Assembled

What it is

Assembled is workforce management for customer support: forecast the contact volume, build schedules against the forecast, and watch adherence in real time. It sits on top of the helpdesk rather than replacing it — Zendesk, Intercom, Salesforce Service Cloud, Gladly, Kustomer, Gorgias and Dixa on the ticketing side; Amazon Connect, Five9, Genesys Cloud, NICE, Talkdesk and UJET on the telephony side. It was founded by three Stripe alumni and raised a $51M Series B led by NEA in May 2022, with Emergence Capital and Basis Set Ventures participating. Stripe, Intuit, Autodesk, Robinhood, Duolingo, HubSpot and Intercom appear on the public customer list.

Four lines sit on one contract: WFM (forecasting, scheduling, adherence), AI Copilot for human agents, AI Agents that resolve chat, email, SMS and voice contacts, and Vendor Management for teams that run BPOs.

Why it shows up in Customer Success stacks

• Three kinds of headcount get planned against one forecast. In-house agents, BPO seats and AI agents are usually planned in three separate places, which stops working the moment containment moves — an AI agent whose deflection rate shifts ten points in a quarter changes how many humans you need and when. Assembled schedules all three populations in the same tool. Gartner listed the company as a Representative Vendor in its 2025 Market Guide for Contact Center Workforce Management and cited exactly that scheduling scope, per Assembled’s press page.
• Vendor Management audits BPO invoices against delivered hours. You upload the capacity plan, vendors confirm or amend their allocation, and the platform reconciles productive hours billed against hours actually worked, including billable non-productive events like training and breaks. It pulls schedules from Aspect, NICE and Verint every few minutes and shows adherence per BPO site and queue on hourly-interval heat maps. For most support orgs the BPO invoice is the largest line nobody can check.
• The MCP server is why this rates above the incumbents on AI. Launched May 2026, and Assembled says it is the first in the contact center WFM category. It exposes read and write access across forecasting, scheduling, intraday, workforce composition, compliance and performance analytics, authenticated by OAuth and scoped per account. An intraday manager asking Claude why chat SLA broke on Tuesday afternoon and then moving a shift in the same conversation is a different job than reading a dashboard.

Pricing reality

Published, per agent per month: Core $25, Pro $45, Enterprise $75. Pro is the practical floor — ML forecasting, forecast accuracy reporting, automated time off and shift swaps all sit above the Core line, and WFM without forecasting is a scheduling spreadsheet with a login. AI Copilot starts at $35 per agent per month. AI Agents start at $0.65 per conversation, with voice available on per-resolution pricing. Vendor Management is quoted, not published.

Vendr’s marketplace data, drawn from 49 purchases, puts the median Assembled contract at $29,400/yr, ranging $12,000 to $46,880, with buyers negotiating about 16% off. That band matches list arithmetic — 50 agents on Pro is $27,000/yr — and that agreement is the real signal. Unlike NICE or Verint, the published number is close to the number you pay, so you can size the business case before the first sales call.

Best for

Support and CX leaders running roughly 50 to 400 agents across a mix of in-house staff, BPO seats and AI agents, on a modern helpdesk, who need forecast-driven scheduling without a six-month enterprise WFM implementation. The fit is strongest where BPO spend is already a line you defend in budget review: the invoice reconciliation and adherence evidence cover a meaningful share of the contract on their own.

Do not buy it if

Your team is under about 20 agents. At that size forecast error swamps the schedule, and a shared calendar plus helpdesk reporting gets you the same result for nothing. Skip it too if Zendesk is your only channel of record and you run no BPO — Zendesk WFM, the former Tymeshift, is a native add-on at $25 per agent per month, and the integration work you avoid outweighs the forecasting depth you give up. The same logic applies to Dialpad shops, which have had native WFM since Dialpad acquired Surfboard in October 2024.

Versus the alternatives

• NICE — the deepest omnichannel forecasting and the largest install base in the category, plus Playvox WFM since 2023. Pick it for thousands of agents, shift bidding, and regulated scheduling rules. Pick Assembled when implementation time and admin headcount matter more than feature depth.
• Verint, now combined with Calabrio — Thoma Bravo completed its $2B Verint acquisition on 26 November 2025 and merged it with Calabrio. Pick it when you want WFM inside one workforce engagement suite alongside quality management and recording. Price in the risk of two overlapping product lines being rationalized on someone else’s schedule.
• Zendesk WFM — the cheapest credible path at $25 per agent per month, and the right answer when Zendesk is the whole support surface. It gives up multi-vendor BPO management and the AI agent scheduling entirely.
• Decagon — the fastest-growing entrant on the overlapping AI-agent side. Pick it when autonomous resolution is the entire project and you have no human staffing problem to solve. Teams buy Assembled’s AI agents because they still own a forecast; they buy Decagon to shrink one.

Watch-outs

• Reporting depth is the recurring complaint. Assembled rates 4.7/5 across 85 reviews on Software Advice (4.85 for support, 4.60 for ease of use), and the cons reviewers name are reporting that is not yet where they need it, clunky schedule updating, and metric definitions that are hard to interpret. Guard: name the six to eight metrics your weekly ops review will run on, and build those exact views during the trial. If any of them needs an export, budget the API work before signing rather than discovering it in month two.
• Per-conversation AI agent pricing meters attempts, not outcomes. On chat and email, $0.65 buys a conversation whether or not it resolves; an escalated contact bills once as automation and again as human handle time. Guard: negotiate a containment floor or per-resolution rate at first signature, and track deflection weekly against the AI Experience Scores metric rather than at renewal, when the ramp is already priced in.
• Forecast quality is downstream of helpdesk hygiene. Volume forecasts learn from historical contact data, so inconsistent queue routing or a mid-year taxonomy change degrades the output that the whole schedule rests on. Guard: freeze queue and channel definitions for the first two forecasting cycles, and hold the accuracy report against a manual baseline you kept from before go-live.
• Vendor Management is an unpublished add-on, and Five9 resells the platform. The BPO module carries no list price, and the Five9 Select ISV partnership means some buyers land on a reseller paper trail. Guard: get Vendor Management quoted in the initial order form, not at expansion, and if you buy through Five9, confirm in writing who owns support escalation and what the renewal uplift cap is.