Reimagining Government Tech

I’ve spent an interesting several months working in government with several different agencies and seen some consistent patterns.   I think my years of experience in industry prior to government help shape my perspective on how government runs. These are my personal thoughts in my personal capacity not related to work.

Here goes…

When a veteran moves states, they must navigate 43 different websites with separate logins to transfer their benefits → Imagine one secure digital ID that seamlessly transfers all benefits nationwide

Medicare still processes 4.6 million paper forms annually → Picture a modern web native  interface like most everything else on the web, immensely more efficient and cost saving. .

Federal employees juggle 3-4 separate computers to access different agency systems → Envision one secure device with smart authentication.

The IRS runs on 60-year-old COBOL code → Transform to modern, maintainable systems like those used by fintech companies.

Government doesn't attract qualified developers because they offer significantly below market rate salaries.

We know this transformation is possible because:
- Estonia serves 99% of government services digitally
- UK's Gov.uk platform unified 1,800 websites into one user-friendly system
- Singapore's digital ID system enables instant access to all government services

The cost of our current system? $100+ billion annually on outdated IT. The cost of transformation? Far less than maintaining the status quo. We pay out extraordinary sums to consulting firms rather than building out the capabilities we need our government to provide.

This isn't just about efficiency - it's about creating a government that works for everyone. When filing taxes takes minutes instead of hours, when veterans get immediate access to care, when businesses can focus on growth instead of paperwork - that's when we'll know we've succeeded. A shared resource and benefit for all as we work together.

The technology exists and our government orgs are extremely overdue for change. Other countries have shown the way. Let's build the digital government Americans deserve and works for the people, not just the rich and wealthy.

AI in 2025: Less Chat, More Action

AI interaction is ripe for transformation. Brute force scaling LLMs with more and more data is showing signs of diminishing returns. I’m looking forward to getting more creative and innovative.

5 key areas I am focusing on in 2025.

1. Moving Beyond the Chat Paradigm

While the chat interface revolutionized how we interact with AI, it's time to break free from these constraints. We need both incremental improvements within existing chat frameworks and entirely new paradigms of interaction. Imagine AI that can understand your context through browser activity, respond to pointing gestures, or seamlessly integrate multiple input modes. The future of AI interaction lies in making these exchanges feel more natural and contextually aware.

2. Empowering AI Agency

The next frontier isn't just about AI that can chat – it's about AI that can act. By giving AI systems access to deterministic APIs and tools, we can move from conversation to collaboration. This isn't about autonomous AI, but rather about AI that can execute specific, well-defined tasks within clear boundaries. The key is building these capabilities thoughtfully, with robust safety measures and user control at the forefront.

3. AI Collaboration

We need to move beyond the current model where AI is merely a passive assistant. Imagine real-time collaboration where both human and AI can simultaneously edit documents, code, or designs. This isn't just about parallel work – it's about creating a genuine partnership where both parties can contribute their unique strengths to a shared objective.

4. Sophisticated Training & Feedback Loops

Current AI systems often lack the ability to learn from user interactions in meaningful ways. We need to develop better mechanisms for capturing implicit feedback from user actions and behaviors. This isn't about collecting more data – it's about understanding the quality and context of user interactions to drive meaningful improvements in AI capabilities.

5. Multimodal Interfaces

The future of AI interaction will be multimodal. Voice interfaces that can detect emotional tone, gestural inputs that feel natural, and the ability to work with common data formats seamlessly – these are not just features, but essential elements of making AI truly accessible and useful. The goal is to make interaction with AI as natural as working with a human colleague.

Thoughts? What other areas of AI innovation do you see as critical for 2025? Let's discuss in the comments.

Big tech, we need the innovation back!

The Early Days: A Time of Pure Innovation

I started my career as an industrial designer in 1995, the tech landscape was quite different. Software hadn't yet eaten the world, hardware design was still hot for investment. A tech companies like Apple, HP, Microsoft existed, but they weren't the giants we know today. The absence of dominant "FAANG" companies meant that talented technologists gravitated toward startups and smaller companies—drawn by the promise of innovation rather than just compensation.

What stood out most was the motivation: people in tech were primarily driven by a passion for innovation, exploration, and creating real value. The financial rewards were important, but secondary to the excitement of building something new.

The Four Waves of Tech Evolution

Looking back, we can roughly divide tech evolution into four major phases:

1. Personal Computers

2. Internet

3. Mobile

4. AI ( just getting started )

Each wave created tremendous value and potential. This success also attracted more profit-focused players to the industry. As technologies matured, companies shifted from innovation to optimization, from creation to control.

The Rise of Big Tech

Today's tech landscape is dominated by what we know as "FAANG":

• Facebook (Meta) - social media

• Apple - phones/laptops

• Amazon - ecommerce

• Netflix - video streaming

• Google - search and ads

Take Google as an example: In its early days, the company's success was intrinsically linked to an open and thriving internet ecosystem. Their business model elegantly aligned with making the web more open and accessible—when the internet prospered, Google prospered. This philosophy drove innovations like Chrome and Android, which helped democratize internet access initially. As Google and these other tech companies grew and matured, they turned into more standard businesses and attracted more people focused on optimization and profit rather than creating new products and user value.  These companies focus on creating protected platforms and ideally monopolies for their products and services, to lock in users and markets so they can raise prices and have little incentive to maintain let alone improve their offerings.

Innovation Paradox

Our current oligopolistic big tech structure creates three significant challenges:

1. The Talent Vacuum: Big tech companies attract and retain top talent with unprecedented compensation packages. However, these brilliant minds often find themselves working on incremental improvements rather than breakthrough innovations.

2. Trapping Users: Locking users into their platforms without creating more value and degrading quality over time for cost reduction.

3. Innovation Barriers: Market dominance and user lock-in create high barriers for new entrants, making it harder for innovative solutions to emerge and gain traction.

Rekindling Innovation

I believe the key to long-term value creation over short-term profit extraction is about fundamentally realigning incentives.  Not a simple ask I know, but really obvious as you consider it.  When we create economic systems that reward short term results, we will get short term optimization.

1. New Funding Models: Traditional venture capital often pushes for rapid growth and quick exits, prioritizing profit over sustainable innovation. Alternative funding structures like long-term capital vehicles, public benefit corporations, and community-owned platforms can help companies stay focused on innovation and value creation.

2. Decentralization Opportunities: Emerging technologies are enabling new organizational structures that distribute power and ownership. DAOs (Decentralized Autonomous Organizations) and other novel governance models can align incentives between creators, users, and investors in ways traditional corporations cannot.

3. Public / Government Owned Platforms: We need platforms and systems designed to generate and distribute economic value across entire communities, not just shareholders: Digital public infrastructure that everyone can build upon Platforms where value generated flows back to all participants Systems that turn users into economic stakeholders Technologies that create shared prosperity rather than concentrated wealth Open protocols that enable broad-based innovation and value capture

Agentic or Bust: We need to get past the Chatbot Paradigm... please! don't be lazy!

The remarkable progress in Large Language Models (LLMs) has ushered in a new era of human-computer interaction. However, we find ourselves at risk of becoming trapped in the comfortable but limiting paradigm of chatbots - just as the Graphical User Interface (GUI) both revolutionized and eventually constrained computer interaction for decades. While natural language interfaces represent a significant leap forward, relegating AI to mere conversation partners severely undermines their potential impact on human productivity and capability augmentation.

The Chat Interface Trap

The chat interface has become ubiquitous in AI applications for good reason - it leverages our most natural form of communication and requires minimal learning curve. As Andy Lulham notes in "The UX of AI" (2023), "Chat interfaces provide an intuitive entry point for users to interact with AI systems." However, this apparent strength may be becoming our greatest limitation.

The problem parallels what Brad Myers documented in "A Brief History of Human Computer Interaction Technology" (1996) regarding GUI interfaces: while the desktop metaphor made computers accessible to the masses, it also "led to a narrowing of imagination regarding alternative interaction paradigms." We risk making the same mistake with AI by conflating natural language interaction with conversation-only interfaces.

The Case for AI Agency

Rather than asking AIs to simply describe actions or provide guidance, we need to empower them to directly execute tasks through well-defined tools and APIs. As Ethan Mollick argues in "Productivity in the Age of AI" (2023), "The real breakthrough will come when AI can seamlessly integrate with existing software tools and workflows, acting as an intelligent agent rather than just an advisor."

Consider mathematical operations - instead of training LLMs to perform complex calculations (which they often struggle with due to their probabilistic nature), we should enable them to utilize calculators, spreadsheets, and mathematical libraries. This mirrors how humans augment their capabilities with tools rather than trying to perform all computations mentally.

The Power of Deterministic Tools

The key insight is that many tasks are better suited to deterministic tools than probabilistic language models. As computer scientist Grady Booch observes, "We should use computers for what they do best - precise, repeatable operations - while using AI for what it does best - handling ambiguity and providing high-level direction."

Examples of deterministic tools that AI agents could leverage include:

- Database queries and operations

- API calls to web services

- File system operations

- Image processing libraries

- Code compilation and execution

- Calendar and scheduling systems

Reimagining Computer Interfaces

The path forward requires fundamentally rethinking how we interact with computers in an AI-first world. Instead of forcing all interactions through a chat interface, we need multi-modal systems that combine:

1. Natural language understanding for high-level direction and clarification

2. Direct manipulation of data and objects

3. Automated execution of tasks through appropriate tools

4. Rich visual feedback showing system state and actions

5. Collaborative interfaces where humans and AI can work together on shared artifacts

Learning from Historical Parallels

The transition from command-line interfaces to GUIs provides valuable lessons. As detailed in "The History of the GUI" by Jeremy Reimer, the GUI succeeded because it made computers more accessible while expanding their capabilities. Similarly, AI interfaces need to balance accessibility with capability expansion.

The smartphone revolution offers another instructive parallel. BlackBerry's failure to move beyond the physical keyboard metaphor left them vulnerable to Apple's radical touchscreen interface reimagining. We must be equally willing to break from familiar paradigms in AI interaction design.

The Path Forward

To realize the full potential of AI systems, several key developments are needed:

1. Standardized APIs for AI tool use and system integration

2. Improved frameworks for managing AI agent permissions and capabilities

3. Better interfaces for collaborative work between humans and AI

4. More sophisticated ways to verify and validate AI actions

5. New design patterns for multi-modal AI interaction

Wrapping up...

The chatbot paradigm, while powerful, must not become another "golden hammer" in interface design. By empowering AI systems with agency - the ability to take direct action through appropriate tools - we can move beyond conversation to true collaboration. The future of human-computer interaction lies not in simply talking to our computers, but in working alongside them as capable partners.

Just as the transition from CLI to GUI fundamentally changed computing, the shift from passive chatbots to agentic AI assistants will transform how we work with machines. The challenge ahead is not just technical but conceptual - we must expand our vision of what AI interfaces can be.

References

Lulham, A. (2023). The UX of AI: Designing for Intelligence

Myers, B. A. (1996). A Brief History of Human Computer Interaction Technology

Mollick, E. (2023). Productivity in the Age of AI

Reimer, J. (2005). A History of the GUI

Booch, G. (2023). Computing the Human Experience

Innovate, don't just iterate

Companies think they are “innovative,” but, in reality, they just aren’t.

Companies seem to fail at innovation for several reasons:

• Focus on short-term revenue: Revenue is important. Without it, a company can’t exist. However, when revenue in the short-term is prioritized above all things, innovation–which can take multiple quarters, even years–often gets sidelined.

• Fear of failure: Innovation isn’t magic. Most new ideas won’t work out. This is part of the innovation process, but creates a false sense of “failure” in most people. Instead, people should see these events as “learning” moments on the long road to innovation.

• It’s just too hard: As an organization gets bigger, things become harder to do. Innovation requires stamina, perseverance, and guts which many do not have the appetite for.

• Performance reviews: Knowing how you’re doing at work and setting goals is all good. However, frequent reviews result in a focus to control outcomes in order to get that “Exceptional” rating.” Putting a rating first tends to produce incremental vs truly innovative product ideas.

I’ve experienced all of the above, in my 20+ years experience as a product designer. However, I’ve had the opportunity to work on some of the most innovative projects at companies large and small. Innovation is possible at any company, but it takes more hard work, discipline, and commitment than you might expect.

Evaluating Ideas: A Critical Step in the Innovation Process

Innovation emerges from generating numerous ideas rather than a single "eureka" moment. However, generating ideas alone is not sufficient. Without a systematic approach to evaluating these ideas, innovation remains elusive. Throughout my career, I've honed a process for evaluating ideas using specific criteria, a practice I refer to as "Eval-Crit."

Eval-Crit goes as follows:

1. Determine an overarching theme or goal for the ideas

2. Develop an initial set of key criteria to evaluate your ideas with your team

3. Generate ideas as a team

4. Evaluate and score ideas based on the theme and criteria previously set

5. Adjust evaluation criteria (if needed) based on learnings from evaluating your ideas

Best practice: 5 criteria is a good number. I like to stay between 3 to 7. More criteria become harder to keep in mind and is likely a signal that you are not clear enough on what factors are important.

Setting Effective Criteria

Determining appropriate criteria is important. Good criteria are specific, and concise yet not overly detailed or complex. They cannot be too vague and general either. It takes experience and skill to get good at developing good criteria quickly. Having someone with experience and intuition is a huge accelerator to success.

For most projects, you will want to include these two foundational criteria:

• Criteria 1: How much effort, cost, time, or expertise will the idea require? (feasibility)

• Criteria 2: How valuable is the idea to the user? (user value)

Now, add 1 to 3 additional criteria that are core to your goals.

Here’s a simple example of the criteria used to evaluate ideas on potential user applications for a company’s AI imaging technologies:

1. Development effort required

2. Amount of our technology used

3. User value

4. How viral or social it may be

5. Novelty

I use a 1 to 5 score for each criteria–I find a 10 point scale to be overkill. I've experimented with weighting certain criteria more than others but it tends to add unnecessary complexity, especially if you’ve created the right set of criteria.

This is likely obvious to most people that regularly do research surveys, but worth mentioning: make sure that a score of 1 is low (bad) and 5 is high (good) for each criteria. This way, when calculating the total score for each concept, they add up and don’t cancel each other out. For example, if you had a criteria such as “amount of engineering effort” a lower score should be more effort and a high score of 5 should be the least amount of engineering effort.

Discussing and Ranking Ideas with Your Team

The team discussion around determining scores for each idea is invaluable. New information will arise and differences of opinion or definition will become clear and be more readily acknowledged, if not completely resolved. It does a great deal to increase the team's alignment and shared knowledge.

Best practice: Reduce confusion and bias by presenting ideas in a consistent and concise format. I recommend my poster format,  in an earlier post.  If ideas differ in format it's hard to evaluate them consistently–e.g. if one idea has a slick video it may score high versus an equally good idea that is just a napkin sketch or research paper. A consistent format ensures comparable rating of ideas.

Don’t get too focused on exact scores, this is a directional tool, not an exact calculation. If any ideas seem out of rank it's okay to investigate and evaluate it. New interesting learnings can be gained when ideas are not where you thought they would be.  The ideas with the  highest score move to the next phase of the innovation process: prototyping. Stay tuned for my next article on that topic.

I’ve found this practice of evaluating ideas using my Eval Crit method highly effective at producing better, more data driven decisions around new ideas and projects. It keeps teams aligned and provides transparency on how ideas are rated and decisions are made.  For those of you wishing to inject more innovation into your day-to-day role or desire to help your company shift away from a “play it safe mentality,” give it a go.

Unleashing Ideas: Visualize and share concepts with Posters

Inventing and developing great products often starts with generating lots of ideas. Many people have written and talked about ways to do this, yet I've not seen much shared around the next important steps – capturing, sharing, and evaluating those ideas. I’ve been fortunate to have spent considerable time in my career generating ideas, as well as pulling great ideas out of other people.

Let’s imagine you and your team have generated a bunch of ideas. Hopefully several good ones, and likely many not so great ones (if you are doing it right). You've done a good job diverging and now you need to converge and decide which ideas to pursue. You have the good problem of information overload. The ideas might be vague or unclear and understood differently by various people. Selecting the best ideas to work on can be emotional and arbitrary, not data driven, or just confusing and overwhelming.

3 key things need to happen next:

1. Capture the ideas: Record and document the ideas in a well-defined and concise manner

2. Share & gather feedback: Show the ideas to others to get their thoughts, feedback, and buy-in

3. Evaluate & select the best: You can’t work on all of them so you need to select the best ideas to pursue

I have developed a format that is an excellent tool for this process. I call the format posters. I originally called them “1-pagers” but I think poster is a better term even though it's perhaps a bit general and could possibly mean different things in different contexts. I’ve used and refined my poster format to consistently help deliver new and innovative products for many companies and teams.

The poster format is a single page that conveys a single concept clearly and concisely.  It should be easy for people to quickly grock. People will often be reviewing several ideas at once, and comparing and discussing them. Clear and concise posters are more efficient at communicating the concepts and facilitate easily scanning and referring to specific concepts in discussions.

Time is scarce and people will likely look through your concepts more quickly than you might prefer. If concepts are long, complicated, or confusing your ideas will get lost. A 2-3 page essay or several pages of images may be complete and thorough; however, this is not the time for that level of detail.

Elements of a poster

1. Title: Succinct, self explanatory, and easy to recall

2. Description: A few sentences that fully explain the idea

3. Visual: An image, series of images, animation loop, or possibly a video. Still images are obviously better for printing but if you’re distributing digitally, animation or video can be useful.

Below is a very basic example of a poster. I used a ChatGPT-like concept for the example because it's well known. For actual concept posters I take care to generate clearer, higher quality images and descriptions.  The title here should be refined a bit, but this example demonstrates the format.

Benefits of posters

• Ideas are captured and less likely to get lost and forgotten

• Posters become a great resource and record

• Making posters can generate more ideas and help you hone your ideation and communication skills

• The format forces you to make your ideas clearer

I have found posters an essential tool for sharing and consuming ideas effectively. The consistent format also facilitates the evaluation and ranking of ideas. (I’ll write more another time about my concept evaluation process.)

Posters may seem simple when spelled out like this. Most elegant solutions appear fairly obvious when explained. While the poster format is quite simple, creating good posters takes practice and skill. After making a few posters you will discover how valuable an exercise they are and how challenging making good posters can be.

I’ve found posters to be an invaluable tool in getting from early fuzzy ideas to shipping new innovative products. I believe this is crucial now, more than ever, in this exciting time of rapid acceleration thanks to AI.

I welcome your thoughts and references to similar or related tools. If you try using posters, I’d love to hear about your experience and results.

- Andy Dahley