The most useful near-term application of AI in go-to-market may be less dramatic than replacing the seller. It is making the work before the conversation substantially better.
Desiree-Jessica Pely reached that conclusion through an unusual combination of finance, computer science, academic research, and startup selling. When she faced a list of thousands of possible buyers with limited capacity to pursue them, the problem was obvious: the team needed to know who deserved attention before it needed another automated message.
A large market is not yet a go-to-market plan
A total addressable market can produce an impressive slide while giving a seller almost no direction. A list of 10,000 accounts still leaves the most important questions unanswered:
- Which accounts have the strongest fit?
- Which are likely to have the problem now?
- Which signals should change their priority?
- What should the seller know before reaching out?
- How should limited capacity be allocated?
Jessie encountered that problem directly while selling to banks and hedge funds as a technical co-founder. She understood the market, but the lead list was too large for one person to pursue intelligently.
We need to do research first to understand who to go after first and how to best penetrate the market.
That sequence is foundational. Outreach cannot fix poor prioritization. It merely scales the consequences.
Research converts an account list into a decision
Traditional account research is valuable and expensive. A seller can examine company information, recent events, strategic initiatives, hiring, technology, and likely pain. Doing that consistently across a large market is difficult.
AI changes the economics of assembling and comparing those signals. It can help teams collect unstructured information, apply common research questions, and surface accounts that deserve human attention.
The objective should not be to produce more facts. It should be to improve a decision: pursue now, nurture, route elsewhere, or remove from the active market.
That requires explicit criteria. An AI research agent cannot compensate for an undefined ideal customer profile or a strategy that treats every possible buyer as equally important.
Start with the bottleneck, not the exciting technology
Jessie's company initially explored a nudging bot that would recommend what a seller should do next. Customer conversations revealed a more important opportunity.
We quickly realized all the problems start top of the funnel. So let's automate just the top of the funnel.
That pivot is a useful product and operating lesson. Teams often begin with the most visible workflow instead of the constraint limiting the whole system. A recommendation inside the sales cycle has limited value when the wrong accounts entered the cycle in the first place.
The stronger sequence is:
- Define the target market.
- Research and prioritize accounts.
- Assign the right coverage.
- Prepare relevant outreach.
- Support the seller through the active cycle.
AI can contribute at every stage. Its value compounds when the earlier decisions are sound.
Automation should remove work, not hide judgment
Jessie distinguishes between nudging people and automating work for them. That difference matters.
A system that generates another alert may add cognitive load without completing anything. A useful automation collects evidence, applies a clear rule, creates the needed record, or prepares a decision for review.
Human judgment remains essential where context is ambiguous or the cost of an error is high. The point is not to remove people from go-to-market. It is to stop spending scarce human attention on repetitive assembly that machines can handle consistently.
The best design makes the boundary visible: what the system did, what assumptions it used, and where a person must decide.
Selling remains a human, inefficient market
Jessie describes sales as a market defined by information asymmetry and uncertainty. Buyers do not reveal every need. Sellers do not know exactly which account will move. Both sides operate with incomplete information.
AI can reduce some of that uncertainty, but it does not eliminate the human dynamics of trust, timing, and communication.
That is why better research should lead to better conversations rather than simply more automated ones. If every company uses AI to increase message volume, relevance becomes more valuable, not less.
The strategic advantage will come from using technology to understand where a human conversation is most likely to matter.
Territory planning and account research should connect
Research and territory design are often treated as separate exercises. In practice, they inform each other.
Account-level signals affect which segments deserve capacity. Coverage constraints affect how deeply each account can be researched. New evidence can change both priority and ownership.
A mature system therefore connects market definition, account scoring, territory capacity, and routing. The goal is not a static ranked list. It is an operating model that keeps attention aligned with the best available evidence.
The practical lesson for go-to-market teams
Before deploying another AI sales tool, answer four questions:
- Which decision should become better or faster?
- What evidence should inform that decision?
- Which work can be completed automatically?
- Where must a person remain accountable?
AI is most useful when it reduces the distance between a large, noisy market and a clear next action. Account research is a strong place to begin because every downstream motion depends on choosing where to focus.
About the guest
Desiree-Jessica Pely is Co-Founder and CEO of Loi. Her background spans finance, computer science, academic research, and entrepreneurship. She applies that research discipline to AI-assisted account prioritization and go-to-market execution.
