Inwise.ai

We built Inwise.ai to help salespeople before, during and after meetings with customers. It joined calls on Zoom, Google Meet and Teams, transcribed them, and turned them into meeting summaries, action items, follow-up email drafts and a timeline of how each stakeholder felt about the deal. It tracked each deal against a qualification framework (SPICED), flagged gaps such as a deal with no champion yet, and synced with Salesforce, HubSpot and Microsoft Dynamics. This page describes the product as it was in 2023–2025.

The team was very small: the CEO, who also ran sales, two engineers and a rotating group of interns, working with a small group of early customers who were trying the product. I worked across product and engineering.

Product

• Helped shape the product vision and value proposition, and where we could stand out from competitors.
• Owned the roadmap and kept it current as priorities shifted. Requests came from four directions: problems from current users, asks from prospects in demos, demo and marketing needs, and new ideas from the CEO. I weighed them against each other, decided what was feasible and worth doing, and said no to the rest.
• Ran sprint planning and backlog grooming, turning vague ideas and rough designs into clear, buildable requirements.
• Gathered user feedback continuously and used it to decide what to build next.
• Compared third-party services for recording, transcription and company data on integration effort, reliability and cost, and chose which ones to use.

Building it

• Built the core screens: dashboards, meeting summaries, deal views and pipeline review, usually from loosely defined ideas without finished designs.
• Designed ways to show complex AI-generated data linked across deals, stakeholders, meetings and action items, with clear loading and error states for features that wait on AI responses.
• Built the AI meeting-insights pipeline with the other engineer, from the recording webhook through transcript processing and LLM calls to structured data on screen. Refined prompts for more consistent results, and reworked the data model as the insights changed.
• Integrated calendar APIs, Recall.ai and Coresignal, and tracked down problems that crossed the frontend, backend, AI workflows and third-party services.