Agents
BidderOps gives you a team of AI agents. Each one has a single job, works from your own data, and reports back so your people can review the work and make the call. You see their names across the app: on the Agents page, in the top-bar activity list, on each AI feature, and in Settings → AI.
Your agent team
| Agent | What it does | Where it works |
|---|---|---|
| Scout | Scores new tenders from your feeds against your bid profile. | Discover |
| Analyst | Reads the tender documents and pulls out scope, criteria, dates and the submission checklist. | The RFx analyzer on an opportunity’s Overview tab |
| Advisor | Recommends go, no-go or conditional from your win history, with the reasons. | The Go/No-Go advice card on an opportunity |
| Researcher | Researches the project, people, competitors and partners on the web, with sources. | Bid intelligence on an opportunity’s Competitive Edge tab |
| Matcher | Imports references and CVs, matches them to each tender and tailors your exports. | Experience, and Capability Match on the Competitive Edge tab |
| Strategist | Finds the patterns in your wins and losses and writes the pipeline briefing. | Portfolio insights on Intel, and the Pipeline report |
| Copilot | Answers questions about your pipeline and makes changes once you approve them. | The chat panel |
Each AI feature shows a small badge with the name of the agent that runs it. Each agent uses the model you pick for its features in Settings → AI, where the features are listed by agent.
Scout, Researcher and Matcher belong to the Discovery, Bid intelligence and Experience modules. If your plan does not include a module, that agent’s card on the Agents page says so.
The Agents page
Open Agents from the sidebar, or choose View all agents in the top-bar list. The page starts with Your team: a card for each agent with what it does, where it works (with a link where it has its own page), and how many runs it has going now or finished in the last 7 days. The Copilot card opens the chat.
Below the team, the page lists every run, each labeled with the agent that did it:
- Running: every run still working, with how long it has been running and the model running it.
- Finished: runs that completed or failed in the last 7 days, with how long each one took and the model that ran it.
The page refreshes on its own every few seconds.
Capability Match and chat replies are not listed as runs: they answer straight away in the panel where you asked.
Watching agents from anywhere
The bot icon in the top bar shows how many agents are running. Click it for a quick list of what is running and what finished in the last few minutes. Each entry shows the agent’s name above what it is doing, with the model running it (for example running · 12s · Claude Sonnet 4). Finished runs link to their result.
Some runs keep going after you click away: bid intelligence reports, section deep dives, CV and reference imports, and experience exports all work in the background, so you can keep working while they finish.
Which model ran it
Each run shows the AI model behind it. Models from the built-in list appear by name, such as Claude Sonnet 4. OpenRouter and local models appear as the model ID followed by the provider, such as anthropic/claude-3.5-sonnet · OpenRouter or mistral · Local.
The model is the one you picked when you started the run, so you can see which provider did the work and compare runs. Runs recorded before this was added show no model.
Stopping a run
Choose Stop next to a running agent on the Agents page, or the stop icon in the top-bar list. The run is marked Stopped by user, and anything it returns afterwards is thrown away. For bid intelligence, your previous report stays in place.
Stopping needs editor access. You can stop your own exports and any import or intelligence run in your organization.
Stalled runs
A run can lose its worker (for example, after a server restart or when the AI provider never answers). Runs going for more than 15 minutes are flagged as stalled in red, because no run legitimately takes that long. A stalled run will never finish on its own. Stop it to clear it, then start the work again.