100 Days to November. The AI Operations That Win in the Final Sprint.
The final 100 days aren't a time to configure. They're a time to execute. The campaigns that win this window have already built the operation. Here's what executing at full capacity looks like.
One hundred days to Election Day.
This is the window where the money gets raised, the voter contact gets done, and the races get decided. It's also the window where you find out whether the AI operation you've been building since January is actually running at full capacity, or whether you spent six months with something that kind of worked.
The campaigns that win this window aren't configuring anything in August. They're executing. They've already done the setup. The voice model is calibrated. The approval workflow runs daily. The email program is at frequency. The donor cultivation sequences are already in motion.
The final 100 days reward preparation. They punish campaigns that are still figuring out their systems while the sprint is happening.
Here's what execution mode looks like for each part of the operation.
Donor Follow-Up: Velocity and Volume
In the final sprint, every call session matters more than it did in May. The major donor who got called in September and received a follow-up the next morning is a different prospect than the one who got called in September and received a follow-up three days later.
Execution mode for donor follow-up: the trigger is running automatically. The batch is generated overnight. The approval queue is reviewed before 9 AM. The follow-ups are sent by noon. Every call session, every time, regardless of who else is traveling or stretched.
The velocity target doesn't change in the final sprint; 24 hours is still the goal. But the discipline to hit it has to be locked in, because the sprint creates constant pressure to let non-urgent work slip. Follow-up velocity is not non-urgent. It's one of the highest-value activities in the operation.
Email Program: Frequency and Precision
The email program in the final 100 days is running at higher frequency and tighter segmentation than it ran in May.
Two emails per week is the floor for an active general election program in August and September. That cadence should hold through October, dropping to one per week in the final two weeks before the election when the audience has shifted to voter contact mode.
The segmentation discipline matters more in the sprint. Your major donors in active cultivation should not be getting the same email as first-time small donors. Event attendees have a different relationship with the campaign than cold prospects. The AI system handles the segmentation if the list is configured, but the list has to be configured.
What execution looks like: the email for Tuesday is in the queue by Friday. It was generated off this week's news, reviewed once, and approved. Nobody spent an afternoon writing it. Nobody missed a send because the comms director was busy.
Major Donor Cultivation: The Ask Window
August and September are the ask months for major donors. The cultivation that happened in June and July (the personal touches, the research conversations, the relationship maintenance) is now converting to formal asks.
An AI-assisted major donor operation in execution mode is running: personalized pre-ask research for each prospect (what changed in their world since the summer conversation), a formal ask email written in the candidate's voice for each ask conversation, and a follow-up sequence for prospects who expressed interest but haven't committed.
The campaigns that run their major donor asks correctly in August and September have a different October than the ones that are still trying to get in front of major donors in the final 30 days. The final 30 days aren't a major donor cultivation window. They're a closing window.
Rapid Response: Already Built
A breaking news development in September (an opponent attack, a local controversy, a policy development) requires a campaign response within hours, not days. Campaigns that have a rapid response path already built can do this. Campaigns that are figuring out the process while the story is moving can't.
Execution mode for rapid response: a designated decision-maker, a template the AI can adapt quickly, an expedited approval path, and a distribution setup that can execute in hours. This path should exist before it's needed, not assembled in real time.
The AI layer handles the drafting speed. A well-configured system can take a breaking story and generate a fundraising response email in 15 minutes. But the approval and distribution path still has to work on the human end. Both sides have to be ready.
Continuous Monitoring: Current Race, Not Primary Race
The news monitoring and content feed should be generating useful content through November: daily digests, issue developments, opponent activity, media coverage relevant to the race.
By August, if the monitoring source list hasn't been updated since the primary, it's generating noise. The general election race has different opponents, different issue dynamics, and different local media than the primary did. The monitoring configuration has to reflect the current race.
Execution mode: daily digest lands in the morning, relevant items are flagged, content team decides what to act on. No manual scanning. No catching up on what you missed. The feed keeps running.
What This Window Rewards
Campaigns that execute well in the final 100 days share a few things:
They made decisions about process before the sprint, not during it. The approval workflow was locked before August, not negotiated in October when everyone is stretched.
They treated configuration as ongoing maintenance, not a one-time setup. The voice model got updated. The source list got reviewed. The email segmentation reflected where the race was, not where it started.
They kept humans in the loop on judgment calls and used the AI layer for the structural work. No campaign ever got into trouble because an AI drafted a follow-up email that a human reviewed and approved. The problems happen when the approval step gets skipped because things are moving fast. In the sprint, things are always moving fast. The approval step has to be non-negotiable.
For Campaigns That Don't Have This
If the operation described here doesn't match what you're running, the question isn't whether to build it. The question is how fast you can get to something that works before the sprint is fully underway.
The setup takes weeks, not months. But it takes weeks, which means the window is August, not September. An operation that starts standing up in September will spend October debugging instead of executing.
The campaigns that win in November are the ones that show up to August with the infrastructure ready. The ones that start building in August can still catch up. The ones that are still standing up systems in September are running a different race.
Eric Linder is a former California State Assemblyman (2012-2016) and founder of AutomatedTeams, an AI operations consultancy for political campaigns and advocacy organizations.
Eric Linder
Former California Assemblyman. Now building AI operations for political campaigns.
ericlinder.com →