Part 1 of 2: The integration gap — and the five workflows separating leading teams from everyone else
A note on timing: this piece reflects the state of AI in government affairs as of June 2026. That caveat isn’t boilerplate; the capabilities described here didn’t exist in their current form a year ago, and by the fall, there will be tools and tactics this article doesn’t cover. Treat what follows as a snapshot and a starting point, not a finish line.
If you’re a government affairs leader still debating whether your team should use AI, the debate has been settled without you. McKinsey’s latest global State of AI survey found that 88% of organizations now use AI in at least one business function, up from 78% a year earlier. In the public sector, a 10-country survey of more than 3,300 public servants found that over 70% are already using AI in their work. The people drafting the bills you track, the staffers preparing the hearings you attend, and the regulators writing the rules you analyze are using these tools today.
But here’s the number that should actually shape your second-half planning: in that same McKinsey research, only about 6% of organizations qualify as “AI high performers” — companies seeing significant, measurable value. And on the government side, while 70%+ of public servants use AI, only 18% believe their institutions are using it effectively.
The gap between using AI and using it well may be the defining business challenge of 2026. Adoption is no longer a differentiator; it is how you integrate tools into your workflows.
What Separates the Top Teams (It Isn’t the Tools)
McKinsey’s data is unusually blunt on this point: high performers are nearly three times as likely as everyone else to have fundamentally redesigned their workflows around AI rather than bolting AI onto existing processes. Wharton professor Ethan Mollick, whose research with the Boston Consulting Group documented 25-40% performance gains for consultants who use AI effectively, describes the failure mode precisely: most professionals use AI like a slightly better search engine, asking it questions and pasting answers. The teams getting transformative results embed it directly in the flow of work, which is why the biggest gains come from AI built into the systems where work already happens, not from a chatbot in a separate tab.
Here’s what that redesign looks like in practice among the government affairs teams pulling ahead:
- Why the adoption debate is over — and what the data actually says about who’s winning
- The five workflows separating leading teams from everyone else
- What a real practitioner’s AI setup looks like in a fast-moving legislative moment
- Why capability is a team sport, not a tool purchase
- The honest caveats that keep AI output trustworthy
Monitoring Becomes Triage, Not Reading
State legislatures introduced more than 1,200 AI-related bills in 2025 alone — and that’s one issue area. No human team reads everything anymore; the leading teams have stopped pretending to. They’ve redesigned intake so analysts start each morning with a triaged, prioritized brief instead of a raw feed.
- What this looks like in practice: In a platform like Quorum, AI-generated bill summaries and version comparisons run automatically against your tracked issues across all 50 states and Congress.
- What your team does instead: Analysts start each morning deciding what matters — not discovering what exists.
- Where the recovered time goes: Relationships and strategy — the work AI can’t do.
Preparation Becomes Systematic
Meeting prep used to consume hours per meeting and relied entirely on whoever held the institutional memory. Top teams have made it a repeatable workflow.
- What used to take an afternoon: A legislator’s committee history, recent statements and votes, district context, and your organization’s full engagement history with that office.
- What it takes now: Quorum’s Meeting Prep Agent automatically assembles that brief from legislative data and your logged interactions — producing in minutes what used to take hours.
- What changes for the lobbyist: Time shifts from assembling context to deciding what to do with it. And the quality of prep no longer depends on who’s available that week.
Institutional Memory Stops Walking Out the Door
The classic government affairs failure: years of relationship intelligence living in one person’s head or inbox, gone when they leave. Leading teams now treat interaction logging as a system, not a chore — and AI has removed most of the friction.
- What’s changed: A CRM Agent like Quorum’s can draft interaction records from meeting notes and surface relationship history on demand.
- Why it compounds: Knowledge builds rather than evaporates — and your logged history is exactly the proprietary data that makes every other AI workflow on this list smarter.
- The bottom line: AI working from your institutional knowledge is an asset. AI working from generic public data is a commodity everyone else has too.
Communications Scale Without Scaling Headcount
The Public Affairs Council has documented this shift among its members: the American Society of Clinical Oncology’s PAC used AI-assisted email programs to more than double its fundraising — twice the dollars and twice the contributors — without adding staff.
- Where AI takes the wheel: Tools like Quorum’s Policy Comms Agent handle the first draft and the mechanics of tailoring an update to different audiences — executives, members, coalition partners.
- Where humans stay in control: Voice, judgment, and the political read.
- The real unlock: It’s not just faster comms. It’s comms that actually go out, because the blank-page problem no longer kills them.
Analysis Moves Upstream
The most sophisticated teams aren’t just summarizing what happened — they’re using AI to model what’s coming.
- What that looks like: Which carryover bills are positioned to move, where amendment language is migrating between states, which committee assignments signal priority.
- What it requires: Monitoring, stakeholder, and engagement data living in one system that AI can reason over.
- Why it matters: This is where AI stops being a productivity tool and starts being a strategy tool.
What This Looks Like on a Real Team
Dustin Perchal, who leads advocacy work at the ALS Association, describes a workflow that hits nearly every pattern above — running both Quincy, Quorum’s AI assistant, and general-purpose AI tools in parallel, each where it’s strongest.
“I’ve used Quincy to help answer questions from lobbyists on state and federal bill hearings, to know what actions have gone to lawmakers that sit on key committees, identify personalized messages, and move advocates up engagement ladders,” Perchal says.
The payoff shows up when the calendar compresses. As the House moved on its research funding bills this spring, Dustin used AI to compare what passed for FY26 against what’s proposed for FY27 — a mixed picture in which neurological research funding is rising while overall budgets shrink, with downstream impacts (like proposed CDC cuts) that affect the ALS community’s programs. When report language drops the same day as a full committee markup, “we have to act fast with analysis and response.” That speed — credible analysis in hours instead of days — is the difference AI makes when it’s embedded in the actual work.
Notice two things about Dustin’ setup. First, it’s a portfolio, not a single tool: a purpose-built platform handles the work that depends on structured legislative and advocate data, while frontier models handle long-document analysis. The teams getting the most from AI in 2026 have stopped asking “which tool should we buy?” and started asking “which tool is right for which job?” Second, none of it is delegated to a single “AI person.” The analysis resides with the policy expert because their expertise makes the AI output trustworthy.
Capability Is a Team Sport, Not a Tool Purchase
That second point deserves its own treatment, because it’s where most AI initiatives in this profession quietly stall. The pattern is familiar: leadership buys licenses, one enthusiast on the team becomes “the AI person,” everyone else nods along, and six months later, usage data shows a single power user and a graveyard of dormant seats.
The research explains why. The Public Sector AI Adoption Index found that in environments with clear guidance and active support for AI use, 91% of public servants feel confident using AI and 79% find it empowering — versus dramatically lower confidence where rules and backing are unclear. Confidence isn’t a personality trait; it’s a function of permission, training, and seeing peers succeed. Mollick’s prescription for building it is disarmingly simple: bring AI to everything you do, legally and ethically, until you’ve personally mapped where it helps and where it fails. He calls that boundary the “jagged frontier.” AI is superhuman at some tasks and embarrassingly bad at adjacent ones, and the only way to learn the shape of the frontier for your work is direct, repeated use.
For a government affairs team, building that shared capability looks like four habits:
- Make experimentation visible. A standing five minutes in the weekly team meeting — “what did you try with AI this week, and did it work?” — does more for adoption than any training deck. Failures are as valuable as wins, because they map the frontier for everyone.
- Pair your skeptics with your enthusiasts. Skeptics catch the hallucinated bill citation; enthusiasts find the workflow nobody thought to automate. A team of all-enthusiasts ships errors; a team of all-skeptics ships nothing new.
- Write down what works. When someone develops a prompt or workflow that reliably produces a good committee-hearing summary or a strong first-draft member update, capture it. A shared library of proven workflows is how one person’s discovery becomes team capability — and it’s the institutional-memory principle applied to AI itself.
- Set the verification norm explicitly. Every team needs a stated rule for what AI output can ship without review (internal brainstorms, formatting) and what always gets expert eyes (anything citing a bill, anything going to a lawmaker, anything public). Ambiguity here is how organizations end up in the news.
The Honest Caveats (Because Credibility Requires Them)
AI in this profession has sharp edges, and teams that ignore them get burned. General-purpose chatbots will confidently cite nonexistent bills and mischaracterize legislative language — one reason purpose-built tools grounded in actual legislative data outperform copy-pasting into a consumer chatbot, and why every AI-generated analysis that informs a position still needs human verification.
Perchal’s experience is instructive here too: “Because of the complexity of some of these issues, I have caught each of the external AI tools providing incorrect information at one time or another — so it’s also keeping them in check and guiding them to appropriate outputs.” That’s the posture of a team using AI well: trust the speed, verify the substance.
And the limits go beyond accuracy. AI cannot read a room, doesn’t know that a chair’s public position differs from her private one, and has no relationships. The teams winning with AI are explicit about this division of labor: AI compresses the time between question and draft; humans own everything that touches judgment, trust, and advocacy itself.
Coming in Part 2: The Clock Starts Now
Everything above is about how leading teams work today. But the second half of 2026 isn’t a neutral window to build capability at your own pace. A midterm election cycle is underway in which AI-generated deep fakes are already a mainstream campaign tactic, and 30 states have enacted their own rules. The 2027 legislative sessions, the heaviest AI-policy calendar the profession has ever faced, start the moment the new legislators are sworn in. And Colorado’s comprehensive AI law is now in effect, with dozens of statehouses watching.
In Part 2, we cover the three election-cycle work streams every team needs before November, the three moves to make before year-end to be ready for 2027, and a concrete 30/60/90-day playbook for getting from “we have licenses” to “we have an operating discipline.”
