Guest Bio
Vadim Peskov is the founder of Diffco and a long-time builder of AI-powered systems for companies that want more than a “we use AI” slide in their pitch deck. He has worked across industries, from fashion and manufacturing to insurance and healthcare, helping teams move from manual workflows to AI-driven operations. Vadim brings a pragmatic, product-minded view of implementing AI that starts with business problems, not shiny tools. vadim-peskov-ai-take
Episode Overview
In this episode of Making Big Shifts, we go deep on implementing AI inside real companies. The conversation starts with a simple question leaders are not asking clearly enough: Are you trying to enhance human work or replace specific workflows? From there, we move into how to use experiments to test AI in marketing, why data fragmentation quietly kills most initiatives, and what it actually takes to keep humans in the loop without slowing everything down.
We also jump forward a few years. Vadim paints a picture of AI agents for hire, where a single marketer might manage 25 agents across strategy, SEO, and PPC, and small teams of 10 to 20 people can operate like a 500-person company. Then we zoom out to the bigger societal questions, like AI characters that act as friends, and how leaders should plan for a future where the capabilities and costs of AI systems change dramatically year over year.
Key Takeaways
- Start implementing AI by choosing your intent
The first decision is whether you are enhancing humans or replacing defined workflows. You can often do both, but without that choice, you get vague mandates like “we need AI” with no ownership and no clear success metric. - Experiments beat big bets
In marketing and product, AI shines when you use it to run more experiments faster. Instead of hand building every landing page, message, or offer test, you can use AI to spin up and validate many variations, then double down on what converts. The company that wins is not the one with the “best” product, it is the one that runs more experiments. - Data readiness is the silent blocker
Most companies underestimate how fragmented their data is. Critical information sits in spreadsheets from someone who left three years ago. When you start implementing AI, you discover that you do not even know where all your important data lives. Until you address this, any AI project will struggle. - Change management matters more than models
The technology is not the hard part. The hard part is the humans. If you roll out new AI powered workflows without clear training, champions, and communication, your team will say the old version was better and ignore the new tools. Implementing AI is change management with a different vocabulary. - AI agents will reshape team design
Vadim expects a near future where AI agents for hire handle focused tasks, from strategy to SEO and PPC, for 500 to 1,000 dollars per month while working 24/7. One strong operator could manage a fleet of agents and produce output that looks like it came from a much larger team.
Favorite Quotes (with timestamps)
“The core question that you need to answer to yourself as a leader is, are you trying to enhance the human work or are you trying to replace some workflows that are done by humans.” (01:59)
“The person that wins in a business is actually the person that is doing more experiments, not just the person that is doing the best product.” (05:57)
“Humans that think they will be replaced are probably right, and humans that think they will not be replaced are probably right as well. The point is, you need to know how to use the tools.” (06:51)
“Old ways will die. We can rethink the interface and let AI prepare the work so humans accept or adjust instead of being slaves to the system.” (20:30)
“We will have a marketer that is an AI agent, or a human that operates maybe 25 different agents at the same time, from strategy to SEO to PPC, for a fraction of the cost of hiring a full team.” (27:35)
“Do not think that things are impossible with technology right now. When you plan, do not think about current tech. Think about how it will work for you a year from now.” (35:06)
Playbook: How to Apply
- Decide what implementing AI means for you
Write down, in one page, where you want AI to enhance humans and where you want it to replace workflows. For each area, define the business problem, not the tool. For example, “reduce average time to respond to customer requests” is clearer than “build a chatbot.” This keeps your implementation grounded in outcomes, not features. - Run small, focused experiments
Pick one or two workflows, such as generating and testing landing page variants or analyzing support tickets, and design simple experiments. Use AI to propose multiple options, then let humans review, approve, and adjust. Track a few core metrics, like conversion rate or handling time, and learn before you scale. Implementing AI through experiments gives you quick wins and clear lessons. - Clean up your data and train your people
Before you roll out AI across the company, map where your key data lives and consolidate what you can. In parallel, invest in training. Do not assume people will figure it out. Assign an internal champion for implementing AI, and give frontline teams practical sessions on how to use the new tools in their actual workflows. - Pilot your “AI agents plus human” model
Choose one function, like marketing or customer operations, and design a pilot where one person orchestrates several AI agents. For example, one operator might run an agent for research, one for drafting content, and one for performance analysis. Make it explicit that the human is in charge of judgment and priorities while the agents handle volume and speed. Use this pilot to model what your org could look like in two or three years.
Subscribe:
🎧 Listen to the full conversation with Vadim Peskov on Making Big Shifts—available on YouTube and Spotify.
Subscribe for more candid conversations with founders and operators rewriting the rules of modern work.
