The first week of October showed where business software is heading. Atlassian and OpenAI expanded their partnership to bring OpenAI models into Atlassian's Rovo and agents across its platform, and Cisco unveiled Webex capabilities that let AI agents take part in spaces, meetings and calls to carry out multistep work. At the same time, CNBC reported that OpenAI told investors it reached about $50 billion in annualized revenue at the end of September, below a widely reported higher figure, and AI-related stocks fell on the news.
Agents Are Moving Into Existing Tools
The pattern is clear: rather than asking people to adopt a separate AI app, vendors are placing agents inside the tools teams already live in, from project trackers to meetings. For SaaS builders, that raises the bar. Customers will expect your product to work with their agents, through clear APIs, sensible permissions and actions an agent can safely perform.
The Economics Matter More Than Ever
The market's reaction to revenue numbers is a reminder that AI is expensive to build and run. For your own product, that means knowing the cost of every AI feature, not just its appeal.
- Measure cost per task and per customer, and price AI features accordingly.
- Use the smallest model that does the job well, and save larger models for hard cases.
- Cache results and avoid repeating the same expensive request.
- Track whether each AI feature actually changes customer behaviour or retention.
What We're Telling Clients
Make your product agent-friendly with good APIs and permissions, add AI where it removes real work, and keep a close eye on what each feature costs to run. The winners will be products that are useful to both people and the agents working alongside them.