loading...
Onward converts AI-generated meeting notes into structured action items and follow-through workflows – where every other transcription tool stops at the summary.
ENTRY ANGLES
Transform meeting outputs into actionable next steps rather than just transcription/summarization · Focus on closing the gap between meeting end and work initiation · Apply outcome-focused AI to decision-dense workflows
VERTICALS
CAPABILITIES
Meeting transcription and understanding, Action item extraction and task generation, Integration with downstream workflow tools
Onward graduated from Y Combinator in summer 2023 and received the standard $500K seed. It's still in the early customer acquisition phase, so granular product details are limited – but the core idea is strong enough to deserve its own analysis.
Over the past couple of years, a wave of AI plugins has emerged for video conferencing platforms, offering automatic transcription and AI-generated summaries of online meetings and team standups. Onward's ambition is to go a step further: to turn those words into actions.
The problem with meetings isn't capturing what was said – it's that decisions and commitments made during a meeting still require significant follow-up work to actually execute. If you don't do that work, the meeting was largely wasted. The more substantive the meeting, the more follow-up it generates, and the more time that follow-up consumes. Research suggests that employees involved in client meetings and internal syncs spend about 30% of their working time on the meetings themselves plus the surrounding administrative work.
Onward's plugin connects to Zoom, Google Meet, Microsoft Teams, and Slack. Like other tools in the space, it creates transcripts and meeting summaries. But its primary function is something different: extracting the commitments made during a meeting and then – critically – acting on them.
- Someone promised to send a follow-up email? Onward drafts it and drops it into the sender's email drafts folder. - A bug was flagged or a feature requested? Onward creates the corresponding ticket in Jira or the team's project management tool of choice. - A follow-up meeting was agreed on? Onward creates the calendar event, writes the meeting description, and adds the attendees. - A team member committed to a deliverable by a specific date? Onward adds it to their task list and sets a reminder. - It was a client call? Onward updates the deal status in the CRM and logs the planned next steps against the account.
AgentsForce – which debuted in the same Y Combinator cohort as Onward, then called Octo, and was [covered here](/review/tri-svojstva-dlja-bolshogo-i-denezhnogo-rynka) in August – took an analogous leap in a different domain. AgentsForce built an AI chatbot for customer support that doesn't just answer questions – it takes action to resolve them:
- A customer wants to return a product? The bot looks up the order, updates its status, generates a return shipping label, emails it to the customer, and initiates the refund. - A customer asks "where's my order?" The bot identifies the carrier, contacts their tracking system, finds out what's happening, and reports back.
AgentsForce has data from a client who previously ran a conventional FAQ-based chatbot: it could autonomously handle about 15% of incoming support tickets. After switching to AgentsForce – which can both take independent action and initiate contact with third parties – the resolution rate hit 72% within three weeks.
A chatbot that does things is five times more useful than one that only talks. The implication for Onward is similar: a meeting assistant that executes on outcomes should be dramatically more valuable than one that only documents them.
That's the conceptual thread connecting Onward and AgentsForce: AI tools that move users closer to the end goal, rather than simply improving the step they're already on.
Online meetings have become the dominant communication channel for client-facing work and team coordination alike. The tool category built around them – transcription, summarization, search – has attracted genuine adoption. There are real users and real revenue in the space.
But competing by building a slightly better transcription engine is a hard game. The players are already well-funded, and the differentiation on that dimension is marginal.
The smarter competitive angle is to ask what meetings are actually for – which is to make decisions and set the next actions in motion. A tool that reduces the gap between "the meeting ended" and "the work is underway" doesn't just improve the meeting experience; it makes the meeting more valuable as an output. That's the bet Onward is making.
This principle generalizes well beyond the meetings category. Look at any widely-adopted AI application and ask: what is the user ultimately trying to accomplish, and how close does this tool actually get them? The answer usually reveals a gap. Closing that gap – even by one step – often creates a product that's dramatically more useful than anything already in the market.
AgentsForce demonstrates this in customer support. Onward is attempting it in meetings. The most promising adjacent applications are in recruiting (moving from job description to candidate pipeline) and legal workflows (moving from contract review to annotated redline) – both high-volume, decision-dense processes where AI is already good at extraction but hasn't yet taken the next step into execution.