monday.com: bring your own agent, built in Make
On 15 October monday.com added a Bring Your Own Agent integration with Make. Your agent's logic runs in a Make scenario, and people can chat with it, @mention it on an item, or assign it work like a teammate. Four new beta modules arrive in Make: Watch Agent Events (fires on chat, mention or assign), Send Agent Response, Run monday.com AI (LLM calls to several providers through monday.com) and Run Platform Agent (an agent that can take actions on monday.com). To read and write board data, the agent connects to monday's MCP server.
Where it shines: teams already running ops on monday.com boards, with Make (or a Make-literate partner) in the stack. Intake triage, chasing missing fields, or summarising an item's history on request are natural first jobs.
Watch out for: every module is beta, and the agent's quality depends on how tidy your boards are. Investors are watching too: OpsControl notes Barclays downgraded monday.com on 19 October citing AI concerns, and predicts a change to its AI credit model before the end of November. Budget for metered AI, not just seats.
Linear: Loops that start from Slack and GitHub
Linear's 22 October changelog extends Loops, its automations that react to changes in Linear and connected tools. Loops can now be triggered from Slack and GitHub events, so a loop could answer product questions in a Slack channel, or draft a QA plan when a pull request gets a "ready for testing" label. Teams can save loops as shared templates, and each loop gets a reasoning level (Auto, Low, Medium or High) to trade speed and cost against depth. Linear Agent can also now read and update Figma designs through Figma's MCP server.
Where it shines: small, fast product and engineering teams who already live in Slack and GitHub. Reasoning levels are a quietly sensible cost control: keep routine triage on Low and save High for gnarly work.
Watch out for: Linear is built mainly for engineering. If marketing, finance or ops need the same agent, you'll be bridging tools rather than replacing them.
Asana: AI Teammates, priced per request
Asana's AI Teammates route intake, fill fields and assign owners across its Work Graph. The commercial model is the bit to read carefully. On Asana's pricing page, paid plans include 5 AI requests per user per month, capped at 50 per account on Starter and Advanced (250 on Enterprise tiers). From Asana's September webinar Q&A: requests are pooled across the organisation, only successful work is charged, simple Dash questions can be free while assigning work uses requests, and an AI Usage Dashboard sits in the Admin Console. Extra requests are listed in US dollars at $0.50 prepaid or $0.60 on demand.
Where it shines: structured, cross-functional processes with clear forms and fields, such as client onboarding, creative requests and approvals.
Watch out for: a 50-request account cap disappears fast once a busy intake form is wired to an agent. Model the monthly volume before you switch it on, and confirm sterling invoicing with Asana.
A five-step test before you turn any agent on
- Pick one job, not a platform. Name a single repetitive task (intake triage, status chasing, QA plan drafts) and the person who owns the outcome.
- Clean the data first. Agents act on your fields, statuses and templates. Messy boards produce confident nonsense.
- Count the meter. Requests, credits or reasoning levels: estimate monthly volume and set a cap or alert before launch.
- Check permissions and audit. Know what the agent can write, where its actions are logged, and how to switch it off.
- Run it for two weeks, then decide. Measure time saved and errors caught against a baseline, then expand or kill it.
The tools are converging fast. The differentiator for a UK SME won't be which logo has the cleverest agent; it'll be whether the agent is pointed at a real bottleneck with tidy data and a budget.
Why this matters for Clarity Growth
Clarity Growth helps UK teams choose the right work management tool, set it up cleanly and connect it to the rest of the business, from websites and ecommerce to CRM and support. October's upgrades make monday.com, Linear and Asana better places to run agents, but each assumes a well-structured workspace, a clear first use case and someone watching usage. That setup and integration work, including Make and MCP connections, is where most of the value is won or lost.
Sources
- New Make modules: build and power monday.com agents — monday.com Developer Changelog (15 October 2026)
- More ways to run and share Loops — Linear Changelog (22 October 2026)
- Asana pricing — Asana (AI Teammates request allowances; accessed 23 October 2026)
- Agentic Work Management Assembly webinar Q&A — Asana Community Forum (September 2026; request pooling, charging and admin controls)
- AI request usage pack (prepaid) — Licenseware (Asana extra request list prices)
- The work graph is being rewritten in public — OpsControl Work Management Market Pulse (19 October 2026; homepage positioning, Barclays downgrade, forecast)
- 10 Best AI Agents for Project Management in 2026 — ClickUp Blog (vendor roundup; Asana AI Teammates capability summary)