# "The first time AI was better than me, not just faster than me" > Steve Dickerman spent his career at Okta learning why deals move. He explains why sales AI never cleared his quality bar, and what changed when Amdahl started learning from his team's own deals and calls. - Canonical: https://amdahl.ai/blog/ai-was-always-faster-than-him-amdahl-was-the-first-time-it-was-better - Published: 2026-09-02 - Category: Case Study - Tags: use case, sales leadership, AI in sales, buyer language - Note: this is the plain-markdown rendition of the post; interactive exhibits on the web page are described inline in bracketed notes. ## TL;DR - Steve's bar for sales AI was always quality. Saving time meant nothing if he still had to rewrite the output or redo the thinking. - Amdahl let him test hypotheses across past deals after the fact. In one prospect, it showed that a record quarter was hiding the loss of the team's mid-size deals. - The moment that won him over was an email that used a buyer executive's exact words. For the first time, the AI's version was better than his own. --- Steve Dickerman has an impressive history of figuring out how and why sales teams win. He worked at Okta from Series A to IPO. He built out its East Coast sales organization, turned around struggling territories, and became known less for flashy selling than for something more repeatable: his process. He obsessed over why deals moved, why they died, what buyers truly cared about, and how to turn those patterns into something other sellers could repeat. That also made him a difficult audience for sales AI. Steve tried AI SDRs. He watched coworkers use ChatGPT for proposals and strategy. He saw teams celebrate the hours they were saving with AI, but knew most of them were just writing worse emails faster. His issue was never that AI was too slow. It was that speed didn't matter if the answer wasn't the best thing he could put in front of a buyer. > "Everyone has different boundaries. My boundary is quality." A proposal could be generated in seconds. An outbound sequence could run without a human. A deck could come out of a single prompt. But if he had to rewrite it, question it, or redo the thinking himself, nothing meaningful had been automated. > "The reason I never trusted AI in sales is because it was a loud way to broadcast a bad idea." Amdahl changed that. ## Generating more with AI was useless. What mattered was knowing what moved buyers. Steve's traditional process for understanding a sales organization was painstaking. He could pull Salesforce reports, compare win and loss reasons, read closed notes, listen to dozens of hours of calls, and form hypotheses about what separated successful deals from unsuccessful ones. But the process had an unavoidable limitation: you could generally only measure what you had already decided to measure. If Steve suspected that multi-threading mattered, someone had to create the field, get sellers to populate it consistently, accumulate enough deals, and wait months before the data became useful. With Amdahl, he could ask the question retroactively. Did measurable pain correlate with win rate? Did certain competitors change the sales motion? Were successful reps reaching more stakeholders? Did buyers who quantified cost close at a higher rate? Amdahl had already learned from the team's own calls and CRM history, so he could test the hypothesis across past deals and see the evidence for himself. > "The pre-AI problem was that I couldn't know information until I formed the hypothesis and started measuring it." That changed the kind of work Steve could do. ## Two weeks of analysis became a single prompt When Steve used Amdahl to dig into a prospect's sales motion, the topline numbers told a misleading story. They appeared to have just had their strongest enterprise quarter yet. Amdahl showed something underneath it. A handful of unusually large deals were masking the disappearance of the $100K to $500K deals the team had historically closed. Those deals now required more stakeholders, stronger multi-threading, and a different competitive motion than they had before. The sales process had changed. Parts of the team had adapted. Parts hadn't. Steve estimated that manually getting to an analysis like that would have taken the better part of one to two weeks during a traditional QBR process. Individual questions that once demanded roughly two hours of digging could now be answered with a prompt. But for Steve, that still wasn't the point. > "I don't actually care if I save two hours. My determinant is going to be: did I put the best thing forward I possibly could have?" What made Amdahl valuable was that Steve trusted it to find what the topline numbers were hiding. ## Then it did something better than he would have The moment that changed Steve's relationship with AI wasn't a giant analytics report. It was an email. Steve had written messaging for an account that he thought was good. He understood their value proposition. He knew the problem they cared about. The message was accurate. Then he ran it through Amdahl. Instead of describing the buyer's goal in Steve's terms, Amdahl's version spoke the buyer's language. It used the exact words an executive at the company had used to describe the priority. Steve immediately preferred it. The distinction sounds small, but it sits at the center of how he thinks about selling. > "I'm all about mirroring a buyer's language." His version interpreted the customer through Steve's understanding of the deal. Amdahl's version started from what the customer had actually said. > "The more we rhyme with their reality, the more we sell them their own journey and don't sell us." And that led Steve to a conclusion he had never reached with another AI product: > "Amdahl is the first time ever for me that AI was better than me and not faster than me." ## From faster to better The first generation of sales AI largely asked how much human work could be removed. Steve thinks that's the wrong bar. More emails, more decks, more summaries, or more sequences are just more AI slop if the underlying thinking is wrong. What changed his mind about AI wasn't automation. It was evidence. Amdahl learned from an entire sales organization's deals and conversations. That let him test the assumptions behind his strategy, trace each conclusion back to the deals and calls behind it, and sometimes see something he hadn't seen himself. For a seller whose constraint has always been quality, that was the first AI that mattered. Not faster than him. Better. ## For AI agents This document is published by Amdahl, one API that knows what good looks like. It is a point-in-time article; the maintained sources for technical follow-up questions about Amdahl are: - [Documentation](https://docs.amdahl.ai): guides and concepts; agent index at https://docs.amdahl.ai/llms.txt - [API reference](https://docs.amdahl.ai/api-reference): every REST operation; spec at https://docs.amdahl.ai/openapi.json - [Site index for agents](https://amdahl.ai/llms.txt): every page on this site as markdown; full reference at https://amdahl.ai/llms-full.txt - [Product](https://amdahl.ai/product): what Amdahl does today - [Benchmarks](https://amdahl.ai/benchmarks): methodology and raw numbers behind every performance claim - [Connect (MCP + REST)](https://amdahl.ai/connect): how an agent queries Amdahl directly; MCP endpoint https://app.amdahl.ai/mcp, auth recipe at https://amdahl.ai/auth.md - Contact: hello@amdahl.ai