OpenAI Academy
Article
August 28, 2026

How Ariso uses OpenAI models to build a workplace coach

How Ariso uses OpenAI models to build a workplace coach
# Startup

At Ariso, Ari helps people prepare for workplace conversations and follow through on commitments.

How Ariso uses OpenAI models to build a workplace coach
Max Heckel uses OpenAI’s voice models to power a workplace coach inside Ari, the AI assistant he is developing at  Ariso . Based in Columbus, Ohio, Heckel is Ariso’s founding engineer and chief technology officer. Ari draws on meetings and connected workplace tools to help people prepare for conversations and track commitments.
Ariso connects with Google Workspace and other services so Ari can understand someone’s work without a long introductory conversation. It generates documents and work-status updates from that context. Its coach speaks with users, helps them think through workplace problems and leaves recommendations and to-dos that it can follow up on.
Before a team review, Heckel wanted to ask why CEO Erkang Zheng was spending so much time coding while the company needed to win customers. The CEO ranked first in the team’s code-change statistics. Heckel talked it through with Ari, which asked: “What’s the core thing you’re holding back from the CEO?”
“It legitimately changed the outcome of the meeting,” Heckel says. He raised the concern without making it sound like an accusation and learned that Zheng was responding to direct customer requests he had fielded during sales calls, and increasing how responsive the team was at a crucial stage. 
When a client requested a document explaining how to use Ari, the assistant captured the commitment in meeting notes. A week later, the day before the next client meeting, it warned Zheng that the document was still outstanding. Ari also checks connected work for evidence that a task is finished. Heckel estimates it can accurately mark about half his follow-ups complete without a manual check-off.
Heckel studied computer science at Ohio University. His AI work began around 2017, helping an e-commerce company recommend products based on what people put in their carts. He took online machine-learning classes, then experimented with GPT models by generating lyrics for imaginary songs by his favorite bands. In 2023, he founded SciSummary, which summarizes research papers and helps researchers organize references. He says it now has about a million users and 2.5 million uploaded papers. It also supplies paper summaries to the Association for Computing Machinery. 
At Ariso, the voice coach runs almost entirely on OpenAI models. Heckel says connecting his software to OpenAI’s models has helped the team launch features faster. Their open-source tool, Ivan, asks Codex to review proposed code changes and sends the feedback to a coding agent for revisions.
Founded in February 2025, Ariso has four full-time employees and more than a dozen early design partners: companies testing the product and helping shape it. It began working with early customers in spring 2026 and has started running intensive pilots as the product improves.