of my time thinking about systems, product, and why so much of the work in government contracting is beset with unnecessary friction. We started with proposal management because it was the most acute pain point: detail-intensive work where small mistakes have outsized consequences, and an area that's been historically underserved by good software. But proposals are just one part of a much larger workflow. Teams work across opportunity discovery, capture, proposals, and contract management, often stitching together unrelated tools and manual processes. Each of these areas could be a company on its own (and historically has been), but teams shouldn't have to absorb the complexity and inefficiency created by that fragmentation. Our aim is to eliminate that friction across the entire GovCon lifecycle. We're building software that supports our customers end-to-end, so information flows cleanly from one phase to the next without constant rework, duplication, or brittle handoffs. By taking on that complexity in the product, we let teams focus on the work that actually differentiates them: delivering critical goods and services to our nation. Proud of the team, grateful to our customers, and excited to keep building. --- P.S. A note for engineers. We're working on problems that require building systems and product from first principles, because best practices for reliable, agent-driven systems in this domain don't exist yet. We invest heavily in appropriately constraining and auditing model behavior (without making outputs feel stilted or generic), building testing and evaluation frameworks that can wrangle nondeterministic systems into producing accurate, useful outputs with high confidence, and encoding deep domain context so results aren't just "correct," but trusted and verifiable by our users. A lot of the work is about making complex systems predictable, observable, and "boring" in the ways that matter. And since we deal with sensitive customer and government data, there's also a significant amount of infrastructure and security work. That includes reducing the surface area of our trusted computing base, eliminating mutable state in our infrastructure, and building automated mechanisms to continuously test our security and reliability guarantees, from Playwright checks to automated red team tests. If you're an engineer who wants to work on these kinds of problems, we're hiring. There's a lot of autonomy here: many of the engineers we've hired are former founders, and they thrive in this environment. If you want to own hard problems, work directly with customers, and see your work's impact on them, you will too.

106

27 Comments

In the public sector, we talk a lot about "responsible AI" and ensuring human review of AI-generated content. Washington's recent issue with "Spanish-Accented English" instead of Spanish menus is exactly why. While it might seem like a gag straight out of Family Guy, it's an unfortunate example of government services beyond the reach of non-English speakers due to a system design error. If I had to guess, Washington DOL used something like Amazon Polly and selected Spanish, but then uploaded scripts/answers in English. Polly isn't a translation tool; it's a text-to-speech tool. You can select Spanish, which will impact the pronunciation, but it still won't translate text. https://lnkd.in/eaGfG3yF

18

2 Comments