30:55The Retention Roadshow
Austin 2026
Practical AI Adoption for Marketing Teams with Rytis Lauris
Rytis Lauris shares lessons from adopting AI inside Omnisend: creating time to experiment, starting small, assigning ownership, and improving tools people will actually use.
Watch it. Put it to work.
ABOUT THIS SESSION
Make AI part of the work.
Rytis Lauris shares lessons from adopting AI inside Omnisend: creating time to experiment, starting small, assigning ownership, and improving tools people will actually use.
For marketing leaders and operators turning AI experiments into dependable team workflows.
- Speaker
- Rytis Lauris
Co-Founder & CEO, Omnisend - Recorded at
- Austin 2026 ↗
April 2026
Speaker roles and platform examples reflect the session’s original context. This is an archived conversation.
TAKE IT BACK TO YOUR DESK
Ideas to put to work.
- 01
Give adoption time and attention
Set aside working time for people to improve a repetitive task with AI.
Read this part · 01:05 ↓ - 02
Draft the work before automating the decision
Start with a reviewable draft or recommendation before allowing an agent to act independently.
Read this part · 05:08 ↓ - 03
Keep the first version small
Choose a narrow first use case and improve it through actual use.
Read this part · 08:17 ↓ - 04
Assign every agent an owner
Name a human owner who reviews output, fixes failures, and maintains the workflow.
Read this part · 14:18 ↓ - 05
Fit AI into an existing habit
Place the tool inside a recurring task rather than ask the team to invent a new habit.
Read this part · 18:33 ↓
READ THE SESSION
The edited transcript.
A condensed, edited reading version based on the YouTube captions and session chapters. Repetition and unclear audience audio have been removed; the discussion is paraphrased for clarity, rather than presented as a verbatim transcript. Timestamps refer to the original video.
Original recording on YouTube ↗Give adoption time and attention
Providing access to AI does not mean a team will change how it works. Rytis describes making experimentation deliberate through dedicated AI days, giving people outside engineering a chance to identify problems and build small solutions.
The emphasis is on participation in real work rather than a separate innovation exercise. People closest to repetitive tasks can see opportunities that leadership may miss. A scheduled opportunity to try those ideas helps turn curiosity into something the team can evaluate.
Draft the work before automating the decision
A sales example prepares a follow-up email from a call, including questions and commitments, then leaves a person to review and send it. The assistant removes preparation work without assuming every generated detail is correct.
Other examples involve influencer review and support. Across them, the useful starting point is a task with a recognizable input and output. Make the first version easy to check so the team can learn where it helps and where it makes mistakes.
Keep the first version small
Rytis warns against making an AI project so complicated that it becomes difficult to launch or maintain. A narrower tool can demonstrate value sooner and reveal what actually needs improvement.
Treat the workflow like a product. Launch a usable first version, watch how people use it, and adjust it in response to their experience. The value comes from repeated use and improvement, not from the complexity of the original design.
Assign every agent an owner
An agent needs a person responsible for its quality and continued usefulness. Rytis compares that responsibility with managing a new employee: someone must know what the tool is supposed to do and notice when it stops doing it well.
Ownership includes iteration, maintenance, and checking failures. A one-time launch is not the end of the project. Without a clear owner, small errors and changing requirements can quietly turn a promising tool into something the team no longer trusts.
Fit AI into an existing habit
Adoption is easier when a tool improves something people already have to do. Rytis points to subject-line and preheader generation as an example of assistance placed inside a recurring marketing task.
The broader lesson is to start from the routine, not from a desire to use a particular technology. Look for work that happens often, is easy to recognize, and creates a clear benefit when made faster or better. That gives the team a reason to return to the tool.
KEEP THE IDEAS COMING
Stay in the conversation.
3 SESSIONSPLAY IT BACK. PUT IT TO WORK.
YOUR NEXT STEP: IN THE ROOM
Bring your questions.
Meet your people.
Join Commerce Roundtable in 2027 for focused learning and conversations with people working through the same challenges.